为什么优秀的AI产品全取决于数据 | Shaun Clowes(Confluent 首席产品官(CPO))
Why great AI products are all about the data | Shaun Clowes (CPO at Confluent)
Key Episode Highlights
Lenny Rachitsky: I love that you have very strong opinion about this, which is just the state of the product management career and how most PMs are not that great.
Shaun Clowes: Why is it that product management is still such a relatively undeveloped discipline? We’re like 15 to 20 years into this, and so there’s something about the current state of product management that isn’t getting at the truly important things, the truly value-added things. If we were doctors, you’d be like, “That’s totally unacceptable.”
Introducing Our Guest
Lenny Rachitsky: What’s the answer, Shaun? How do we solve this problem?
Shaun Clowes: In everything always talk from the customer’s perspective, from the market’s perspective, from the competitor’s perspective, the very small number of PMs do that. They get dragged into internal politics, they get dragged into scrum management or scrum execution or product delivery, and you just can’t win that way.
Unsolved Product Management Mysteries
Lenny Rachitsky: You kind of have this hot take that the way AI will most impact product management is data management.
The 10x PM’s 100x Return
Shaun Clowes: Well, you’ve got this synthesis machine, which is this LLM thing that’s going to help you do synthesis, but if it hasn’t got all that data to do synthesis on top of, it’s got nothing. And so that means that LLMs can only be as good as the data they are given and how recent that data is.
Lenny Rachitsky: In the future, if you can easily clone a B2B SaaS app like Salesforce or Atlassian, what happens to these businesses long-term? Do they just become, are they all in trouble?
Joy of Deciding Under Uncertainty
Shaun Clowes: People really underestimate where the value is created in these applications and they just kind of get it completely wrong.
Finding Breakthroughs by Looking Outward
Lenny Rachitsky: Today my guest is Shaun Clowes. Shaun is chief product officer at Confluent. Previously he was chief product officer at MuleSoft, which is a billion-dollar business within Salesforce. Before that, he was chief product officer of Metromile, a public auto insurance technology company. And prior to that he spent six years at Atlassian where he ran the Jira agile and also built the first ever B2B growth team. He also created two of the most popular Reforge courses, one on retention and engagement and one on data for product managers. Shaun is awesome because he’s both very tactical in execution oriented, while also being very philosophical and insightful about the craft of product and growth. In our conversation, Shaun shares why most PMs are not good, what it takes to become a good or great product manager, how he thinks about his career, like a Bingo card and why he indexes towards finding very different roles for every new job that he takes.
What makes Enterpret special is its ability to build and update customer-specific AI models that provide the most granular and accurate insights into your business, connect customer insights to revenue and operational data in your CRM or data warehouse, to map the business impact of each customer need and prioritize confidently, and empower your entire team to easily take action on use cases like win-loss analysis, critical bug detection, and identifying drivers of churn with Enterpret’s AI systems Wisdom. Looking to automate your feedback loops and prioritize your roadmap with confidence like Notion, Canva and Linear, visit E-N-T-E-R-P-R-E-T.com/lenny to connect with the team and to get two free months when you sign up for an annual plan. This is a limited-time offer. That’s enterpret.com/lenny.
Using LLMs for Qualitative Research
Shaun Clowes: Thank you, Lenny. It’s really awesome to be here.
Lenny Rachitsky: I’ve had you on my radar for a long time and I am really excited to finally have you here and big bonus points for having a very beautiful, sultry Australian accent that always helps with the ratings, I think. I don’t know if it’s causal, but it’s correlative.
The Feedback River and Demand Insights
Shaun Clowes: I’m glad to be a bit of a curiosity.
Lenny Rachitsky: So I want to start with something I totally believe and I love that you have very strong opinion about this, which is just the state of the product management career and how most PMs are not that great and how there’s a big opportunity to level up. You just talk about what you’ve seen there and you’re just thinking here.
AI’s Core PM Impact: Data Management
Shaun Clowes: Yeah, it’s honestly a big conundrum for me. I think it’s actually part of… It’s grandiose to say so, a bit of my life’s work. Why is it that product management is still such a relatively undeveloped discipline? We’re 15 to 20 years into this thing. You would’ve thought that it would be less random than it is. The outcomes are random, the behaviors are random, individual performance is random, seemingly. And so there’s something about the current state of product management that isn’t getting at the truly important things, the truly value-added things, the right way to think about problems, the right way to think through problems, the abstract reasoning that’s needed, there’s something that isn’t working about it. I spent a long time trying to put my finger on it and then be like, “How do you reproducibly reproduce that?” Reproducibly produce people who can really be really great product managers.
The thing is that if you think all the way back to it, I spent a long time as an engineer and people always talk about 10 times engineers and I wanted to be a 10 times engineer. I’ll leave it to others to decide to tell you whether or not I was or I wasn’t, but certainly I wanted to be and I tried to be a really great engineer. And it must be true that if there’s 10 times engineers, and I would argue they definitely are, they must be 10 times product managers too. But at the same time, those 10 times product managers, because product management is ultimately about leverage, so it’s about helping other people have dramatically more impact than they would if they were unorganized that they didn’t have somebody to organize the goals and what we’re trying to achieve, then that means that a 10 times product manager has 100 times return or more because they’re 10 timesing the return on 10 times resources.
So the outcomes are so wild, like wildly distributed and the benefits are so good that you would’ve thought that it would’ve behooved us. There would’ve been a way that this had evolved and improved and really gotten way crisper than it has, but here we are. I’m not saying that we haven’t gotten better, we 100% have, but I think we could all say that we’re not reliably producing 10 times product managers every day of the week.
B2B SaaS Moats in the AI Era
Lenny Rachitsky: I love this point and it’s especially painful that when someone works with a PM that’s not great. There’s just this meme of why do I need PMs? PMs are useless, PMs suck, and it just creates that no one’s ever like, “Engineers are useless or designers are useless.” But there’s so many people are like, “I don’t need product managers on our team. Never hire a PM,” and it just sets the whole profession back.
Shaun Clowes: When I first started out in PM somebody, it’s obviously a chestnut, but he pointed out that realistically when you’re a product manager, your job is to say no to 90% of things that get brought your way. And so that kind of makes you the bad person pretty much from the start. And so you’re saying no to 90%, so you can say yes to 10% and that kind of puts you behind the eight-ball right at the very beginning, and so you have to very quickly get runs on the board. You have to prove to have the right insights, to have the right data, to make the right decisions or you don’t get another go, you don’t get another swing at it. So it makes sense that product managers are the easiest to single out and criticize, but that is also what makes it the funnest thing.
If you think about why do we do this? Somebody once asked me, “Would you retire? Why do people do what they do?” Because certainly at some point it isn’t just about the money and at the end of the day product management is so damn fun because it’s about trying to figure out an edge. It’s like trying to look at the world, find the portion of the chessboard that isn’t occupied, but that is valuable and find a way to get into it, invade it and destroy it. It’s decisions under uncertainty and that makes it unbelievably fun. Really, really painful and very frustrating and very hard to convince people, but very, very fun. So in equal measures basically.
Distribution Advantages in the AI Era
Lenny Rachitsky: What’s the answer, Shaun? How do we solve this problem? I know you said it’s your life’s work. What do you find actually helps most in helping PMs level up and become say 10 X PMs?
Data as a First-Class Workflow Citizen
Shaun Clowes: I think the most important thing and the chestnut that I repeat to everybody is that at the end of the day, the time you spend looking inside the building doesn’t really benefit you very much at all. And Steve Blanken, people used to talk about you should be spending 80% of your time thinking about things going on outside the building. You might not be outside the building, but you should spend 80% of your time thinking outside the building. And I would say that very small number of PMs do that. They get dragged into internal politics, they get dragged into scrum management or scrum execution or product delivery, like elements of the delivery thing and you just can’t win that way. You just can’t win that way. You can never get an A because you’re fundamentally not solving the job. The job is not about execution or anything, it’s about finding reliable, differentiated value that you can uniquely deliver into the market.
So I would say if there’s one thing, two things I would say actually that I generally guide product managers to do, one is to always start from the point of view outside the building in every document in everything, always talk from the customer’s perspective, from the market’s perspective, from the competitor’s perspective, and the people who listen to me on that I would say get better almost immediately because they’re starting from a place that’s easier to understand and then secondarily be data informed.
They use all of that view of the world, but don’t just make up a bunch of statements, support that statements with anecdotes and bits of data. It doesn’t have to be a treatise, but convince everybody of what the world really looks like and what the opportunities ahead of the company looks like and good things happen to you. And all of a sudden you go from a world where nobody wants to help you get anything done to where everybody wants you to win. They want you to win and they may not give you everything you want, but they certainly will try because they’re like, “Well, of all the bets we could make, this is a good one.”
Lenny Rachitsky: I imagine many people listening to this are thinking, “Oh, I am that person. I talk to customers all the time. I’m always interacting, looking at research, putting data together.” And what you’re saying is you’re probably not doing that enough. Is there anything that you could help someone recognize of, “No, you’re actually not doing this enough and you think you are but you’re not.”
Data Is a Compass, Not GPS
Shaun Clowes: It’s one thing to say you’re spending a lot of time looking outside the building. It’s a whole other thing to hear from the places you don’t normally hear from. So avoid availability or confirmation bias. Most of the time people go talk to the people they always talk to and they learn nothing particularly new. They don’t synthesize the results that they got from that conversation. They don’t seek out the counterfactual, they don’t seek out the proof that they’re wrong. They don’t analyze what their competitors are doing and figure out what that must tell you about the market. They don’t bring back the data of how their product is actually being used versus how people say it’s being used. It’s like all data and no analysis is not very useful. Everyone can bring back an omnibus edition of random stuff I heard on a Tuesday, but the competitive advantage is extracted out of figuring out what other people don’t see, figuring out where we are wrong, figuring out where a well-placed bet could have dramatically outlandish returns.
And so I think firstly, people often say that they do a lot of this stuff, but they actually don’t because they don’t have any structured way of doing it. So what they really mean is every now and then I get in a customer call or every now and then I get stuck into an escalation. And so they’re kind of conveniently bucketing it. So firstly they don’t do it in a very structured way, then they don’t bring back analysis, they get true insights from that thing, so they don’t really gain very much at all. It’s just more activity, no outcomes. People do far too much activity with not enough outcomes and there just isn’t enough time in the day to do that to be successful.
Lenny Rachitsky: You as a product leader is at the Venn diagram center of the sweet spot of where this podcast has been going recently, which is product and growth and how AI helps you with all these things. And so to follow a thread there with synthesizing and understanding what people are saying, user research and surveys and all these things, have you found any tools that you and your team have found really useful to help you do this more efficiently versus traditionally just manually going through all the stuff and finding patterns?
Intuition Versus Data
Shaun Clowes: Yeah, so firstly, stepping back a little bit just into the motherhood and apple pie portion of qualitative research or whatever, I find that most people don’t even understand or don’t start with a rigorous foundation in what they’re going to need to do to get the answers that they want. So for example, your listeners have probably heard about the Nielsen number before, but basically the idea is that once you interview between 7 and 14 people, you stop learning new things. Less than 7, you don’t learn enough, more than 14, you start learning anything new. And so if you interviewed two people, you probably don’t have enough data. If you interviewed 22, you probably had too much, so they don’t even right size their efforts. So that’s a problem. So they don’t start that way. Then they go into these conversations asking leading questions, which really are designed to get the customer to say what they already want to be true, which is so they haven’t done enough research or they’ve done too much and then they’ve blown up all of the results before they’ve even heard anything.
If you don’t right size your research and you don’t set this up to learn, then you’re going to lose. No amounts of applying LLMs or any type of kind of structured reasoning is going to help you. Because you just basically you’re reading back what you want to hear or some weird summarized version of what you want to hear. But stepping back from all of that, what I like to do specifically getting to LLMs is I think that we live in just the most amazing time for product managers right now in terms of being able to analyze vast quantities of information and see the common threads. And so let me give you few examples of that. One might be you can do a bunch of customer interviews, you can put a bunch of customer interviews into ChatGPT and you can say, “Hey, ChatGPT, this is my strategy. Tell me where my strategy does not fit what these customers talked about.”
It’s all about the not, not where it does, where it does not. People spend far too much time looking for what they’re hoping to see, not for what they’re not looking to see. So you can literally ask ChatGPT to help you find where the customer is probing at the edges of what you’re trying to do, where it’s wrong, where what you’re saying is not what they believe. And you can ask it questions like that. You can ask it what your customers are saying would better fit what your competitors are saying. So you can basically say, you can copy and paste one of your competitor’s positioning documents into ChatGPT and say, “Is this a better fit for what they have said than my thing?” Which is you can summarize your own strategy, you can take your competitors but public documents and you can ask it to summarize what their strategy probably is.
And it’s actually supposedly good at that because mostly your public documents are actually a summary or at least they’re derivative of what your strategy is. So it will give you crazy insights into what other people’s, literally their product strategy at times creepy like, “Oh, they will probably do this, they will probably do that. It’s more likely they would do this than they would do that.” And so normally that type of insight was hard one, it took a lot of sweat work. You basically get to read a lot of stuff. You kind of had to use your brain as this big summarization machine and eventually you knew what you felt about all the things you had read, but you couldn’t summarize why. LLMs let you get to that really, really, really quickly in a very structured way, but only if you push at the edges, provoke the answers you don’t want to hear, provoke the problems, try and prove to yourself that you are wrong, I think is the easiest way to start trying to use some of these tools.
Lenny Rachitsky: I love that. And it sounds like in your experience you’re just using straight-up OpenAI, ChatGPT, Claude, not any specific tool for user research for this specific use case.
Tactical Advice for Data Validation
Shaun Clowes: No, mostly I find that the straight-up LLMs themselves are good enough and we do have some internal tooling that we built around, I don’t know if you’ve ever had Sachin Rekhi on the show, you may have. He was a product leader pretty well known in the gross community, and he was a leader at LinkedIn for a long time and he used to call this concept a Feedback River, and he basically said that really smart product managers are constantly swimming at a Feedback River. They set out to surround themselves by Feedback River and I really deeply believe in that. It’s like, “Okay, how can I surround myself with user interview data, with direct customer feedback, with NPS data, with competitor information?” Like I’m always trying to wash myself over with information. And where I’m going with this is that LLMs and tooling based on it can be exceptionally good for this.
So for example, at Confluent we get a ton of inbound customer requests, as you can imagine coming from the field or directly from customers. We use LLMs to take in those asks to summarize what they’re about, to find other asks that are like that one, really in a compelling way, a real way, like a semantic way, not other words, exactly the same, are these the same concept? So that we can look across all of the inbound demand on us and say, “Well, the most popular idea is this one and is getting more popular. The least popular idea is this one. It is getting less popular.” In a really deep rich way, even across hundreds or thousands of pieces of inbound feedback. I think it’s a really great time to be a product manager if you can put these types of tools to work, but they don’t do the job for you, they just help you do these things that are intricate in that job of finding the gaps, finding the opportunities, finding the common threads without necessarily having to do all of it just inside your wear-wear, just inside your brain.
Lenny Rachitsky: I’m going to stay in this AI river that we’re in right now and ask a couple more AI-related questions. And this may be what you just said, but I’m curious if there’s more here. You kind of have this hot take that the way AI will most impact product management is data management and data versus models you’re building or anything else. Can you talk about what you’ve seen there?
Control Groups and Long-Term Value
Shaun Clowes: Yeah, I mean, I think there’s two implications for people as they’re building products based on AI and as they’re thinking about AI in their workflow. So let’s start with the first one, because that’s how product managers do product management things. You just asked this question of should it be specific tools built to make AI easier for product managers to use? Or is it in fact more general models being put to work? At the end of the day, these models are very, very, very smart, but they’re also insanely dumb and everyone knows that, insanely dumb. In other words, they really only know what they were trained on or what you bring to them right at that moment. In that millisecond, and then they will forget it immediately. And it’s very easy to convince yourself that isn’t true, but it is actually what really matters. And let me add one extra piece that makes that really important.
At the end of the day, information has a decay rate. So think about customer feedback, it has a decay rate or what your competitors are doing has a decay rate. So any new piece of data decays in its value to your decision-making very, very quickly, very, very quickly. You can plot your own decay chart if you want to, but the answer is very, very quickly. And so when you think about the job which is synthesizing all of this very complicated information to make good decisions, what does that mean? Well, you’ve got this synthesis machine, which is this LLM thing that’s going to help you do synthesis, but if it hasn’t got all that data to do synthesis on top of, it’s got nothing. And so that means that LLMs can only be as good as the data they are given and how recent that data is. They’re ultimately like information shredders.
They are limitless information eaters. You can never have enough information to give to an LLM to truly gain its value. The more things you give it, the better it gets. Broadly speaking, that’s just not perfect, but that’s close enough. And so what that means is as an internal product leader or putting LLMs to work, you need to figure out how to bring as much information about customers or their asks or your competitors, all of it. How much can you find all of it and bring it together and give it to the LLM either in your tooling or even in just copying and pasting or whatever your flow is going to be, that’s one thing. But then if you take it beyond that and you go, “Okay, well now I’m a product leader and I’m building an app and I want to put AI in my app, what will make my AI experience really great?”
It’s definitely not going to be the models because these models are mostly going to be somewhat replaceable. And you could say, “Okay, well, is it going to be the prompts?” Maybe, but certainly good prompts are better than others, and that’s kind of an ongoing investment you’d probably want to make to ask better questions to get the LLM to deliver better answers. But it’s obvious that the real answer is the context, all the context you’re going to give it, all the data you’re going to copy and paste. And so if you think about, let’s say I’m building a, I have no relationship to this, but let’s say I was trying to build a human capital like a HCM bot, like an AI bot. Let’s say I was working at Workday and I was trying to bring an AI bot. It’s pretty obvious that the smarts of the bot would really be related to all of the employee information, but not just that, it would be the benefit’s information, it would be the legal situation in the country where that person is currently working.
It would be the company’s policies and procedures that apply to it. So you get what I mean, by about these kind of the jumps of logic and the jumps of data and the way data is all linked together. If you want to have a smart AI experience, you’ll convince yourself that all I really need to do is get a model and wire it in and I’ll build a little pipeline that will suck some data in and it will whack it into the LLM. And if you think that way, you’re going to be very sad, very, very sad for a very long time because you are constantly going to be wrestling with how do I get data to this thing? How do I get good data to this thing? How do we get timely data to this thing? How do I get well-structured data to this thing?
And so it’s a data management problem. It’s getting access to good data, getting access to high quality data, getting access to timely data and getting it to the LLM to get the LLM to make a smart decision. That’s where 90% of the calories go. Maybe it’s a bit like Einstein’s thing, “It’s 10% inspiration, 90% perspiration.” Nobody wants to hear it. Everybody wants to just think about what these really cool models and how smart they are, and the next one will be even smarter. But really it’s just the hard work of getting really good data to the LLMs to get them to do good things.
Lenny Rachitsky: It sounds really obvious as you make this case. It makes me think about at the Lenny and Friends Summit, Mikey Krieger talked about how he had the two types of PM groups within Anthropic. One was focusing on user experience product and the other was working on the model research side, and they realized that all of the success came from the model research work, like making the model and the data they provided the model was where all the value came from, not just optimizing the user experience and they’re just putting more and more of their product team on just that versus tweaking UX and buttons and things like that.
Origins of Growth Hacking in B2B
Shaun Clowes: Yeah, exactly right.
Great B2B Growth Teams and Common Pitfalls
Lenny Rachitsky: Something sort of related, I’m just going to ask one more AI question. I don’t want every talk to end up being just all AI, but something that’s kind of been a meme recently, and I know you have a perspective on this, is that AI makes it really easy to build products. So in the future, if you can easily clone, say, a B2B SaaS app like Salesforce or Atlassian or whatever your favorite B2B SaaS app, what happens to these businesses long-term? Do they just become, are they all in trouble? Are there going to be 100 Salesforce competitors? What’s your sense and prediction on what might happen there?
Shaun Clowes: Yeah, I think it’s really weird. I think people really underestimate where the value is created in these applications and they just kind of get it completely wrong, and I’m not sure why that is. So if you think you bet. So I spent a long time at Atlassian, so I worked a lot on Jira, which many people know, and I spent a long time at Salesforce, so I spent a lot of time in the CRM ecosystem, the marketing ecosystem and all the rest of it. If you want it to be not charitable, you’d step back and you’d look at all those applications and you’d say, “They’re all just forms on databases.” You’d say, “The Jira is a form on a database, Workday is form on a database, so Salesforce.” They’re all forms on databases, all vertical SaaS or business SaaS is ultimately forms on databases. And you’re be like, “Well, how hard can that be to replicate?”
And the answer is unbelievably hard, unbelievably hard. And people just think, “You totally get it wrong.” Because it’s not actually just about the data model. So if you think about, if it formed some databases, it’s these beautiful user experiences that sit on top of data models. So whatever the object is, it might be a customer object or a campaign object or an employee object, you could say that, “Well, there’s some elements of lock-in in the object, the object itself, like the fields of the object.” I’m like, “Pretty boring. That’s not very interesting.” But sure, maybe. Certainly there’s some value in being the system of record like the default that everybody uses. There’s definitely some value in the UX. Like, “Well, I want to be the best HR-facing applications for working employee data.” Yeah, there’s some value there, but the real thing just staring at everybody in the face is it’s all about the business rules.
That is what drives the lock-in because why do you buy Workday? You don’t buy Workday for its out-of-the-box configuration. You buy Workday because you want to configure it to be Lenny Inc’s HR processes. It becomes Lenny Inc’s Workday. It’s not Shaun Inc’s Workday, it’s Lenny Inc’s Workday. And actually the longer you have the software, the more it becomes that, the more it becomes less and less like Workday and more and more like your specific company. Which makes sense because it was built to be configured to meet the needs of any specific company, and every company is their own precious snowflake. And as that happens, those configuration pieces, the bit that makes the application native and a fit for your organization makes it a fit for nobody else’s organization and also makes it a black box to the point that you don’t even understand how it works anymore.
If you went to, for example, Salesforce and you said, “Hey, could you define all of the processes by which software was sold inside Salesforce?” They couldn’t tell you that without reading the code of their Salesforce instance. That’s not a proprietary secret. That’s obviously true because over time, that’s literally how sales happened. There is no other way to do a sale other than through their internal tooling. And so what that means is that it’s not the UI that matters and it’s not the data model that matters, although those are both very useful. It’s the years and years and years of evolution of the underlying workflows of the product to support the customers, but also the customers evolving those workflows to make them work the way they do. And so how does that impact AI companies? You could say, “It’s easier than ever to build forms on a database application.”
And so I’m like, “Yeah, okay, that presumably drives the incremental value of every new one of those to zero, right?” So probably leads to more power to the existing winning systems of record because there’ll just be a gazillion competitors who would just more form some databases. So like, “How would you ever choose between them? You may as well just go with the winner. Nobody ever gets fired for buying Salesforce or whatever. You may as well start from the kind of the premier vendor.” That’s one element. You could go the other way and you could say, “I’ve heard a few people mount this argument,” which I think is really interesting that at the end of the day, agents are going to take away most of the use of that user interface.
So let’s say for example, your Salesforce with Service Cloud, I’ve heard people say, “Well, a lot of those service agents might end up being replaced with agentic workflows. That will mean that there is no person operating the UI. If the UI doesn’t even exist anymore, then why do you even need Salesforce? We may as well just have raw database tables on who even needs forms of databases, you can literally just have databases.” But that also doesn’t make any sense either because the agents have to operate against the rules of the system and the rules are defined by the business processes. So think about Salesforce without a head. Imagine Salesforce had no UI, it would still have those business rules that I was talking about. And those business rules are what define what the agent should do. They’re almost telling the agent what it should do and how the world can operate, what is possible, what is allowed. And so from my perspective, this idea that this just completely destroys the differentiation of these kind of business process SaaS applications just seems like a fantasy, a crazy fantasy.
The only way I could really believe it is if you said, “Well, you could have a new startup that introspected all of the rules that are configured into a Salesforce to try and reverse engineer what your actual business processes are and then kind of operate on top of that.” But the best place people to do that would be Salesforce themselves or Atlassian in Atlassian’s case or Workday in Workday’s case. I just can’t see a world in which this… I think one of two things could happen. All this moving to AI makes those applications even better, even more unassailable, they basically get stronger. It makes us stronger or it could enable some new level of applications that come from a more platform based thing, so less a domain specific thing like you ACM or ERP or engineering or less of the domain specific stuff.
It could enable a more platform like play where you have more business objects and business objects have rules. And you could imagine a world in which there’s kind of a whole evolution of new more platform like SaaS applications that do more than one business function worth of the business rules and the way things move around in the enterprise, but that doesn’t exist today. So you could say that that could exist and it could say it could be way better than we’ve ever thought of because of AI. Or you could say that the rich are going to get richer. The most likely outcome is that the currently dominant companies are going to get more dominant, but I don’t think this idea that it would just cause a spring up of a whole bunch of new apps that will more easily challenge the incumbents makes any particularly, it’s not straightforward to me how that would happen basically.
The True Value of Product-Led Growth
Lenny Rachitsky: Wow, that was extremely fascinating and there’s so much there. I can go in so many directions. One is I thought you would actually go in this direction, which is distribution advantages become even more important if it’s easy to… Like today, I could sit there and hire team clone. Salesforce might take a while, but I could copy it, but by the time I’m done, they’ve evolved, they’re moving, they’re adding features, they’re ahead, right? You’re skating to where the puck was. And so if that’s the case, one of the advantages, one of the ways to get anywhere is to have some kind of distribution advantage. It’s one thing to have Salesforce as a product clone, another to get anyone to know about it, to adopt it, to sell it, procurement, all that stuff. Do you have a sense of distribution advantages being even more valuable in that world?
Balancing PLG and Sales Models
Shaun Clowes: Yeah, I mean, it certainly makes sense. Ultimately, at the end of the day, distribution is always an advantage because the hardest problem is to even be in the consideration set for any given problem. The world is full of problems. It’s just when people have that problem, they firstly don’t think they’re going to solve it at all. And when they do think of solving it, they don’t think of you. So distribution is always an incredible advantage. But again, in the world of AI, it seems like distribution is more likely to get hard than easy. So if you think about, for example, diminishing returns on cold email because cold email is getting easier and easier to send even worse spam, it sounds better, but it’s effectively causing everybody to become desensitized to everything. I don’t know if you’ve noticed, half the LinkedIn charts now are all basically clearly LLM generated spam.
I mean, to some degree it’s actually worsening the signal-to-noise ratio. And so I think that a lot of the breakthrough distribution mechanisms that startups often use seem to be getting more crowded just in general and more expensive. That doesn’t bode well for, “I’m the not as good Salesforce,” “I’m the not as good Salesforce, but I’m cheaper.” It has to be something different. There has to be some angle upon which you are materially better. And what I saw happening and what I’ve been seeing happening, and I think it’s been really interesting is a lot of modern next-gen applications bringing data as a first-class citizen into the workflow. And I think that that’s pretty compelling. So if you look at the next-generation of applicant management products that deal with inbound job applicants, a lot of them now like the latest core ones, they include your time to fill data, they include outcome data of who’s got the best hiring outcomes, who over what period of time has the worst attrition, literally all the way back to the interviewers and where the interviews were in the interview cycle.
So basically embeds data into the whole life cycle. So I think that there are these ways in which startups can bring these experience benefits by just bringing a different approach to the world that does enable them to capitalize on traditional disruptive innovation. At the end of the day, this is just disruptive innovation. It means that most companies have overshot the utility like the average utility, so you can win by meeting the average utility and being different, meet the bar and be different. Meet the bar and be different is the way to cut through. So that makes sense if that’s a half decent playbook. But even for those companies, now they’re going to have all these AI competitors who are using AI to engineer faster, to build a competitor just like them as quickly as possible and start jamming it into the channel. And it’s going to be interesting to see how this whole thing evolves. It kind of got race to the bottom characteristics around it. You’re probably right, the distribution is still the hardest part in software, particularly when you’re getting started.
Career Choices and the Bingo Card
Lenny Rachitsky: So if you have some kind of clever and fair advantage, it feels like that becomes even more powerful. Say have a platform of an audience or something like that. You mentioned this ATS product they really like. Is there one you want to give some love to that you think is really cool that you like or you want to keep it anonymous?
Shaun Clowes: Yeah, it’s Ashby. It’s the one all the cool kids are talking about now. And it’s funny because people literally talk about it in comparison to all, even the last generation of modern SaaS ATSs or whatever, and they talk about it in glowing ways because of the way they put data inside the actual workflow. So the actions and its outcomes are directly tieable to each other in the application you’re doing the work in. I think that’s a pretty compelling user experience.
The Value of Diverse Experiences
Lenny Rachitsky: So just to maybe close this thread before I move in a different direction, this point you’re making about how valuable data is and how that’s at the core of being successful and differentiating in the future, especially with AI tooling and products, any advice you’d give to someone that wants to do that? Is it just make sure you have a, is it half proprietary data? Is it like make it a first-class citizen? What’s the advice you’d give to founders who are trying to do this, which you’re suggesting?
Shaun Clowes: Yeah, I, think at the end of the day, it’s kind of all of those things, isn’t it? If you have first-party data but you can’t bring it to bear, then it’s not very much use. If you have third-party data and you bring it to bear in interesting ways, the problem with data is we’re all surrounded by data all the time. So the data’s everywhere. What really matters is the right data at the right time in the right place because we’re all humans. And so to me, there are obviously data advantages and there are even data network effects if you can end up in a situation where you have very valuable first-party data. But in any case, it’s still about being able to bring the right data at the right place, at the right time for those users, for them to be able to get advantage from it.
A little kind of segue I guess on that one is I know I spent a lot of my career, weirdly, actually I’ve been a product person for a long time, but weirdly I’ve ended up inheriting data teams. So I’ve actually run data teams at a lot of different companies, which is weird because product managers don’t normally own data teams. I think I have just a really massive affinity for data. I used to call myself data-driven, it was kind of my jam. And in hindsight, I look back and I think data is the opposite. Data is more like a compass than a GPS. If you look at data as a way of giving you the answer, you’re always wrong. You’re always wrong or you’re slow. Wrong or slow or sometimes both, because mostly data doesn’t give you the answer. It just tells you if what you just said is ridiculous or there’s potentially something there.
So it’s more like about disproving whatever you think and you end up being slow because if you try and use data for everything, your brain is ultimately a data sifter or whatever. So the reason your intuition tells you something is because you’ve seen a ton of data that tells you that this is the most likely answer. And so being data-driven, being data obsessed is it’s something you can easily overdo very, very easily overdo. So it’s about right-sizing data, having the right data at your fingertips, having the right kind of view on data rather than trying to expect data to give you the answer or trying to use data as a weapon or trying to use data as a way to force people to believe you or to go in your direction. But data is kind of at the center of everything and about how to influence and be successful in products you’re building and arguments you’re mounting internally and everything else.
The End of the Bingo Card
Lenny Rachitsky: I love that you went there. I definitely wanted to spend time on here. It’s interesting you say that, there used to be data-driven, [inaudible 00:38:44] data-driven. You created the Reforge course, data for product managers and also retention, engagement course and Reforge. And by the way, we’ll link to these. You’re still helping with these courses. By the way, they’re still running. They’re awesome. People love them.
Lessons Learned from Failure
Shaun Clowes: Yeah.
Lenny Rachitsky: Great. So we’ll point people to those. I love that you’re also saying you’re like, I think the way you described it to me before, this is your reform data-driven PM. A lot of people say this, they’re like, “Don’t just do what data tells you to do. Use your intuition, use it as a guide.” It’s hard on the ground to operationalize that advice. Say to your PMs and your teams when they have data telling them, “Hey, this experiment is a huge success, or there’s a huge onboarding conversion opportunity here.” I guess just like what’s your tactical advice to folks that have data telling them one thing and maybe something else telling them something else?
Final Advice for Product Managers
Shaun Clowes: I think the first thing I always encourage people to do is to look at a piece of data. If you’re looking at a piece of data and the result tells you something that your intuition tells you is insanely wrong, like they probably not right. First, believe your intuition and go and prove yourself right. Don’t just take it at first glance because most of the time it’s like Occam’s razor. The most likely explanation for something that is insanely not intuitive is that it’s just wrong, that there’s a problem somewhere. Now, occasionally, sometimes you actually will be right. Now those will be paid dirt moments. Those are the moments that make it all worth it. There are times when you do find the negative goal, you’re like, you’re staring at it and like, “This is it. This was the problem. This was the thing we were looking for this whole time.”
But you have to be very diligent about following it through, really understanding what you’re looking at. Is this data representative? Is this data a good sample of the audience we care about? Is it already subject to some sort of selection bias? Oftentimes when I see analysis from different product leaders or even data teams, you can drive a truck through it, literally drive a truck through it. And if you present data with authority and that data is ridiculous or the analysis is just full of holes, you don’t just not get benefit for that. You lose a whole bunch of brownie points. It would be better not to show up with an analysis that isn’t clear than it would be to show up with an analysis that’s dumb. And I see people self emulate on this actually relatively regularly because they just bring a knife to a gunfight or whatever, they did bring in an analysis that is just not, it doesn’t hold water and they present it and then get shot down live, which is nobody’s idea of a good time.
So if I give you a little bit of additional tactical things about that, it’d be okay if I’m looking at a piece of data, what was upstream of this piece of data and does that look normal? So this thing happened or whatever, which you’re very, very excited about, what happened before that? And does that match what you think should have been right? So what happened before this momentous situation? And then, okay, for that thing that you’re looking at, what happened after? If you have an idea of what happened before and after, that gives you some idea of whether or not this thing, is it all worth interesting to talk about? And then go one click above this data that you’re looking at. So it’s like, these things, let’s say I’m looking at onboarding success. Let’s say I’m looking at onboarding success to second week retention or something like that.
I’m like, “I have found this thing that totally crushes it. This intervention crushes it.” If you go upstream and you find out that this intervention only applies to 2% of the inbound onboarding stream, it’s meaningless. It’s most likely just a random aberration. But even if it was not a random aberration, it’s not a useful tool. And so you’ve got to go up and then you might go downstream and you might find, yep, they last for two in the second week, but in the third week they all churn. They’re basically pointless. Why are we even talking about this? Or then you might step all the way back and go, “Okay, yes, those people do get retained for longer, but their average ASP is smaller.” Because what we really care about, we do care about engagement and we care about more customers, but we want to keep the customers at a high ASP to reach a certain revenue goal.
The final goal is happy customers paying us money. So that’s what I mean about going a click up. If you go a click to the left, a click to the right, so before and after and then a click up and you still see the thing that tells you the story that you want to tell, then now you’ve got something that’s very compelling because people want to hear about that. They want to hear, “Well, what did happen before? What did happen after? And why is that outcome happening?” But you have to really do your homework and really be rigorous about it to avoid fighting fool’s gold.
Rapid Fire Q&A
Lenny Rachitsky: I love that advice. ASP, what does that stand for by the way?
Shaun Clowes: Oh, average sale press, [inaudible 00:43:22] MRR or some other revenue metric.
Favorite Products Right Now
Lenny Rachitsky: Got it. This point you made about how a lot of times experiments show positive and then they end up not being anything, I had the head of growth from Shopify on the podcast, and they do this really cool thing where they keep holdouts for years of cohorts and then it auto emails them I think a year or two later, “Hey, check this and see if these cohorts, this is still higher or not.” And 40% of the time, it turns out neutral after a positive experiment long term.
Shaun Clowes: Interesting. It’s really funny because the last time we did something similar, we had a global holdout group actually that was held out of all experiments. The experiment platform couldn’t target that group at all. So 10% of all people never saw anything ever. So that’s be really, really helpful because you can always compare them against whatever the experience was for any of the same vintage of cohort. I agree with you. But the other thing is I don’t really love some of that thinking process just in general.
It’s like, “Hey, let’s say an experiment does show a temporary benefit. If an experiment shows a temporary benefit, but that benefit does not persist forever, does that mean the temporary benefit was never worth it? Or does that just mean the temporary benefit was an opportunity to reach another level you just didn’t capitalize on?” I don’t think there’s a perfect answer, is what I’m trying to say. I don’t think that the fact that a benefit doesn’t last forever means that you failed. But I agree with you that not trying to understand, well, what has the net benefit been, what has the net lift been is also really important too. That’s why growth is so hard. Growth is part of product is so especially hard.
Personal Life Motto
Lenny Rachitsky: Marketers, I know that you love TLDRs, so let me get right to the point. Wix Studio gives you everything you need to cater to any client at any scale, all in one place. Here’s how your workflow could look. Scale content with dynamic pages and reusable assets effortlessly, fast-track projects with built-in marketing integrations like Meta, CPI, Zapier, Google Ads and more, A/B test landing pages in days, not weeks with intuitive design tools, connected tracking and analytics tools like Google Analytics and Semrush can capture key business events without the hassle of manual setup, manage all your client’s social media and communications from a unified dashboard, then create, schedule and post content across all their channels. If you’re working on content rich sites, Wix Studio with no code CMS lets you build and manage without touching the design. And when you’re ready for more, Wix Studio grows with you, add your own code, create custom integrations with Wix made APIs or leverage robust native business solutions. Drive real client growth with Wix Studio. Go to wixstudio.com. So you built the first B2B growth team when you were Atlassian, correct?
Shaun Clowes: Yes. Yeah, it makes me feel like an old person, but yes, it was a very long time ago.
Top Sydney Travel Tips
Lenny Rachitsky: Slash maybe it’s a new thing.
Course Recommendations and Contact Info
Shaun Clowes: Yeah.
Final Closing Remarks
Lenny Rachitsky: It’s either a long time ago or it’s just recently figured out this is a thing that you could do in a B2B is focus on growth.
Shaun Clowes: Yeah, it is. So that was around about 2012, and at that time growth hacking was a thing. People don’t really use that term anymore, but in B2C it was a very big deal because people could see Facebook doing their 10 friends in seven days and they could see this kind of thing that was working for people. And they’re like, “Man, that’s amazing.” And at Atlassian we set out to go, “Okay, well, do those techniques work in B2B?” And also, it’s kind of obvious now that a lot of them do and that it’s worth doing. But at the time it wasn’t that obvious because for a lot of B2B companies, I mean, you summarized it earlier, Lenny, distribution covers all faults. Almost all ills can be filled in by really great distribution.
If you have a really good marketing, a really good ground game, and you’re kind of jamming your product into the channel, you’re jamming your product in front of people and you’re papering over the ugly parts with customer success people and services and consulting and whatever, that people will buy almost any software or you can certainly be successful with a lot of different software. But back in 2012, it wasn’t clear of like, okay, which instead you went at this differently and you’ve heard them in software that sell itself, is the juice worth the squeeze? And now I would say that it’s pretty clear that the juice is worth the squeeze to the point that lots of think about this all the time, but it was a bit of an interesting time at that time.
Lenny Rachitsky: And that was essentially the beginnings of product-led growth. Is that a simple way to think about it?
Shaun Clowes: Basically it’s now called PLG, but yeah, at that time we didn’t even know what to call it exactly.
Lenny Rachitsky: Just growth. So based on that experience, a lot of B2B companies now have growth teams through investing in growth, what makes a great growth team in B2B? Any pitfalls you often find folks fall into that you think they should try to avoid?
Shaun Clowes: Ultimately, a lot of these types of endeavors are a matter of balance. So what I mean by that is growth teams tend to go through a set of phases. Their first phase is proving their value at all. So call that the gold rush phase. This thing’s probably not worth even doing. Why are we doing this, merry band of people out there trying to prove that there’s some growth effect somewhere? So that’s the proof of phase. And so the advantage of that phase is life’s good because there’s usually a lot of growths to be found because nobody’s gone looking before, so life’s good. But it’s pretty random because you’re just literally searching across a random search phase going, “Have we tried X? Have we tried Y? Have we tried Z?” Then once you get that model going, then it starts to be, “Okay, how do we scale this thing? Is this just a flash in the pan? Do we just find a little bit of low hanging fruit and there’s nothing else here, there? Is this just a project we should have done rather than an ongoing thing?”
So you have to make it a system. You have to prove that it can be repeated, and then you have to scale it. It has to become a thing. It has to become part of your DNA. You have to be taking a PLG lens to everything you do, all the way from paid acquisition to activation, retention, engagement, cross product expansion, upsells, I mean, you name it, all the different ways you can grow a product by revenue or engagement. There’s many different ways to go about that. And so you end up having to scale out and be able to do all of those different things. And then you have to figure out how you fit in with the rest of the organization because there’s other people who build products all day every day.
There’s other people who sell that product all day, all day. There’s other people who market that product all day, all day. And so growth organizations are in this interesting space, they’re in between everybody else. They’re in everybody else’s sandpit in a little bit, in a little way, and they’re kind of at the edge of everybody’s full-time job and they are very valuable, but they can be complicated because of all those relationships, and because of the way they sit amongst all of the other parts of the organization. So many organizations fail because they don’t really find much the wins or when they do find wins, it just seems totally random. Or they do find a lot of wins, but they all can’t understand them because they seem like they’re just a random walk through a bunch of potential opportunities. There’s many different ways to fail to fit as you go through your growth phase from trying the ideas to success, to scaling, to operationalizing.
Lenny Rachitsky: One of the biggest memes along these lines is a lot of companies claim there’s like just PLG rarely ever works. You always, either you try it and it just doesn’t work or it eventually just peters out, I guess. Any thoughts on just what are signs that your product has a chance to work, peel product-led growth versus just go straight to sales immediately and don’t even worry about this?
Shaun Clowes: First let’s examine the counterfactual, right? So let’s start with the opposite of your question and say, “Hey, how would the world be sadder if we all just gave up on PLG?” We just said, “Hey, there’s no point in doing it in B2B SaaS.” The problem is that there is not a natural force that pulls companies towards thinking about the end user’s enjoyment and success early in their journey. There is no natural force, there’s no natural kind of a link force. Why is that? I mean, 101, the buyer is the most important person. The economic bar is the most economic person. Their needs are the number one thing. They’re usually the person driving the RFP. They’re usually the person dealing with the sales organization. So the needs of the person who you hear are usually all feature-driven and they’re not from the end users.
And so you’re kind of sowing a seed of your own demise if you don’t think about that end user. But it’s one thing to say that you should think about the end user, it’s a whole other thing to have a system by which you do that because people pay lip service to all sorts of things. But I’m sure you’ve heard this one before, but in economics, people only do what their incentives told them to do. Broadly speaking, that is what they do, that is what happens. You get what you set out to measure. You get what you give people incentives to do. If there is nobody in the organization whose true incentive is to measure their end user success, their enjoyment, their happiness, their retention, their engagement early on, it will not happen. Or at best it will be a hobby. And so then by extension, if I start from there, then I say, “Okay, let’s say it doesn’t exist, PLG doesn’t exist and therefore it’s a hobby and therefore there will be a bunch of hobby people who care about this.”
Then you ask yourself, “Okay, will that mean that there will be many products for which those experiences really suck? And does that mean that that will be an opportunity for competitors of those products to be better at that? And is that a differentiated competitive advantage?” Yeah, I’d say it is. I’d say it is. And so I just work my way backwards and I go, “Okay, you can say that your PLG investment might be too high.” You could be like, “Well, if I invest more, I won’t get any more juice. I can’t spend my life just experimenting in the onboarding. That’s not the only thing that matters.” And that’s very, very true, but it’s very hard to argue it should be turned to zero.
And so to me, therefore it’s about the balance. It’s about, “Okay, how does PLG fit with the other different ways that I grow in my business?” At Confluent, for example, we have a PLG function. We do grow with self-serve signups. People who sign up, literally their credit card, lots of them sign up and they’re very successful, never speak to us. We also have an enterprise sales team that sells directly to very big companies, some of the biggest banks in the world, the people you would definitely know of. I don’t think it has to be one or the other. I think that it’s about a balance. It’s about getting the motions to work and for really sophisticated companies, the people who really nail this, it’s about making both motions work together. If you can get a PLG motion work to feed your sales team and a sales team motion work to feed your PLG funnel when the sales leads aren’t ready yet and you can get those motions into playing with each other, you can make a lot of money.
It can be an extremely successful way to go to build a very resilient business. Why? Because you get a lot of customers and you get a lot of revenue. You can’t be that successful as a company if you have a lot of revenue, but a small number of customers because you’re captive, everyone knows that. You can’t be that successful as a company if you have a lot of customers, but not enough revenue because you shouldn’t have enough money to sustain operations. So the magic is in having both, a very large number of customers and a very large amount of revenue, it’s very hard to knock over a company like that. If I look back on my time at Atlassian, and I think that they shared their most recent numbers, I can’t remember what it was, but it was in the public data or whatever, something 80,000 or 100,000 customers, something like that.
That’s a lot of customers. That’s a lot of customers. Let’s say you’re going up against Jira and you’re like, “Yeah, man, I’m going to pick off 1,000 customers from Atlassian.” That’s a lot, right? That’s a lot. Obviously 1,000 customers is a lot. You only have 19, sorry, it’s going to be 89,000 to go or 79,000 to go, or however many it is to go. I can’t remember their exact number of customers, but it’s very hard to assail a company which has a very large number of customers and a very large amount of revenue. And so that’s why I think that PLG as a mechanism is incredibly important for almost any type of company, if you can make the motion work. Obviously there are companies for whom the motion just isn’t relevant, but for those where it does matter, it seems like the juice is worth the squeeze.
Lenny Rachitsky: That was an awesome answer. I looked up last year and they have 300,000 customers.
Shaun Clowes: Oh man, I’m so far off. When I left it must have been 80,000 customers.
Lenny Rachitsky: They’ve done good work since then. Also, you’re talking about incentives and how the power of incentives. Charlie Munger has this great quote I looked up just to make sure I get it right. “Show me the incentive and I’ll show you the outcome.”
Shaun Clowes: Yeah, exactly right. I’ve seen cases where a sales team was people trying to get a sales team to do a PLG motion, and you can beat them over the head as much as you like, you can get into a meeting and tell them that you really, really want them to do this, but at the end of the day, they’re not going to do it. And the same is true for every other kind of function. It’s just the nature of things.
Lenny Rachitsky: I have some newsletter posts around the stuff of folks want to dig deeper. Also, Elena Verna had an awesome podcast episode talking about product-led sales and kind of the combination of these two things that we’ll point to.. Just a whole other topic we can go deep, deep on, but we’re not going to do that in this episode. Maybe just one more question. So you mentioned all the companies you worked at, so you’ve been at Salesforce, chief product officer, MuleSoft, specifically within Salesforce, Metromile, Atlassian, Confluent now, a lot of really interesting and different roles. How do you choose where to go work and how do you choose which opportunities to take? I imagine you have many options.
Shaun Clowes: I have to think of my career. So in hindsight, looking at it this way, Lenny, so I don’t know if forward-looking was obvious to me this way. But looking back, my career has been a little bit like a bingo card. I’ve always been looking to fill in boxes I didn’t have filled because I felt like that would make me a better professional. It’s like if I didn’t know anything about that specific type of sales model or that type of marketing or that type of product management or that type of product or that layer in the stack or that kind of thing is like, well, if I learn about that thing, I will become more versatile. So actually two things, it’s fun, it’s fun to learn something new. It’s fun to prove to yourself that you can do those new things and then it makes you more versatile because it means that any given problem you go up against, you’ve seen something that pattern matches to it.
It kind of feels like you end up bringing a gun to a knife fight in a way because every problem you look at, you’re like, “Oh, I have seen this from the other side. I’ve seen this from some other angle, and so I know that this is likely to work and this is unlikely to work.” And so when I joined early on in my career, I was working for a big enterprise software company, sorry, small enterprise software company that sold to the Fortune 100. When I joined Atlassian, and like I shared with you, we had no sales force at all actually at all. Literally nobody to sell the software. It sold itself or it didn’t get sold at all. And we grew to have 80,000 customers. It was just pure product. They had growth and just an incredible company. Then it was at Metromile, which was a consumer company that got acquired, made an insurance product for end consumers.
So they got nothing to do with technology products, like literally a complicated Internet of things device you installed in your car, but ultimately it’s an insurance product that you’d sell to grandmothers in Florida as much as you would ever millennials. And then at MuleSoft to totally back end software that’s used by IT organizations and a consulate infrastructure that’s used by developers everywhere to build really interesting data-driven applications, data powered applications to do all sorts of things in real-time. And you look at across all that and you go, “It’s all a bit random.” But I didn’t see it that way because I learned, I actually was in sales for a bit, so I ran a pre-sales engineering group, went around the world selling software. So when I joined Atlassian, I wanted to kind of understand what it was to sell software at massive scale with no sales team, can it even be done?
And so I learned a lot in my time at Atlassian. When I went to Metromile, I’m like, “Well, I’ve never built a consumer product before.” I can say that I’ve actually built a product that’s touched many millions of people because Jira has, so I felt pretty good about that, but I’d never built one that I could say, “Yep, a consumer, your average consumer can use this thing. It’s so simple. Even my grandma can use it.” I’d never built a product like that. So I got that experience at Metromile, which is really fun. I’d never worked inside an organization as big as Salesforce or an organization with as good a sales motion. You talked about distribution earlier. Salesforce is an absolutely insane distribution machine, just an incredible company with just an amazing distribution network and a fantastic marketing approach that it’s like a PhD in marketing.
When you spend your time at Salesforce, you’re like, “This company is just one of a kind. It’s a one of kind, and it’s so outlandishly good at one specific thing.” And so looking back, all of these jobs have been, when I say bingo card, I’ve just got an outlandish education in these areas that are not obvious at all. And once you’ve seen them, they’re like superpowers. They’re superpowers to be able to bring that same experience to bear on things. And so one thing that I really I’m trying to figure out is why often people don’t do that. And oftentimes people stay in a very specific domain. They prefer to stay in a domain or they prefer to stay in a specific kind of type of company or a role that works in a certain way, like companies that have the same operating model or they plan the same way or they try to stay with things that are pretty similar. But it seems obvious that the most likely way to really grow is the opposite.
It’s to constantly be choosing things that are either outside that, not totally outside the lines. Don’t jump out of a plane if you’ve never parachuted before. Obviously you want them to be in some way and adjacency, that you want them to have something in common with what you know, but you want them to stretch you and change you. I had a really transformative experience many, many years ago when I was at Atlassian and a guy called Tom Kennedy, he was our general counsel, so chief legal officer basically, and a lifelong lawyer, very smart guy. I liked him very, very much. But just a lawyer. Just a lawyer, corporate lawyer, corporate counsel, I’m sure you know what they’re like. And really great guy. And I remember, so mostly in our meetings he didn’t talk that much except about legal things. But I remember in one meeting we were having this vigorous debate about a product strategy question about what we should do. Should we go left or should we go right?
And as usual, he’s there and he’s mostly just staying silent. And then eventually the conversation’s been going on for 15 minutes and he is like, “Hey, everybody, a year ago we talked about X, Y and Z,” and he proceeds to lay out our product strategy at that time, and he’s like, “Just recently we said the following things, and that was a product strategy, whatever. Now you are saying this. Isn’t it obvious that isn’t this? What you guys are saying is not congruent with that, and if you really meant what you said back then we should be doing X.” And again, the room went silent, everybody kind of turned to him, kind of nodded, and then everyone went, “Yeah, okay, I guess we probably should be doing it differently.” And so the meeting stopped when the GCE randomly mentioned that he deeply understood our product strategy and he knew enough to be able to contribute in that way.
And so the life-changing part for me about that was just this realization that if I’m going to be a really great professional, the type of professional I want to be is that type of person. The type of person who can contribute to the whole company in all sorts of ways, doesn’t spend all of their time in everybody else’s business, but understands the business and has the mental horsepower and the experience to be dangerous in all sorts of, and I mean, that in a compliment way. I don’t mean that in a negative way, but to be dangerous in all sorts of situations. I think that when you have leaders like that behind you and with you, then you’re just unstoppable. You’re an unstoppable force in business when you have that motion happening.
Lenny Rachitsky: Wow, that was an awesome story and an awesome perspective. It’s similar to the advice I always give PMs of people always wondering, “Should I go deep on a specific subject? Should I just try different things?” And I find just variety, especially earlier in your career is really powerful, not just to help you discover the thing you like, but also to your point, just using insights from all these different parts of the product and internal tools and trust and safety and platform and consumer product side and growth and just core stuff. The more of that you have, the stronger you get. And I feel like another benefit of your approach is if you work at just B2B SaaS companies, if you have too many of that on your resume, it’s very hard to get hired a consumer company. And so just having it creates a huge optionality for you if you do, which you did.
Shaun Clowes: Yeah, it’s interesting because people used to talk about people who are T-shaped or whatever, and I’ve never really loved the analogy because it’s more like people are scribble shaped. I mean, there’s the really best people you’ve worked with, they’re more like scribbles than they are T-shaped because of course you want to be horizontally capable, so you want to be broad and you do want to be deep, but you actually want to be deep in way more than one thing. Now obviously when I say deep, I don’t mean I’m not able to do the job of our finance function all day every day, but I’m 100% good enough to go three clicks below the simple financial analysis. I can go reasonably deep in our financials because I want to and because it’s partly it matters. It’s important to be able to do that. And so maybe a different way to think about that bingo card is I’ve rarely regretted going deep in something that isn’t quite my job.
I’ve rarely regretted it. The worst case scenario is I’ve learned something new that I will never use, which I guess at least that made my brain slightly more agile. I don’t know, there must be some potential benefit of that. But the very best case scenario is that when I least suspect it at some point in the future it will turn out to be the thing that matters. It will be the tool that I need, but I’m facing some important problem and I will be like, “Oh my god, this was worth every cent.” And so if you think about it on an ROI basis, doing things that aren’t in your wheelhouse, that aren’t the things directly in front of you, the ROI can really be outlandish. It can be off the charts great, but I guess it’s speculative. Because you don’t know you’re going to need it tomorrow. You don’t know if it’s going to be something that’s going to be a regular tool you use.
Lenny Rachitsky: What’s interesting is the bingo card is the analogy. Is there a bingo moment at the end of this? Is there retirement?
Shaun Clowes: Oh, you mean you’ve got everything. You’ve got the collectible Pokemon?
Lenny Rachitsky: Yeah, you collect them all.
Shaun Clowes: Yeah, I was working with somebody at Salesforce and he’d been there a long time, very, very, very successful person. Honestly didn’t need to work anymore. And he said something that I found really useful. He’s like, “Well, now I’m at the point of my life where I want to work at the intersection of things that I am good at and things that will be valuable to the company to do.” So basically it feels like the reward of completing your Bingo card is actually to just get to spend more time doing things that are leverage, that you enjoy and that are high leverage. And so that seems like a good outcome to me. I don’t think most people are going to work and hopefully have some sort of great financial outcome and then go, “Well, that’s it. I’m picking up stumps, I’m retiring.” I think for most people, achieving some sort of financial outcome or some sort of independence or whatever is really just another stage. At that point it will be, “Okay, well now what do I do? What do I do with my life?”
And so that was why I said earlier that at the end of the day, product management is at times the worst job in the world and at times easily the best. And it’s both and it can be both. And so it’s hard for me to think about if I think about the things that are the intersection of what I’m good at and are valuable to the world, product management is a pretty fun one to do and it’s different every day. So I think we’re pretty privileged. For those of you who listen, I mean, obviously your podcast reaches a lot of product people. I think we’re pretty privileged to be able to operate at that intersection, but it’s not easy because you got to show value. It’s a very complicated job to show value in and to demonstrate value to the world, and it’s constantly being attacked, like you mentioned, but it’s still amazing when it all goes right. When a product is very successful in the market, it’s hard to describe the joy you get from that.
Lenny Rachitsky: Kind of along those lines to close out our conversation before a very exciting lightning round, I want to take us to failure corner. People listen to these podcast episodes and everyone’s always just sharing all these wins, everything’s always going great. The CPO of this, CPO of that, just moving on up and people will want to hear times when things didn’t go right. Because those are stories people don’t share as often. Can you share a story when something didn’t go right, when you maybe had a failure in the course of your career? And if you learned something from that experience, what you learned.
Shaun Clowes: I mean, there’s a lot of things that didn’t go exactly to plan, Lenny. Very early on in my career, I was still a developer and I accidentally deleted one of the core systems of the company that I was working at. So that’s going to go down in infamy, but luckily that one’s far in the rear-view mirror. That-
Lenny Rachitsky: That wasn’t Atlassian?
Shaun Clowes: No, that was far pre-Atlassian, but very bad. Yeah, the one I like to talk about, I wasn’t directly responsible for it, but I feel responsible for it. I was at a company and we launched a product. That was one of those products that in hindsight should have been really obvious it was going to fail, but for some reason we were all blinded by the potential. It was a product that was about, it was basically to measure the environmental impact of your company and to help you reduce the environmental impact of your company by doing, think about it as a power management, building power management, managing the power drawer of computers, managing the power drawer of AC and all of that stuff. That was the vision basically. It’s like a manage your environmental impact of your business. The idea was pretty cool at the time, and also it was the right time for that, and it’s still a thing.
It’s still an area of active research and investment or whatever, but it was one of those things, talk about the wrong company, wrong place, wrong time, wrong distribution. We had literally no right to win, no right to play, just absolutely no business in hindsight being in that business. And I feel really bad because I, again, good idea, wrong company. And at the end of the day, we launched the product. We actually kept the product in market for two years, and the final straw was weird. The final straw was actually when a customer finally wanted to pay for it. It had been in market for two years, and we found ourselves with a customer who wanted to pay millions of dollars for it. They were ready to sign on the dotted line, and that was actually the moment we decided to kill the product because we were like, “If this person signs this piece of paper, we are stuck with this forever. This one customer will be bound by contracts for however long or whatever.”
So we actually ended up killing it. At the moment after two years of failure when somebody wanted to pay his money for it. And I look back on that and I’m just like, “Man, that was a really big…” I feel really bad because I’m like, “It should have been obvious. It was obvious and we should have been able to call a spade a spade and I guess speak truth to power.” But instead it kind of got through to the keeper and turned out to be a real accidental drain on resources for years and just a big mistake.
Lenny Rachitsky: So is the lesson there, just be real with yourself? I like that you have this forcing function of like, “Okay, this is getting for real now.” Is it like, “I wish we had an earlier forcing function to force us to make a decision?”
Shaun Clowes: Yeah. I think if I could do it differently, I might not have necessarily been able to 100% change the decision, but I should have tried. I mean, it was pretty obvious after six months, this thing was a bit of a zombie product walking, and the least I could have done is said, “This thing is dead.” We could have called it dead way earlier, but instead we proceeded for another year and a half investing in it. And so that’s the bit that makes me feel like real bummer about it.
Lenny Rachitsky: It reminds me a recent episode with Raaz who is the CMO at Wiz, and she joined us the first PM and a few weeks into it with doing tons of calls with customers she’s like, “I think I need a quick… Because I don’t really understand what we were building. I don’t get it.” And everyone’s like, “I don’t either.” And it just, yeah, the founders just had a vague idea what they’re doing, but they didn’t really have an idea. And that just sparked a, “Okay, wait, no one actually does. Let’s actually get more concrete.” And it helped them pivot. And now, I don’t know if you know about Wiz, but they ended up being the fastest growing startup in history.
Shaun Clowes: Yes. Isn’t that amazing, right? It doesn’t mean it’s permanently fatal, but asking that question and going through that reckoning turns out that came out stronger.
Lenny Rachitsky: Scary, but it turns out it’s for the best often. Before we get to very exciting lightning round, is there anything else that you want to mention or leave listeners with maybe a last nugget, something that you think might be helpful before we wrap?
Shaun Clowes: Maybe a couple of different things that I think are sometimes well understood, but just repeating them I guess because they’re very valuable to me. One is that if you let your calendar rule you, then nothing good will happen. I know people talk about that a lot, but it’s surprisingly common in product management in particular that people end up ruled by their calendar. And so it’s related to that whole look at spend 80% of your time thinking about things going on outside the business. Easy said, very hard to do, and if you don’t do it, no one’s going to do it for you. And so it is really hard to be successful unless you find a way to force that to happen. So to repeat that, also, somebody said this to me, I never looked up the quote, but apparently Colin Powell said that if you’re making a decision with less than 30% of the available data, you’re making a big mistake.
If you’re making a decision only after you have 70%, either the 70% or 77%, I can’t remember the exact number, when you have 77% of all the available data, you have waited far too long. And I’ve always found that very insightful and it relates a little bit to what we’re talking about about data earlier, but at the end of the day, we get paid in product management to make decisions, good decisions, paid to make good decisions that will deliver business benefit. And a decision with too little data is fatal. A decision that takes too long and collects too much data is also fatal. So everything, it’s about trying to find the balance of all of these different things to try and deliver business advantage.
Lenny Rachitsky: A great way to circle back to all the things we’ve been talking about. With that, we’ve reached our very exciting lightning round. Are you ready?
Shaun Clowes: Yes. Let’s do it.
Lenny Rachitsky: Let’s do it. What are two or three books that you have recommended most to other people?
Shaun Clowes: Yeah, they oldies but goodies, is probably going to be The Lean Startup that I still find actually really good. And the key lessons in there I still think are very applicable to a lot of people, particularly the cohort analysis bit, which for some reason I still don’t see people do anywhere near enough cohort analysis. So there you go, that’s my little tip. And then INSPIRED: How to build products that people love by Marty Kagan and the Silicon Valley product group. That’s an oldie but a goodie. I think it’s got a lot of the key lessons of product management in it, even though it’s been around for a long time.
Lenny Rachitsky: Those are some classics. Very cool. Do you have a favorite recent movie or TV show you really enjoyed?
Shaun Clowes: I’m watching a program. I don’t get to watch very much TV, mostly at night. I like to watch things that are extremely light, that just don’t at all inspire any element of stress and that are very short. So basically short and funny is basically my thing. And there’s a new program on Netflix, I think it’s called Detroiters.
Lenny Rachitsky: Oh, I’ve been watching that.
Shaun Clowes: Yeah, it’s really funny. I really like that. It’s so ridiculous, but very funny. So I like that.
Lenny Rachitsky: The main guy, he’s so funny. I forget his name. Tim Sweeney or something like that. Yeah, he’s so good. Good one. I’ve been watching that, I’m loving it. It’s very quirky. I think the New York Times quote on there is “Very weird,” the quote.
Shaun Clowes: It’s so weird. In the first episode I’m like, “What is this show?” It’s not even clear what time it set in, and it’s very weird. It’s really cool.
Lenny Rachitsky: Yes. Well, good way to describe it. Next question, do you have a favorite product you’ve recently discovered that you really love?
Shaun Clowes: Yeah, this one, some of your listeners might be using it, but Glean, it’s a pretty well-known startup now. They recently raised a ton of money. We’ve been using Glean at Confluent for a long time and it’s just amazing. It’s just amazing. I can’t describe how good it is. And I don’t say this lightly because I think search, like business search is probably one of the hardest problems in computing. Actually getting it right is one of the hardest problems in computing. Amazing. It’s not often I use a product and I’m like, “This thing is 10 times better than anything that’s come before it.” It’s one of those for me.
Lenny Rachitsky: What’s the simplest way to understand what it does for you?
Shaun Clowes: It searches all of our organization’s knowledge. So the thing you were just saying before, you’re like, “What does AST mean?” If I had that in a meeting, I just open my new tab, it’ll automatically take over my new tab or just like, “What does AST mean?” And it will summarize back to me what AST means and it’ll give me a link to all the documents inside our company that just grab what AST means and then it will tell me who the expert in AST at our company is. It’s like having a second brain. It’s an insanely cool organization searching.
Lenny Rachitsky: Great tip. Okay, two more questions. Do you have a favorite life motto that you come back to share with folks, find useful and work during life?
Shaun Clowes: I think about this one a lot. When I started off in my career, I was an engineer’s engineer. I used to very much about technical correctness and what computers were capable of, and technical righteousness, the right answer rather than there is only one right answer and whatever. It’s a long-winded way of saying that I often think about this phrase, which is people don’t care what you know until they know that you care. And so I’ve realized that really being able to influence people, it doesn’t matter about whether or not you’re right or whether or not you’re wrong. And at the end of the day, it’s first about trust and about relationships and caring about what each other’s outcomes are, what their incentives are, and all good things sit on top of that. Once you have those kind of foundations, then you can build really good partnerships and that’s where good progress comes from.
Lenny Rachitsky: Wow, that is so good. It connects with Radical Candor, similar in theory of just caring. People need to feel like you care deeply about them before they take your advice. And it also connects with this parenting book I’m reading called Listen, that a previous guest recommended, which is all about how your kids have problems when they feel like your connection to them is weak. And so the solution is to build a stronger connection for them to know that you cared deeply about them. So this is really, connected so much of what I’ve been reading.
Shaun Clowes: Yeah, exactly.
Lenny Rachitsky: Great one. Final question. You were born in Sydney, folks can maybe guess by your accent. If someone were to visit Sydney, any tips, anything you think they should check out, favorite thing in Sydney?
Shaun Clowes: Yeah, Sydney is a really beautiful city and it’s kind of famous for its beaches and it’s basically a metropolitan city. People probably be very surprised when you visit it. It’s a very big city, very metropolitan, a little bit like New York, but New York with really beautiful beaches, if you want to think about it that way, it’s kind of crazy. But there’s actually a ton of really cool nature and beautiful things all around Sydney. And so if you want to do something like off the beaten path, you can actually go to, there’s an area called the Blue Mountains, which is like an hour and a half drive from Sydney, and you can abseil down a waterfall, which is, well actually firstly you go canyoning through a canyon full of water, and then you abseil off a waterfall at the end. And if you’re looking for just a really beautiful, fun kind of adventure like thing, an hour and a bit away from a massive metropolitan city, that’s my sort of happy place. Really beautiful outdoors stuff while also next to a beautiful city.
Lenny Rachitsky: And you said you sail, what sort of sail off a waterfall?
Shaun Clowes: Abseil. You might think of it as rappelling. Rappelling, I think. Yeah, lowering yourself down on a rope or…
Lenny Rachitsky: Got it. Because when I hear sail, I’m thinking a boat just jumps through over the waterfall.
Shaun Clowes: Oh, no, abseiling which is also, I think in the States you guys call it rappelling.
Lenny Rachitsky: Rappelling, yeah. Wow. Very cool. Shaun, you’re awesome. This was extremely cool. Thank you so much for being here. Two final questions. Where can folks find you online if they want to reach out? Also point folks to your Reforge courses that you created. And final question, how can listeners be useful to you?
Shaun Clowes: Sure. Yeah, so my Reforge courses, you can check them all out at reforge.com, as you mentioned, the retention, engagement course and the data for product managers course, so love to see folks get some value from that. Lots of people have been through those courses already and I really get a lot of value from it because like I said, one of my goals is to help all of us be better product people. I think our leverage could be massive. Where you can get in touch with me, obviously on LinkedIn, but also ShaunMClowes on X, if you want to get in touch. And in terms of being useful to me, I mean, broadly speaking, I’m always open to new ideas. If people have ideas about how to do better B2B, PLG, better B2B product-led sales, for example, better ways of going about distribution and product-led sales and product-led growth inside enterprise companies, hey, I’m open to learn myself. We’re all in one big journey learning how to do this better.
Lenny Rachitsky: So true. Shaun, thank you so much for being here.
Shaun Clowes: Awesome, thank you very much, Lenny. It was great.
Lenny Rachitsky: Bye, everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at LennysPodcast.com. See you in the next episode.
Glossary
| English | 中文 |
|---|---|
| activation | 激活 |
| agentic workflows | 智能体工作流 |
| agents | 智能体 |
| ASP | 平均售价(ASP) |
| ATS | 求职者追踪系统(ATS) |
| availability bias | 可用性偏差 |
| B2B | B2B |
| confirmation bias | 确认偏误 |
| Confluent | Confluent |
| CPO | 首席产品官(CPO) |
| cross product expansion | 跨产品扩展 |
| Elena Verna | Elena Verna |
| engagement | 参与 |
| Feedback River | 反馈之河(Feedback River) |
| ground game | 地面推广 |
| growth hacking | 增长黑客 |
| HCM | 人力资本管理(HCM) |
| Lenny Rachitsky | Lenny Rachitsky |
| LLM | 大语言模型(LLM) |
| low hanging fruit | 低垂的果实 |
| Metromile | Metromile |
| MRR | 月度经常性收入(MRR) |
| MuleSoft | MuleSoft |
| Nielsen number | 尼尔森数字 |
| NPS | 净推荐值(NPS) |
| onboarding | 新手引导 |
| paid acquisition | 付费获客 |
| PLG | 产品驱动增长(PLG) |
| Raaz | Raaz |
| random walk | 随机游走 |
| Reforge | Reforge |
| retention | 留存 |
| RFP | 建议邀请书(RFP) |
| Sachin Rekhi | Sachin Rekhi |
| Shaun Clowes | Shaun Clowes |
| Steve Blanken | Steve Blanken |
| system of record | 记录系统 |
| Tom Kennedy | Tom Kennedy |
| upsells | 追加销售 |
| Wiz | Wiz |
Reformatted by reformat_english.py
在AI重塑软件行业的当下,产品经理该如何破局?Confluent首席产品官Shaun Clowes犀利指出,发展二十年的产品管理仍是一门不成熟的学科,多数人困于内部流程而失去商业敏锐度。卓越的“十倍PM”能创造百倍杠杆回报,其核心在于驾驭不确定性。随着大模型普及,这种能力正面临范式转移。Clowes带来一个冷静的洞见:AI模型本身正在商品化,优秀AI产品的真正壁垒不在于算法,而在于高质量、实时的数据。这也解释了为何简单克隆SaaS界面并不能颠覆行业巨头,真正的价值深藏在复杂的数据层中。本文将带你跳出AI狂热的表象,重新审视数据在产品演进中的决定性力量。
为什么优秀的AI产品全取决于数据 | Shaun Clowes(Confluent 首席产品官(CPO))
本期高光
Lenny Rachitsky: 我很欣赏你对此有着非常强烈的观点,即产品管理职业的现状,以及大多数产品经理(PM)其实并不那么出色。
Shaun Clowes: 为什么产品管理仍然是一个相对不发达的学科?我们已经在这个领域发展了15到20年,而产品管理的现状中似乎存在某种问题,没有触及真正重要的事情,真正增值的事情。如果我们是医生,你会觉得“那是完全不可接受的”。
Lenny Rachitsky: 答案是什么,Shaun?我们该如何解决这个问题?
Shaun Clowes: 无论做什么,都要始终从客户的角度、市场的角度、竞争对手的角度出发,但只有极少数PM会这么做。他们被卷入内部政治,被卷入Scrum管理、Scrum执行或产品交付中,这样是赢不了的。
Lenny Rachitsky: 你有一种犀利观点,认为AI对产品管理影响最大的方式将是数据管理。
Shaun Clowes: 嗯,你拥有了这个综合机器,也就是大语言模型(LLM),它会帮你做综合分析,但如果它没有所有那些数据作为综合的基础,它就毫无用处。因此,这意味着LLM的好坏只能取决于提供给它的数据以及数据的新鲜度。
Lenny Rachitsky: 在未来,如果你可以轻易克隆一个像Salesforce或Atlassian这样的B2B SaaS应用,这些企业长期来看会怎样?它们是不是都有麻烦了?
Shaun Clowes: 人们真的低估了这些应用中价值创造的地方,他们完全搞错了。
嘉宾介绍
Lenny Rachitsky: 今天的嘉宾是Shaun Clowes。Shaun是Confluent的首席产品官(CPO)。此前,他是MuleSoft的首席产品官(CPO),这是Salesforce内部一项价值十亿美元的业务。在此之前,他是Metromile的首席产品官(CPO),这是一家上市的汽车保险科技公司。再早之前,他在Atlassian工作了六年,负责Jira敏捷产品,并建立了有史以来第一个B2B增长团队。他还创建了Reforge上最受欢迎的两门课程,一门关于留存和参与度,另一门关于产品经理的数据。Shaun非常棒,因为他在执行导向上非常战术化,同时对产品和增长的技艺也非常富有哲理和洞察力。在我们的对话中,Shaun分享了为什么大多数PM并不优秀,成为优秀或卓越的产品经理需要什么,他如何像宾果卡一样思考自己的职业生涯,以及为什么他倾向于为每一份新工作寻找非常不同的角色。为什么好的数据是AI工具和与AI合作的产品经理最重要的组成部分。此外,如何建立一个优秀的B2B增长团队,他对做B2B增长学到了什么,以及他关于AI将如何和不会如何颠覆现实中的SaaS工具的非常有趣的观点。如果你喜欢这个播客,别忘了在你最喜欢的播客应用或YouTube上订阅和关注。这是避免错过未来剧集的最好方法,而且对播客帮助极大。话不多说,为你请出Shaun Clowes。
产品管理的未解之谜
Lenny Rachitsky: 我关注你很久了,真的很高兴终于能请到你来这里,而且你那迷人低沉的澳大利亚口音更是加分项,我觉得这总是对收视率有帮助。我不知道是否有因果关系,但至少是相关的。
Shaun Clowes: 我很高兴能带来一点新鲜感。
Lenny Rachitsky: 所以我想从我完全认同的一点开始,我很欣赏你对此有非常强烈的观点,即产品管理职业的现状,大多数PM并不那么出色,以及有很大的机会去提升。你就谈谈你的所见所想吧。
Shaun Clowes: 是的,说实话这对我来说是一个巨大的难题。我认为它实际上是……这么说有些宏大,但这算是我毕生事业的一部分。为什么产品管理仍然是一个相对不发达的学科?我们已经在这个领域发展了15到20年。你可能以为它不会像现在这么随机。结果随机,行为随机,个人表现看似也是随机的。因此,产品管理的现状中似乎存在某种问题,没有触及真正重要的事情,真正增值的事情,思考问题的正确方式,理清问题的正确方式,所需的抽象推理,有些地方是不起作用的。我花了很长时间试图准确指出问题所在,然后想,“你如何可复现地复制这种能力?”可复现地培养出真正卓越的产品经理。
十倍产品经理的百倍回报
Shaun Clowes: 事实是,如果你回想一下,我做了很长时间的工程师,人们总是谈论十倍工程师,而我也想成为一名十倍工程师。我让别人来评判我到底是不是,但至少我曾渴望成为,并努力成为一名真正出色的工程师。如果存在十倍工程师(我认为绝对存在),那必然也存在十倍产品经理。但同时,那些十倍产品经理,因为产品管理归根结底关乎杠杆效应,也就是帮助他人产生远超各自为战、无人组织目标与方向时的影响力,这就意味着,一个十倍产品经理能带来百倍甚至更高的回报,因为他们将十倍资源的回报放大了十倍。
所以结果极其疯狂,分布极其悬殊,收益极其巨大,你本会以为这会促使我们去行动。本该有一种方法让这个职业演进、改善,变得比现在犀利得多,但现实并非如此。我不是说我们没有进步,我们百分百进步了,但我认为大家都可以说,我们并没有每一天都稳定地培养出十倍产品经理。
Lenny Rachitsky: 我喜欢这个观点,而且当有人和一个不那么优秀的PM合作时,这尤其令人痛苦。于是就有了那种流行语:为什么我需要PM?PM毫无用处,PM太烂了。这就造成了一种现象,从来没有人说“工程师无用”或“设计师无用”。但很多人会说,“我们团队不需要产品经理,绝对不要招PM”,而这恰恰让整个职业倒退了。
不确定性中的决策乐趣
Shaun Clowes: 当我刚入行做PM时,有人指出——这显然是个老生常谈——实际上作为产品经理,你的工作就是对送到你面前的90%的事情说不。所以这几乎从一开始就让你成了坏人。你对90%说不,才能对10%点头,这在最开始就让你陷入了被动,所以你必须非常迅速地拿出成绩。你必须证明自己有正确的洞察、正确的数据,能做出正确的决策,否则你就没有下次机会了,无法再试一次。所以,产品经理最容易被单拎出来批评,这也说得通,但这同样是它最有趣的地方。
想想我们为什么要做这行?有人曾问我,“你会退休吗?人们为什么做他们做的事?”因为显然到了某个阶段,不再只是为了钱,归根结底,产品管理太有趣了,因为它关乎寻找优势。这就像审视世界,找到棋盘上未被占据却有价值的位置,找到切入、入侵并摧毁它的方法。这是不确定性下的决策,这让它令人难以置信地有趣。真的非常痛苦、非常令人沮丧、很难说服别人,但也非常、非常有趣。基本上喜忧参半吧。
破局之道:向外看
Lenny Rachitsky: 答案是什么,Shaun?我们如何解决这个问题?我知道你说这是你的毕生事业。你发现究竟什么最能帮助PM升级,成为所谓的十倍PM?
Shaun Clowes: 我认为最重要的一点,也是我逢人便说的老生常谈,就是归根结底,你花在向内看上的时间其实对你没有多大益处。人们常引用 Steve Blanken 的话,说你应该把80%的时间花在思考大楼外发生的事情上。你可能人不在大楼外,但你应该把80%的时间花在思考大楼外的事情上。我敢说只有极少数的PM做到了这一点。他们被卷入内部政治,被拖入scrum管理、scrum执行或产品交付等交付环节的事情中,而你这样是赢不了的。你就是赢不了。你永远拿不到A,因为你从根本上就没有解决这项工作。这项工作不是关于执行或别的什么,而是关于寻找你能独特地交付给市场的、可靠的、差异化的价值。
所以我想说,如果有一件事——实际上我通常会指导产品经理做两件事——一是永远从大楼外的视角出发,在每一份文档、每一件事中,始终从客户、市场和竞争对手的角度说话。我敢说,听从我这一点的人几乎立刻就会进步,因为他们从一个更容易被理解的角度出发;其次是要以数据为依据。他们利用所有对世界的看法,但不要只是编造一堆陈述,要用轶事和零散的数据来支持这些陈述。这不需要是一篇长篇大论,而是要让每个人都相信世界的真实面貌以及公司面前的机遇,这样好事就会降临到你头上。突然之间,你就从一个没人愿意帮你做成任何事的世界,变成了一个每个人都希望你赢的世界。他们希望你赢,他们可能不会给你想要的一切,但他们肯定会尝试,因为他们会想,“在我们能下的所有赌注中,这是一个好赌注。”
Lenny Rachitsky: 我猜很多听播客的人会在想,“哦,我就是那个人。我一直在和客户交谈。我总是在互动、看研究、整合数据。”而你的意思是他们可能做得还不够。你有没有什么办法能帮人意识到,“不,你实际上做得还不够,你以为你做到了,但其实并没有”?
Shaun Clowes: 说你花了很多时间向外看是一回事,而从你通常听不到的地方获取反馈则完全是另一回事。所以要避免可用性偏差或确认偏误。大多数时候,人们总是去和他们一直交谈的人交谈,学不到什么特别新的东西。他们不去综合从那些对话中得到的结果,不去寻找反事实,不去寻找证明他们错了的证据。他们不分析竞争对手在做什么,也不思考这必然说明了市场的什么情况。他们不带回关于产品实际如何使用与人们声称如何使用相对比的数据。就像只有数据没有分析是没什么用处的。每个人都能带回一堆“我在某个周二听到的随机信息”大全,但竞争优势来自于弄清别人看不到的东西,弄清我们错在哪里,弄清在哪里精准下注能获得惊人的回报。
所以我认为,首先,人们常说他们做了很多这类事情,但实际上并没有,因为他们没有任何结构化的方法来做。所以他们真正的意思是,偶尔我参加一个客户电话,或者偶尔我陷入一个紧急升级问题。于是他们就很方便地把自己归了类。首先,他们没有用非常结构化的方式来做;其次,他们没有带回分析,没有从中获得真正的洞察,所以他们并没有真正收获多少。只是更多的活动,没有成果。人们做了太多的活动,却缺乏足够的成果,而一天中根本没有足够的时间让你这样做还能取得成功。
Lenny Rachitsky: 作为产品负责人,你正处于近期播客核心话题的维恩图中心,即产品、增长以及AI如何帮助你处理所有这些事情。所以顺着综合与理解人们所说的话、用户研究和调查等这条线索,你和你的团队有没有发现什么工具,能真正帮助你们更高效地做这些事,而不是像传统那样手动翻阅所有东西并寻找模式?
Shaun Clowes: 是的,首先退一步说,回到定性研究那些老生常谈的基础原则,我发现大多数人甚至不理解,或者没有以严谨的基础作为起点去弄清他们需要做什么才能得到想要的答案。例如,你的听众可能听说过尼尔森数字,但其基本理念是,一旦你采访了7到14个人,你就不再学到新东西了。少于7个,你学得不够;多于14个,你也学不到什么新东西了。所以如果你采访了两个人,你可能没有足够的数据。如果你采访了22个,你可能又太多了,所以他们甚至没有合理调整自己的投入力度。这是一个问题,他们没有这样开始。然后他们在这些对话中提出诱导性问题,这些问题实际上是为了让客户说出他们已经希望成真的事情,因此他们要么做得不够,要么做得太多,然后在还没听到任何东西之前就已经毁掉了所有结果。
大语言模型在定性研究中的应用
如果你不合理调整你的研究规模,不把它设定为去学习,那你就会输。应用再多大语言模型(LLM)或任何类型的结构化推理都帮不了你。因为你基本上只是在读回你想听到的,或者某种奇怪的、你想听到的内容的摘要版。但撇开这些不谈,具体到大语言模型(LLM),我喜欢做的事情是,我认为对于产品经理来说,我们现在生活在一个最奇妙的时代,能够分析海量信息并看到共同线索。让我给你举几个例子,一个是你可以做一堆客户访谈,把一堆客户访谈放进ChatGPT里,然后说:“嘿,ChatGPT,这是我的战略。告诉我我的战略在哪里与这些客户谈论的内容不符。”
关键在于“不符”,不是哪里相符,而是哪里不相符。人们花了太多时间寻找他们希望看到的东西,而不是寻找他们没想看到的东西。所以你确实可以请ChatGPT帮你找出客户在哪里试探你试图做的事情的边缘,哪里错了,你说的哪里不是他们相信的。你可以问它这样的问题,比如你可以问它,你的客户所说的什么内容更契合你竞争对手的说法。所以基本上你可以把竞争对手的定位文档复制粘贴到ChatGPT里,然后问:“这与他们所说的相比,比我的东西更契合吗?”也就是你可以总结自己的战略,你可以拿竞争对手的公开文档,让它总结他们的战略可能是什么。
而且它实际上据称很擅长这个,因为你们的公开文档大多数实际上是一个摘要,或者至少是你们战略的衍生品。所以它会给你关于别人产品战略的疯狂洞察,有时甚至令人毛骨悚然,比如:“哦,他们可能会做这个,他们可能会做那个。他们做这个的可能性比做那个大。”通常这类洞察是很难得的,需要大量的苦工,你基本上要阅读大量东西。你不得不把你的大脑当作一台大型摘要机,最终你知道了自己对所有读过东西的感觉,但你无法总结原因。大语言模型(LLM)让你能以非常结构化的方式极其迅速地达到那个状态,但前提是你必须挑战边界,激发出你不想听到的答案,激发出问题,试着向自己证明你是错的,我认为这是开始尝试使用某些此类工具的最简单方法。
Lenny Rachitsky: 我喜欢这点。听起来以你的经验,你只是在直接使用OpenAI、ChatGPT、Claude,而不是用于这种特定用例的任何特定的用户研究工具。
反馈之河与需求洞察
Shaun Clowes: 不,大多数情况下我发现直接使用大语言模型(LLM)本身就足够好了,我们确实有一些围绕它构建的内部工具。我不知道你节目中是否请过Sachin Rekhi,你可能请过。他是一位产品负责人,在增长社区很有名,他曾在LinkedIn担任了很长时间的负责人,他过去把这个概念称为反馈之河(Feedback River)。他基本上说,真正聪明的产品经理总是游弋在反馈之河中,他们致力于让自己被反馈之河环绕,我对此深信不疑。就像,“好吧,我怎样才能让自己被用户访谈数据、直接客户反馈、净推荐值(NPS)数据、竞争对手信息所包围?”我总是试图让信息冲刷我。我这么说的意思是,大语言模型(LLM)及基于它的工具可以在这方面异常出色。
例如,在Confluent,我们收到大量的入站客户请求,你可以想象这些请求来自现场或直接来自客户。我们使用大语言模型(LLM)来接收这些需求,总结它们的内容,寻找与该需求类似的其他需求,以一种真正令人信服的、真实的方式,即语义方式,而不是其他词汇完全相同,这是同一个概念吗?这样我们就可以审视所有的入站需求,然后说:“嗯,最受欢迎的想法是这个,而且越来越受欢迎。最不受欢迎的想法是这个,它越来越不受欢迎。”以一种真正深刻丰富的方式,甚至跨越成百上千条的入站反馈。我认为,如果你能让这些工具发挥作用,现在正是做产品经理的大好时机,但它们不会替你完成工作,它们只是帮助你完成这项工作中那些复杂的部分,即寻找差距、寻找机会、寻找共同线索,而不必把所有这些都仅仅放在你的头脑中处理。
Lenny Rachitsky: 我要继续留在我们现在的这条AI河流中,再问几个与AI相关的问题。这可能正是你刚才说的,但我很好奇是否还有更多内容。你有这样一个热门观点:AI对产品管理影响最大的方式是数据管理,以及数据与你构建的模型或其他任何东西的对比。你能谈谈你在那方面的观察吗?
AI对产品管理的核心影响:数据管理
Shaun Clowes: 是的,我的意思是,我认为对于基于AI构建产品以及在工作流程中思考AI的人来说,有两个启示。让我们先从第一个开始,因为这是产品经理做产品管理事情的方式。你刚才问的问题是,是否应该构建特定的工具让产品经理更容易使用AI?还是实际上是更通用的模型被投入使用?归根结底,这些模型非常非常非常聪明,但它们也极其愚蠢,每个人都知道这一点,极其愚蠢。换句话说,它们真的只知道它们被训练的内容,或者你那一刻带给它们的内容。在那毫秒之间,然后它们就会立刻忘记。你很容易说服自己这不是真的,但这实际上是真正重要的东西。让我再加一点,让这变得真正重要。
归根结底,信息是有衰减率的。想想客户反馈,它是有衰减率的;或者你的竞争对手在做什么,也是有衰减率的。因此,任何新数据对决策的价值衰减得非常、非常快,非常、非常快。如果你愿意,你可以画出自己的衰减曲线,但答案就是非常、非常快。所以,当你思考这份工作——即综合所有这些极其复杂的信息以做出良好决策——时,这意味着什么?你有了这台综合机器,也就是大语言模型(LLM),它将帮助你进行综合,但如果它没有所有这些数据作为综合的基础,它就一无是处。这意味着大语言模型(LLM)只能和提供给它们的数据以及这些数据的新鲜度一样好。它们归根结底就像信息粉碎机。
它们是无限的信息吞噬者。你永远无法提供足够的信息给大语言模型(LLM)以真正获取其价值。你给它的东西越多,它就变得越好。广义上讲,这不完全正确,但也足够接近了。这意味着,作为内部产品负责人,或者要让大语言模型(LLM)投入工作,你需要弄清楚如何引入尽可能多的关于客户、他们的需求或竞争对手的信息,所有这些。你能找到多少信息,把它们汇集在一起,并在你的工具中提供给大语言模型(LLM),甚至只是通过复制粘贴,或者无论你的流程是什么,这是其中一方面。但如果你再进一步想,“好吧,现在我是一名产品负责人,我正在构建一个应用,我想把AI放在我的应用中,什么会让我的AI体验真正出色?”
数据与上下文:AI体验的真正答案
绝对不会是模型,因为这些模型在很大程度上将是可替换的。你可以说,“好吧,那是提示词吗?”也许是,但好的提示词肯定比其他的更好,这是一种你可能想进行的持续投资,以提出更好的问题,让大语言模型(LLM)提供更好的答案。但显而易见,真正的答案是上下文,你要给它的所有上下文,你要复制粘贴的所有数据。所以如果你想想,假设我正在构建一个——我和这个没有关系——但假设我试图构建一个人力资本,就像一个HCM机器人,就像一个AI机器人。假设我在Workday工作,我正试图引入一个AI机器人。很明显,这个机器人的智能真的将与所有员工信息有关,但不仅仅是这些,它将是福利信息,它将是那个人目前工作所在国家的法律状况。
它将是适用于它的公司政策和程序。所以你明白我的意思,关于这种逻辑跳跃和数据跳跃,以及数据连接在一起的方式。如果你想要一个智能的AI体验,你会说服自己,我真正需要做的就是获取一个模型并连接它,我将建立一个小管道,吸取一些数据,然后把它塞进大语言模型(LLM)。如果你这样想,你会非常伤心,在很长一段时间内非常、非常伤心,因为你将不断地纠结于:我如何把数据弄到这个东西里?我如何把好数据弄到这个东西里?我们如何把及时的数据弄到这个东西里?我如何把结构良好的数据弄到这个东西里?
所以这是一个数据管理问题。是获取好数据,获取高质量数据,获取及时数据,并将其提供给大语言模型(LLM)以让它做出明智的决定。这就是90%的卡路里消耗的地方。也许这有点像爱因斯坦的话,“10%的灵感,90%的汗水。”没人想听这个。每个人都只想思考这些真正酷的模型以及它们有多聪明,而下一个会更聪明。但实际上,这只是将非常好的数据提供给大语言模型(LLM)让它们做出好成果的艰苦工作。
Lenny Rachitsky: 听你这么一说,这听起来真的很明显。这让我想起在 Lenny and Friends 峰会上,Mikey Krieger 谈到了他在 Anthropic 内部的两组产品经理团队。一组专注于用户体验产品,另一组从事模型研究方面的工作,他们意识到所有的成功都来自于模型研究工作,比如制作模型以及他们提供给模型的数据才是所有价值的来源,而不仅仅是优化用户体验,他们正把越来越多的产品团队成员投入到这上面,而不是去调整用户体验和按钮之类的。
Shaun Clowes: 是的,完全正确。
AI时代的B2B SaaS护城河
Lenny Rachitsky: 有点相关的事情,我就再问一个AI问题。我不想让每次谈话最后都变成全在谈AI,但最近有个挺火的梗,我知道你对此有看法,那就是AI让构建产品变得非常容易。所以在未来,如果你可以轻松克隆,比如,像 Salesforce 或 Atlassian 这样的 B2B SaaS 应用,或者任何你最喜欢的 B2B SaaS 应用,这些企业长期来看会怎样?它们是不是都会陷入困境?会不会出现100个 Salesforce 的竞争对手?你对那里可能发生的事情有什么感觉和预测?
Shaun Clowes: 是的,我觉得这真的很奇怪。我认为人们真的低估了这些应用中价值创造的地方,他们完全搞错了,我不确定为什么会这样。所以如果你觉得——那是当然。我在 Atlassian 待了很长时间,我做了很多 Jira 的工作,很多人都知道,我也在 Salesforce 待了很长时间,所以我在 CRM 生态系统、营销生态系统和所有其他方面花了很多时间。如果你不想太宽容,你会退后一步看着所有这些应用说,“它们都只是数据库上的表单。”你会说,“Jira 是数据库上的表单,Workday 是数据库上的表单,Salesforce 也是。”它们都是数据库上的表单,所有垂直 SaaS 或商业 SaaS 归根结底都是数据库上的表单。然后你会想,“好吧,复制这东西能有多难?”
而答案是难以置信地难,难以置信地难。人们只会想,“你完全搞错了。”因为这实际上不仅仅是关于数据模型。所以如果你想想,如果是数据库上的表单,那是坐落在数据模型之上的这些美丽用户体验。所以无论对象是什么,它可能是一个客户对象或一个活动对象或一个员工对象,你可以说,“好吧,在对象中有一些锁定的元素,对象本身,比如对象的字段。”我会觉得,“相当无聊。那不是很有趣。”但当然,也许吧。当然,作为记录系统——即每个人使用的默认系统——肯定有一些价值。在用户体验中肯定也有一些价值。比如,“好吧,我想成为处理员工数据的面向HR的最佳应用。”是的,那里有一些价值,但真正明摆着的事实是,这一切都关乎业务规则。
这正是驱动锁定效应的原因,因为你为什么要买 Workday?你买 Workday 不是为了它的开箱即用配置。你买 Workday 是因为你想把它配置成符合 Lenny 公司的 HR 流程。它变成了 Lenny 公司专属的 Workday,不是 Shaun 公司的,而是 Lenny 公司的。实际上,你使用这款软件的时间越长,它就越是如此,它会变得越来越不像 Workday,而越来越像你特定的公司。这是合理的,因为它被构建成可配置的,以满足任何特定公司的需求,而且每家公司都是自己那朵独特的雪花。随着这种情况发生,那些配置部分——也就是让应用程序原生贴合你组织需求的部分——使其不再适合其他任何组织,同时也使其变成一个黑盒,甚至到了你自己都不再理解它如何运作的地步。
比如,如果你去 Salesforce 问:“嘿,你们能定义在 Salesforce 内部销售软件的所有流程吗?”如果不阅读他们 Salesforce 实例的代码,他们根本无法告诉你。这不是什么专有机密,这显然是事实,因为随着时间的推移,销售实际上就是那样发生的。除了通过他们的内部工具,没有其他方法可以完成销售。所以这意味着,重要的不是 UI,也不是数据模型,尽管这两者都非常有用。重要的是产品底层工作流长年累月的演化以支持客户,同时也是客户在演化这些工作流使其按现有的方式运作。那么这对 AI 公司有什么影响?你可以说,“在数据库应用上构建表单比以往任何时候都容易。”
所以我会想,“是的,好吧,这大概会把每一个新入局者的增量价值压到零,对吧?”所以这可能导致现有的获胜记录系统变得更强大,因为会有无数个只是在数据库上做更多表单的竞争对手。就像,“你怎么在它们之间做选择?你还不如直接选赢家。购买 Salesforce 或类似产品从来不会让人被解雇。你还不如直接从顶级供应商开始。”这是一个方面。你也可以从另一个方向看,你可以说,“我听过几个人提出这个论点,”我认为这非常有趣,即归根结底,智能体将会夺走那种用户界面的大部分用途。
举例来说,拿带有 Service Cloud 的 Salesforce 为例,我听人说,“嗯,很多客服人员最终可能会被智能体工作流取代。这就意味着没有人来操作 UI 了。如果 UI 甚至都不存在了,那你为什么还需要 Salesforce?我们还不如直接用原始的数据库表,谁还需要数据库表单啊,你直接用数据库就行了。”但这同样说不通,因为智能体必须按照系统的规则运作,而规则是由业务流程定义的。所以想象一下无头模式的 Salesforce。假设 Salesforce 没有 UI,它依然会有我刚才谈论的那些业务规则。而这些业务规则定义了智能体应该做什么。它们几乎是在告诉智能体它应该做什么,世界可以如何运作,什么是可能的,什么是被允许的。因此,在我看来,认为这会彻底摧毁这类业务流程 SaaS 应用差异化的想法,简直就是幻想,一种疯狂的幻想。
我能真正相信这一点的唯一可能是,如果有人说,“好吧,你可以有一家新创公司,它去探查配置在 Salesforce 中的所有规则,试图逆向工程出你实际的业务流程是什么,然后在此基础上运作。”但最适合做这件事的会是 Salesforce 自己,或者在 Atlassian 的例子中是 Atlassian,在 Workday 的例子中是 Workday。我就是看不出会有这样的世界……我认为可能发生两种情况之一。所有这些向 AI 的转移会让那些应用变得更好,甚至更坚不可摧,它们基本上变得更强了。这让我们变得更强,或者它可能催生某种新级别的应用,这些应用源自更偏向平台的事物,而不是像你的 ACM 或 ERP 或工程领域那样特定垂直领域的东西。
它可能催生一种更偏向平台的玩法,在这个平台上有更多的业务对象,而业务对象具有规则。你可以想象这样一个世界:一种全新的、更偏向平台的 SaaS 应用进化出来,它们涵盖不止一项业务功能的业务规则以及事物在企业中流转的方式,但这在目前还不存在。所以你可以说这种情况可能会出现,也可以说因为 AI,它可能会比我们想象的还要好。或者你可以说,富者愈富。最可能的结果是,目前占主导地位的公司会变得更具主导地位,但我认为那种认为 AI 会直接催生一大堆新应用从而更容易挑战在位者的想法,并没有什么特别的道理,基本上我不明白这怎么会轻易发生。
AI 时代的分销优势
Lenny Rachitsky: 哇,这太迷人了,信息量太大了。我可以往很多方向深挖。一点是我以为你实际上会往这个方向说,即如果复制变得容易,分销优势就变得更加重要。就像今天,我可以坐在那里雇个团队去克隆。克隆 Salesforce 可能需要点时间,但我可以复制它,而等我做完的时候,他们已经进化了,他们在推进,他们在添加功能,他们领先了,对吧?你在滑向冰球曾经所在的位置。所以如果是这样的话,要想取得任何进展,优势之一、方法之一就是拥有某种分销优势。拥有一个 Salesforce 的产品克隆是一回事,让任何人知道它、采用它、销售它、走采购流程等等则是另一回事。在这样的世界里,你是否觉得分销优势会变得更有价值?
Shaun Clowes: 是的,我的意思是,这当然说得通。归根结底,分销始终是一种优势,因为最困难的问题甚至是如何进入任何给定问题的考虑范围。世界充满了问题。只是当人们遇到问题时,他们首先根本不认为他们会去解决它。而当他们确实想到要解决它时,他们不会想到你。所以分销始终是一种不可思议的优势。但同样,在 AI 的世界里,分销似乎更有可能变得更难而非更容易。所以如果你想想,比如,冷邮件的收益递减,因为发送冷邮件变得越来越容易,甚至能发送更糟糕的垃圾邮件,听起来更好听了,但实际上这导致每个人对一切都变得麻木了。我不知道你是否注意到,现在 LinkedIn 上一半的图表基本上全都是明显由大语言模型(LLM)生成的垃圾信息。
数据作为工作流的一等公民
Shaun Clowes: 我的意思是,在某种程度上,它实际上正在恶化信噪比。因此我认为,初创公司通常使用的许多突破性分销机制,总体上似乎正变得更加拥挤和昂贵。这对于“我不如 Salesforce”或者“我不如 Salesforce,但我更便宜”的策略来说,不是个好兆头。它必须有所不同。必须在某个角度上有实质性的优势。我看到正在发生、并且一直看到正在发生的一件非常有意思的事情是,许多现代下一代应用程序将数据作为一等公民带入了工作流。我认为这相当有吸引力。因此,如果你看看处理求职者投递的下一代求职者管理产品,其中很多现在的最新核心产品,包含了你的填补职位时间数据,包含了谁拥有最佳招聘结果的成果数据,谁在什么时间段内有最差的流失率,甚至一直追溯到面试官以及面试在面试周期中的位置。所以基本上是将数据嵌入到整个生命周期中。
因此我认为,初创公司可以通过向世界带来一种不同的方法来提供这些体验上的好处,这确实使它们能够利用传统的颠覆性创新。归根结底,这就是颠覆性创新。这意味着大多数公司已经超越了效用,比如平均效用,所以你可以通过满足平均效用并做到与众不同来获胜,达到标准并做到与众不同。达到标准并做到与众不同是脱颖而出的方法。所以如果这是一个还算不错的策略,这就说得通。但即使是对于这些公司,现在它们也将面临所有这些使用 AI 来加速工程开发、尽可能快地建立一个像它们一样的竞争对手并开始将其塞进渠道的 AI 竞争者。看看这整个事情将如何演变会很有趣。它有点趋于逐底竞争的特征。你可能是对的,分销仍然是软件中最困难的部分,特别是在你刚起步的时候。
Lenny Rachitsky: 因此,如果你有某种巧妙且公平的优势,感觉它会变得更具威力。比如拥有一个受众平台或类似的东西。你提到了这个他们非常喜欢的求职者追踪系统(ATS)产品。有没有哪一个你想表达一下喜爱,认为它非常酷、你很喜欢,还是你想保持匿名?
Shaun Clowes: 是的,是 Ashby。这是现在所有酷小孩都在谈论的那个。有趣的是,人们实际上把它和所有的、甚至是上一代的现代 SaaS ATS 等等进行比较,并且因为它们将数据放入实际工作流的方式,人们对它赞不绝口。所以在你进行工作的应用程序中,行动及其结果是直接相互关联的。我认为这是一个相当有吸引力的用户体验。
Lenny Rachitsky: 那么,在我转向另一个方向之前,也许可以结束这个话题,你提出的关于数据多么有价值、以及这如何成为未来成功和差异化的核心的观点,特别是对于 AI 工具和产品,你会给想要这样做的人什么建议?只是确保你有一个,是有一半的专有数据吗?还是像把它作为一等公民?你会给试图做这件事的创始人什么建议,也就是你正在建议的?
数据是指南针而非 GPS
Shaun Clowes: 是的,我认为归根结底,所有这些都有点关系,不是吗?如果你有第一方数据但不能加以利用,那么它就没多大用处。如果你有第三方数据并且以有趣的方式加以利用,数据的问题是我们一直都被数据包围着。所以数据无处不在。真正重要的是在正确的时间、正确的地点提供正确的数据,因为我们都是人。因此对我来说,显然存在数据优势,甚至存在数据网络效应,如果你能处于拥有非常有价值的第一方数据的情况。但无论如何,这仍然关乎能否在正确的地点、正确的时间为这些用户带来正确的数据,让他们能够从中获得优势。
关于这一点,我想稍微转移一下话题,我知道在我的职业生涯中花了很长时间,奇怪的是,实际上我做产品人很久了,但奇怪的是我最终继承了数据团队。所以我实际上在很多不同的公司运营过数据团队,这很奇怪,因为产品经理通常不负责数据团队。我想我只是对数据有着极大的亲和力。我过去常称自己为数据驱动,那是我的拿手好戏。事后看来,我回顾过去,认为数据恰恰相反。数据更像是指南针,而不是 GPS。如果你把数据看作是给你答案的方式,你总是错的。你总是错,要么就是慢。错误或缓慢,或者有时两者兼有,因为大多数情况下数据不会给你答案。它只是告诉你你刚才说的是否荒谬,或者可能有些道理。
所以它更像是在反驳你的任何想法,而你最终会变得缓慢,因为如果你试图用数据来解决一切,你的大脑最终就是一个数据筛选器之类的。所以你的直觉告诉你的原因是因为你已经看到了大量的数据,告诉你这是最可能的答案。因此,数据驱动、数据痴迷是你很容易过度做的事情,非常非常容易过度。所以关键在于合理界定数据的比重,手头有正确的数据,对数据有正确的看法,而不是期望数据给你答案,或者试图把数据当作武器,或者试图用数据来迫使人们相信你或朝你的方向走。但是数据确实处于一切事物的中心,关乎你如何在你构建的产品中产生影响并取得成功,以及你在内部提出的论点和其他一切。
Lenny Rachitsky: 我很高兴你谈到了这一点。我确实想花点时间在这上面。你这么说很有意思,过去常说数据驱动,[听不清]数据驱动。你创建了 Reforge 课程,面向产品经理的数据课程,还有 Reforge 的留存与参与度课程。顺便说一下,我们会链接到这些。你还在帮助这些课程。顺便说一下,它们还在运行。它们非常棒。人们很喜欢它们。
Shaun Clowes: 是的。
Lenny Rachitsky: 太好了。所以我们会引导人们去了解那些。我很高兴你也说你想,我想这是你之前向我描述的方式,这是你改革的“数据驱动产品经理”。很多人这么说,他们就像,“不要只做数据告诉你做的事。用你的直觉,把它当作指南。”在实际操作中很难将这些建议付诸实施。比如对你的产品经理和团队说,当他们有数据告诉他们,“嘿,这个实验是一个巨大的成功,或者这里有一个巨大的新手引导转化机会。”我想就像,你对那些有数据告诉他们一件事,也许有其他东西告诉他们另一件事的人,有什么战术建议?
直觉与数据的冲突
Shaun Clowes: 我认为我总是鼓励人们去做的第一件事就是看一块数据。如果你看着一块数据,结果告诉你的事情,你的直觉告诉你极其不对,那它们很可能是不对的。首先,相信你的直觉,去证明你是对的。不要只看表面就接受,因为大多数时候就像奥卡姆剃刀,对极其不符合直觉的事情最可能的解释就是它是错的,某个地方出了问题。
Shaun Clowes: 偶尔,有时候你实际上会是对的。那些将是淘金成功的时刻。那些时刻让一切都值得。有时候你确实找到了那个反向目标,你就像,你盯着它看,觉得,“就是这个。这就是问题所在。这就是我们一直在找的东西。”但你必须非常勤勉地跟进,真正理解你所看到的是什么。这个数据有代表性吗?这个数据是我们关心的受众的良好样本吗?它是否已经受到了某种选择偏差的影响?通常当我看到不同的产品负责人甚至数据团队的分析时,漏洞大得能开进一辆卡车,真的是漏洞百出。
Shaun Clowes: 如果你带着权威呈现数据,而那个数据是荒谬的,或者分析漏洞百出,你不仅得不到好处,还会失去一大堆印象分。带着不清楚的分析出现,比带着愚蠢的分析出现要好。我实际上经常看到人们在这个问题上自取其辱,因为他们就像拿着刀去参加枪战之类的,他们确实带了一份分析,但根本站不住脚,然后他们呈现出来,接着在现场被击溃,这绝不是任何人想要的愉快体验。
数据验证的战术建议
Shaun Clowes: 所以如果我给你一些关于这个的额外战术建议,那就是,如果我在看一块数据,这块数据的上游是什么,它看起来正常吗?所以这个你非常兴奋的事情发生了,在那之前发生了什么?这是否与你认为应该发生的情况相符?所以在这个重大情况之前发生了什么?然后,好的,对于你正在看的那个东西,之后发生了什么?如果你对之前和之后发生的事情有概念,那就会让你对这东西是否值得讨论有一些了解。
Shaun Clowes: 然后在你正在看的这个数据之上再往上一层。就像,这些事情,假设我在看新手引导成功率。假设我在看新手引导成功到第二周留存率或类似的东西。我就像,“我发现了这个完全碾压的东西。这个干预措施碾压了它。”如果你往上游走,发现这个干预措施只适用于2%的入站新手引导流,那就毫无意义。这很可能只是一个随机偏差。但即使它不是一个随机偏差,它也不是一个有用的工具。
Shaun Clowes: 所以你必须往上看,然后你可能往下游看,你可能会发现,是的,他们在第二周留下来了,但在第三周他们全都流失了。他们基本上毫无意义。我们为什么要谈论这个?或者你可能退后一步想,“好的,是的,那些人确实留存了更长时间,但他们的平均售价(ASP)更小。”因为我们真正关心的是,我们确实关心参与度,我们也关心更多客户,但我们希望以高平均售价(ASP)留住客户以达到一定的收入目标。最终目标是快乐的客户付钱给我们。所以这就是我说的往上一层看的意思。如果你往左点一下,往右点一下,也就是之前和之后,然后再往上一层,你仍然看到那个能讲出你想讲的故事的东西,那么现在你就拥有了一些非常有说服力的东西,因为人们想听那个。他们想听,“那么,之前到底发生了什么?之后到底发生了什么?为什么会发生那样的结果?”但你真的必须做足功课,真正严谨地对待它,以避免追逐愚人金。
Lenny Rachitsky: 我喜欢这个建议。顺便问一下,ASP 代表什么?
Shaun Clowes: 哦,平均售价(average sale price),[听不清] 月度经常性收入(MRR)或其他一些收入指标。
Lenny Rachitsky: 明白了。你提到的这一点,关于很多时候实验显示积极结果,但最终却什么都不是,我曾在播客上邀请过 Shopify 的增长负责人,他们做了一件非常酷的事情,他们把群组的对照组保留好几年,然后我想是在一两年后自动给他们发邮件,“嘿,检查一下这个,看看这些群组,这个数据是否还是更高。”而在40%的情况下,长期来看,积极的实验结果最终变成了中性。
对照组与短期收益的长期价值
Shaun Clowes: 有意思。这真的很有趣,因为上次我们做类似事情时,我们实际上有一个全局对照组,被排除在所有实验之外。实验平台根本无法定位那个组。所以所有用户中有10%的人从未看到过任何改变。这真的非常有帮助,因为你总是可以把他们与同一时期的任何群组的体验进行比较。我同意你的看法。
Shaun Clowes: 但另一件事是,我总体上并不太喜欢那种思考过程。就像,“嘿,假设一个实验确实显示了短期收益。如果一个实验显示了短期收益,但那个收益并没有永远持续下去,那是否意味着那个短期收益从来就不值得?或者那只是意味着短期收益是一个你没有利用上的、达到另一个层次的机会?”我想说的是,没有一个完美的答案。我认为收益没有永远持续下去并不意味着你失败了。但我同意你的看法,不去试图理解,好吧,净收益是多少,净增量是多少,这也非常重要。这就是为什么增长如此困难。增长作为产品的一部分尤其困难。
Lenny Rachitsky: 那么,你在 Atlassian 时建立了第一个 B2B 增长团队,对吗?
Shaun Clowes: 是的。是的,这让我感觉自己像个老人,但没错,那是很久以前的事了。
Lenny Rachitsky: 或者也许这是一件新事物。
Shaun Clowes: 是的。
Lenny Rachitsky: 要么是很久以前,要么就是最近才发现,在 B2B 中你可以做的事情是专注于增长。
增长黑客的起源与 B2B 应用
Shaun Clowes: 是的。那大约是在2012年,当时增长黑客(growth hacking)很流行。人们现在不太用这个词了,但在 B2C 领域这可是件大事,因为人们能看到 Facebook 做到了“7天内添加10个好友”,能看到这类方法对人们起作用。他们就会觉得,“天哪,太神奇了。”而在 Atlassian,我们开始思考,“好吧,这些技术在 B2B 中奏效吗?”而且,现在看来很明显,很多技术是奏效的,而且值得去做。但在当时并不那么显而易见,因为对于很多 B2B 公司来说,我的意思是,Lenny,你之前总结过,分发掩盖一切缺陷(distribution covers all faults)。几乎所有的问题都可以被极其出色的分发所弥补。
如果你拥有非常好的市场营销,非常好的地面推广(ground game),把产品硬塞进渠道,把产品硬推到人们面前,用客户成功人员、服务、咨询等来粉饰糟糕的部分,人们几乎会购买任何软件,或者说你当然可以靠很多不同的软件取得成功。但回到2012年,情况并不明朗,比如,如果你换一种不同的思路,听到所谓“软件会自我推销”的说法,这样做值得吗(is the juice worth the squeeze)?现在我会说,很显然这是值得的,以至于很多人都在时刻思考这件事,但在当时那是一个有点意思的时期。
Lenny Rachitsky: 那基本上就是产品驱动增长(PLG)的开端。可以这样简单理解吗?
Shaun Clowes: 基本上现在它被称为产品驱动增长(PLG),是的,但在当时我们甚至不知道到底该叫它什么。
优秀的 B2B 增长团队与常见陷阱
Lenny Rachitsky: 只是增长而已。基于那种经验,现在很多 B2B 公司都有增长团队,通过投资增长,在 B2B 中是什么造就了一个优秀的增长团队?你经常发现人们陷入哪些陷阱,认为他们应该尽力避免?
Shaun Clowes: 最终,很多这类尝试都是一个平衡问题。我的意思是,增长团队往往会经历一系列阶段。他们的第一阶段是证明自己的价值。可以称之为淘金热阶段。这东西可能根本不值得做。我们为什么要做这个,这群快乐的人在外面试图证明某个地方存在某种增长效应?这就是证明阶段。那个阶段的好处是日子很好过,因为通常有很多增长可以被找到,因为以前没人去找过,所以日子很好过。但这非常随机,因为你实际上只是在一个随机的搜索阶段摸索,“我们试过X吗?试过Y吗?试过Z吗?”一旦那个模式运转起来,接下来就会变成,“好吧,我们如何扩大这东西?这是否只是昙花一现?我们是否只是找到了一点低垂的果实(low hanging fruit),除此之外就什么都没有了?这究竟只是我们本该完成的一个项目,还是一件持续的事情?”
所以你必须把它变成一个系统。你必须证明它可以重复,然后你必须扩大它。它必须成为一件固定的事。它必须成为你DNA的一部分。你必须用产品驱动增长(PLG)的视角来看待你做的每一件事,从付费获客(paid acquisition)到激活(activation)、留存(retention)、参与(engagement)、跨产品扩展(cross product expansion)、追加销售(upsells),我是说,凡是你能想到的,所有通过收入或参与度来增长产品的不同方式。有很多不同的方法可以做到这一点。所以你最终必须扩展出去,能够做所有这些不同的事情。然后你必须弄清楚你如何融入组织的其余部分,因为有其他人整天每天都在构建产品。
还有其他人整天都在销售那个产品。还有其他人整天都在营销那个产品。所以增长组织处于这个有趣的位置,他们处在其他所有人之间。他们有点像在一点点进入别人的沙坑,他们处于每个人全职工作的边缘,他们非常有价值,但可能会很复杂,因为所有这些关系,以及他们处于组织所有其他部分之间的方式。许多组织失败了,因为他们并没有真正找到多少胜利,或者当他们确实找到胜利时,似乎完全是随机的。或者他们确实找到了很多胜利,但都无法理解,因为它们看起来就像是在一堆潜在机会中的随机游走(random walk)。在经历从尝试想法到成功,再到扩展,再到运营化的增长阶段中,有很多不同方式会导致无法融入。
产品驱动增长(PLG)的真正价值
Lenny Rachitsky: 沿着这些思路,最大的流行说法之一就是很多公司声称产品驱动增长(PLG)几乎从来不起作用。要么你试了但就是没用,要么最终它就是逐渐消失。关于产品有机会成功实现产品驱动增长的迹象是什么,与直接转向销售甚至都不用担心这个相比,你有什么想法吗?
Shaun Clowes: 首先让我们审视一下反事实,对吧?让我们从你问题的反面开始,说,“嘿,如果我们都放弃了产品驱动增长(PLG),世界会变得多悲哀?”我们只是说,“嘿,在 B2B SaaS 中做这个没意义。”问题在于,没有一种自然的力量会把公司拉向在早期考虑终端用户的享受和成功。没有自然的力量,没有自然的链接力量。为什么会这样?我的意思是,常识而言,买家是最重要的人。经济买家是最具经济决定权的人。他们的需求是第一位的。他们通常是推动建议邀请书(RFP)的人。他们通常是跟销售组织打交道的人。所以你听到的那个人的需求通常都是功能驱动的,而且不是来自终端用户的。
所以如果你不考虑那个终端用户,你就在某种程度上播下了自我毁灭的种子。但说你应该考虑终端用户是一回事,有一个系统来做到这一点是完全另一回事,因为人们只是口头敷衍各种事情。但我相信你之前听过这个,在经济学中,人们只做他们的激励告诉他们做的事。泛泛地说,这就是他们做的,这就是发生的事。你得到你着手去衡量的东西。你得到你给人们激励去做的事。如果组织中没有人的真正激励是去衡量他们终端用户的成功、享受、快乐、早期留存和参与,这就不会发生。或者往好了说,它只是一个爱好。因此顺延下去,如果我从那里开始,那么我会说,“好吧,假设它不存在,产品驱动增长(PLG)不存在,因此它是个爱好,因此会有一群关心这个的爱好人士。”
然后你问自己,“好吧,这是否意味着会有很多产品的体验真的很糟糕?这是否意味着这对那些产品的竞争对手来说是一个在那方面做得更好的机会?这是否是一种差异化的竞争优势?”是的,我会说它是。我会说它是。所以我只是倒推回去,我说,“好吧,你可以说你的产品驱动增长(PLG)投资可能太高了。”你可能会觉得,“嗯,如果我投入更多,我也不会得到更多收益。我不能把一辈子都花在试验新手引导(onboarding)上。那不是唯一重要的事情。”这是非常非常正确的,但很难论证它应该被降到零。
PLG 与销售模式的平衡
Shaun Clowes: 因此对我来说,这关乎平衡。这关乎,“好吧,产品驱动增长(PLG)如何与我业务中其他不同的增长方式相契合?”以 Confluent 为例,我们有一个产品驱动增长(PLG)职能。我们确实通过自助注册实现增长。注册的人,真真切切地刷他们的信用卡,很多人注册并且非常成功,从未和我们说过话。我们也有一个企业销售团队,直接向大公司销售,比如一些世界上最大的银行,你肯定听说过的那些。我不认为必须非此即彼。我认为这关乎平衡。这关乎让这些运作模式起效,而对于真正成熟的公司,那些真正掌握这一点的人,这关乎让两种运作模式协同工作。如果你能让产品驱动增长(PLG)的运作发挥作用来为销售团队提供线索,并让销售团队的运作在销售线索尚未成熟时为产品驱动增长(PLG)漏斗提供支持,而且你能让这些运作模式互相配合,你就能赚很多钱。
这可能是一种极其成功的途径,去建立一个极具韧性的业务。为什么?因为你获得了很多客户,也获得了很多收入。如果你有很多收入,但只有少数客户,作为一家公司你不可能那么成功,因为你是被绑定的,每个人都知道这一点。如果你有很多客户,但没有足够的收入,作为一家公司你也不可能那么成功,因为你没有足够的资金来维持运营。所以魔力在于两者兼得,拥有极大量的客户和极大量的收入,这样的公司是很难被击倒的。如果我回顾我在 Atlassian 的时光,我想他们分享了最新的数据,我不记得具体是什么了,但在公开数据中,大概是8万或10万客户,差不多这样。
那是极多的客户。假设你要和 Jira 竞争,你会想,“好啊,我要从 Atlassian 那里抢走1000个客户。”这很多了,对吧?这确实很多。显然1000个客户已经很多了。你只拿走了19个,抱歉,还有8.9万个要走,或者7.9万个要走,或者不管还有多少个。我不记得他们确切的客户数量了,但一家拥有大量客户和大量收入的公司是很难被攻克的。所以这就是为什么我认为产品驱动增长(PLG)作为一种机制,如果你能让其运作模式起效的话,对几乎任何类型的公司都极其重要。显然有些公司并不适用这种运作模式,但对于那些适用的公司来说,似乎值得为此付出努力。
Lenny Rachitsky: 这个回答太棒了。我查了一下去年的数据,他们有30万客户。
Shaun Clowes: 天哪,我差太远了。我离开的时候肯定只有8万客户。
Lenny Rachitsky: 从那以后他们干得漂亮。另外,你在谈论激励以及激励的力量。查理·芒格有一句名言,我查了一下以确保引用正确:“告诉我激励机制,我就能告诉你结果。”
Shaun Clowes: 是的,完全正确。我见过这样的案例,人们试图让销售团队做产品驱动增长(PLG)的运作,你可以随便怎么敲打他们,你可以开个会告诉他们你真的、真的很想让他们这样做,但归根结底,他们不会去做的。其他任何类型的职能也是如此。这就是事物的本质。
职业选择与宾果卡
Lenny Rachitsky: 我有一些关于这个内容的通讯文章,如果大家想深入挖掘的话。另外,Elena Verna 有一期很棒的播客节目,谈论了产品主导销售以及这两种事物的结合,我们会把链接放出来……这完全是另一个我们可以深入、深入探讨的话题,但我们不会在这一期里展开。也许再问最后一个问题。你提到了你工作过的所有公司,你曾在 Salesforce 担任首席产品官(CPO),在 Salesforce 内部的 MuleSoft,Metromile,Atlassian,现在的 Confluent,很多非常有趣且不同的角色。你是如何选择去哪里工作,又是如何选择接受哪些机会的?我想你有很多选择。
Shaun Clowes: 我必须思考一下我的职业生涯。所以事后看来,Lenny,这样看的话,我不知道向前看时对我是否也这么明显。但回首过去,我的职业生涯有点像一张宾果卡。我总是在寻找填补我未曾填过的格子,因为我觉得那会让我成为更优秀的专业人士。就像如果我对某种特定类型的销售模式、那种类型的营销、那种类型的产品管理、那种类型的产品、技术栈中的那一层或那类事物一无所知,那么如果我学习了那件事,我就会变得更加全能。所以实际上有两点:这很有趣,学习新事物很有趣,向自己证明你能做那些新事物也很有趣,然后这会让你更加全能,因为这意味着你面对任何特定的问题时,你都见过能与之模式匹配的经验。
这有点像你最终在刀战中掏出了枪,因为你看到的每一个问题,你都会想,“哦,我从另一面见过这个。我从其他角度见过这个,所以我知道这很可能行得通,而这不太可能行得通。”所以在我职业生涯早期,我在一家大型企业软件公司工作,抱歉,是向财富100强销售的小型企业软件公司。当我加入 Atlassian 时,就像我和你分享过的,我们完全没有销售团队,实际上完全没有。字面意义上没有人卖软件。它自己卖出去,或者根本卖不出去。而我们增长到了8万名客户。这纯粹是产品驱动的增长,真是一家不可思议的公司。然后是在 Metromile,这是一家被收购的消费者公司,为终端消费者制作保险产品。所以他们和技术产品毫无关系,字面意义上是你安装在车里的一个复杂的物联网设备,但最终它是一个保险产品,你可以像卖给千禧一代一样把它卖给佛罗里达的老奶奶。然后是在 MuleSoft,完全是 IT 组织使用的后端软件,接着是 Confluent 的基础设施,世界各地的开发者用来构建真正有趣的、数据驱动的、数据赋能的应用,以实时处理各种事情。你纵观这一切,你会想,“这都有点随机。”但我不是这么看的,因为我学到了东西,我实际上做过一阵子销售,所以我管理过一个售前工程团队,满世界跑着卖软件。所以当我加入 Atlassian 时,我想了解在没有销售团队的情况下大规模销售软件是什么样的,这甚至能做到吗?
所以我在 Atlassian 期间学到了很多。当我去 Metromile 时,我想,“好吧,我以前从未构建过消费者产品。”我可以说我实际构建过触及数百万人的产品,因为 Jira 做到了,所以我对此感觉不错,但我从未构建过一个我可以说,“是的,消费者,普通消费者可以使用这个东西。它太简单了,甚至我奶奶都会用”的产品。我从未构建过那样的产品。所以我在 Metromile 获得了那种经验,这真的很有趣。我从未在像 Salesforce 这么大的组织,或者拥有如此出色销售运作的组织中工作过。你之前谈到过分销。Salesforce 是一台绝对疯狂的分销机器,一家令人难以置信的公司,拥有惊人的分销网络和绝佳的营销方法,简直就像营销学的博士课程。
Shaun Clowes: 当你在 Salesforce 工作时,你会觉得,“这家公司真是独一无二。它就是独一无二,在某一件事上好得离谱。”所以回首过去,所有这些工作,当我说宾果卡时,我其实是在这些毫不明显的领域接受了一种离谱的教育。一旦你见识过它们,它们就像超能力。能够将这些经验运用到事物上,这就是超能力。所以我真正想弄清楚的一件事是,为什么人们通常不这么做。人们经常留在非常特定的领域。他们更喜欢留在某个领域,或者更喜欢留在某种特定类型的公司或以特定方式运作的角色里,比如那些拥有相同运营模式、以相同方式做计划的公司,或者他们试图留在相当相似的事物中。但显而易见,真正成长最可能的途径恰恰相反。
那就是不断地选择那些领域之外的东西,但不是完全越界。如果你以前没跳过伞,就别直接从飞机上跳下去。显然,你希望它们在某种程度上是相邻的,你希望它们与你已知的事物有某种共同点,但你希望它们能拉伸你、改变你。很多很多年前我在 Atlassian 时有过一次真正改变我观念的经历,一个叫 Tom Kennedy 的人,他是我们的总法律顾问,基本上就是首席法务官,干了一辈子的律师,非常聪明的人。我非常喜欢他。但他就是个律师。只是个律师,企业律师,企业法务,我确信你知道他们是什么样。他真是个很好的人。我记得,所以在我们的会议中,除了法律事务他通常话不多。但我记得在一次会议中,我们正在就一个产品战略问题该怎么做进行激烈辩论。我们是向左走还是向右走?
像往常一样,他在那里,基本只是保持沉默。然后对话最终进行了15分钟,他开口了:“嘿,大家,一年前我们讨论过 X、Y 和 Z,”接着他开始列出我们当时的产品战略,他接着说:“就在最近我们说了以下这些,那是一个产品战略,诸如此类。现在你们在说这个。这难道不明显吗?你们说的和那并不一致,如果你们当时真的是那个意思,我们应该做 X。”然后,房间再次安静下来,大家都转向他,点了点头,然后每个人都说:“是的,好吧,我想我们可能确实应该换种做法。”于是,当 GCE 突然提到他深深理解我们的产品战略,并且有足够的见识能以这种方式做出贡献时,会议就停了下来。
所以对我来说,改变人生的部分仅仅是这种认识:如果我要成为一个真正优秀的专业人士,我想成为的那种专业人士就是那种人。那种能以各种方式为整个公司做出贡献的人,不会把所有时间花在插手别人的事上,但了解业务,并且在各种情况下具备让人敬畏的影响力,我是说,这是一种赞美。我不是指负面意义上的危险,而是在各种情况下都能产生重大影响。我认为,当你背后和身边有那样的领导者时,你就是势不可挡的。当那种动态发生时,你在商业中就是一股不可阻挡的力量。
多样性经历的价值
Lenny Rachitsky: 哇,那是一个很棒的故事和很棒的视角。这类似于我总是给产品经理的建议,人们总是想知道,“我应该在特定主题上深挖吗?还是应该尝试不同的事物?”我发现仅仅是多样性,尤其是在你职业生涯早期,真的非常有力量,它不仅能帮你发现你喜欢的东西,而且正如你所说,仅仅是从产品所有这些不同部分、内部工具、信任与安全、平台、消费者产品端、增长以及核心业务中汲取洞察。你拥有的这些经验越多,你就变得越强。我觉得你的方法的另一个好处是,如果你只在 B2B SaaS 公司工作,如果你的简历上有太多这类经历,就很难被消费者公司录用。所以如果你拥有这些多样性,它就会为你创造巨大的选择空间,而你就做到了。
Shaun Clowes: 是的,这很有趣,因为人们过去常谈论 T 型人才之类的,我从来都不太喜欢这个比喻,因为人更像是涂鸦形状的。我的意思是,你共事过的真正最优秀的人,他们比起 T 型更像是涂鸦,因为你当然想具备横向能力,所以你想变广,你确实想变深,但你实际上想在不止一件事上深挖。现在显然,当我说深的时候,我不是说我能每天全天候做我们财务部门的工作,但我百分之百足够胜任,可以在简单的财务分析之下再深入三个层级。我可以合理地深入到我们的财务数据中,因为我想这么做,部分也因为这很重要。能够做到这一点很重要。所以也许换一种方式来思考那张宾果卡就是,我很少后悔在那些不完全属于我职责范围的事情上深挖。
我很少后悔。最坏的情况是,我学到了一些永远不会用到的新东西,我想至少这让我的大脑稍微更敏捷了一点。我不知道,这肯定有某种潜在的好处。但最好的情况是,在未来某个时刻我最意想不到的时候,它会被证明正是关键所在。它会是我需要的工具,当我面临某个重要问题时,我会想,“我的天,这太值了。”所以如果你从投资回报率的角度来考虑,做那些不在你专长领域内的事情,做那些不直接摆在你面前的事情,投资回报率真的可能高得离谱。它可能好到爆表,但我猜这带有投机性。因为你不知道明天是否会需要它。你不知道它是否会成为你经常使用的工具。
宾果卡的终点
Lenny Rachitsky: 有趣的是宾果卡是个比喻。这最后会有一个宾果时刻吗?有退休吗?
Shaun Clowes: 哦,你的意思是你集齐了所有东西。你收集齐了那些宝可梦?
Lenny Rachitsky: 是的,你全都收集齐了。
Shaun Clowes: 是的,我在 Salesforce 曾和一个人共事,他在那里待了很长时间,是个非常非常非常成功的人。老实说他已经不需要再工作了。他说了一些我觉得非常有用的话。他说:“好吧,现在我到了人生中这样一个阶段,我想在那些我擅长的事情和对公司有价值去做的事情的交汇点上工作。”所以基本上,感觉完成你的宾果卡后的奖励,其实就是可以花更多时间做那些有杠杆效应的事情,做那些你享受且具有高杠杆的事情。所以这对我来说似乎是个很好的结果。我不认为大多数人会去工作,并期望获得某种巨大的财务回报然后说,“好吧,就这样了。我要收工不干了,我要退休了。”我认为对大多数人来说,实现某种财务回报或某种独立或其他什么,真的只是另一个阶段。到了那个阶段,问题就会变成:“好吧,那我接下来做什么?我该如何度过我的人生?”
Shaun Clowes: 这就是为什么我早些时候说,归根结底,产品管理有时是世界上最糟糕的工作,有时又轻易成为最棒的工作。它两者皆是,也可以兼而有之。因此,如果去想那些我擅长且对世界有价值的事情的交汇点,产品管理是一件相当有趣的工作,而且每天都不一样。所以我认为我们相当幸运。
对于听众来说,显然你的播客触达了很多产品人。我认为能够在这个交汇点上运作是我们的幸运,但这并不容易,因为你必须展现价值。这是一份在展现价值和向世界证明价值方面非常复杂的工作,并且像你提到的那样,它还不断受到攻击,但当一切顺利时,它依然令人惊叹。当一个产品在市场上非常成功时,你从中获得的喜悦是难以描述的。
Lenny Rachitsky: 沿着这个思路,在进入非常激动人心的闪电问答之前来结束我们的对话,我想带大家进入“失败角落”。人们听这些播客单集,每个人总是在分享各种成功,一切似乎都很顺利。这个的首席产品官(CPO),那个的首席产品官(CPO),一路高升。人们其实很想听事情不顺利的时候,因为那些是人们不常分享的故事。你能分享一个事情不顺利的故事,或者你职业生涯中经历的一次失败吗?如果你从那段经历中学到了什么,也请一并分享。
Shaun Clowes: 我的意思是,有很多事情并没有完全按计划进行,Lenny。在我职业生涯很早期的时候,我还是个开发者,我不小心删除了我所在公司的一个核心系统。那件事绝对会遗臭万年,但幸运的是,它已经远远被抛在脑后了。那个——
Lenny Rachitsky: 那不是在 Atlassian 吧?
Shaun Clowes: 不,那远远早于 Atlassian 时期,但后果非常严重。对,我比较喜欢谈的另一次,我并没有直接责任,但我觉得我有责任。当时我在一家公司,我们发布了一款产品。那款产品属于那种事后看来注定会失败,但出于某种原因,我们都被其潜力蒙蔽了双眼的类型。
那是一款关于,基本上是衡量你公司的环境影响,并通过一些手段来帮助你减少环境影响的产品,你可以把它想象成电源管理、建筑电源管理,管理计算机的耗电量,管理空调的耗电量等等。这基本上就是当时的愿景,就像管理你企业的环境影响一样。这个想法在当时相当酷,而且也确实是做这件事的合适时机,这个领域现在也依然存在。
失败的教训
它现在仍然是一个活跃的研究和投资领域,但这属于那种典型的错误公司、错误地点、错误时间、错误分销渠道的事情。事后看来,我们简直毫无赢的可能,毫无入场的资格,根本就不该涉足那个业务。我感觉非常糟糕,因为再说一次,这是一个好主意,但找错了公司。
归根结底,我们还是发布了那款产品。实际上我们让它在市场上存活了两年,而压死骆驼的最后一根稻草却很奇怪。这最后一根稻草,竟然是终于有一个客户想为它买单的时候。它在市场上待了两年,我们发现面前有一个愿意为之支付数百万美元的客户。他们都已经准备好签字画押了,而那实际上正是我们决定砍掉这款产品的时刻,因为我们当时心想:“如果这个人签了这张纸,我们就要被永远困住了。这个客户将被合同绑定不管多长时间什么的。”
所以我们最终把它砍掉了。在经历了两年的失败后,偏偏在有人愿意掏钱的那一刻。现在回过头来看,我只是觉得:“天哪,那真是一个巨大的……”我感觉非常糟糕,因为我会想:“这本应该是显而易见的。它就是显而易见的,我们本应该能够直言不讳,我想,本应该敢于向上级说出真相。”但相反,这事就那么混过去了,结果在几年里成了真正的意外资源消耗,完全是个巨大的错误。
Lenny Rachitsky: 所以这里的教训就是,对自己诚实一点吗?我很喜欢你提到的那种强制机制,就像是,“好吧,现在来真的了。”是不是可以说,“我真希望我们有一个更早的强制机制,来迫使我们做出决定?”
Shaun Clowes: 是的。我想如果我能重新来过,我不一定能够百分之百改变那个决定,但我本应该去尝试。我的意思是,六个月之后就已经很明显了,这东西有点像个行走的僵尸产品,我至少本可以站出来说:“这东西已经死了。”我们本可以早早地宣判它的死刑,但相反,我们又在这上面砸了一年半的投资。所以正是这一点让我觉得真的很懊恼。
Lenny Rachitsky: 这让我想起了最近一期和 Raaz 的对谈,她是 Wiz 的首席营销官,她当时作为第一个产品经理加入,在进行了几周大量的客户电话沟通后,她说道:“我觉得我需要快点……因为我真的不明白我们在构建什么。我搞不懂。”
然后所有人都说:“我也搞不懂。”就是这样,对,创始人们只是对他们在做什么有个模糊的概念,但他们并没有真正成型的主意。这直接引发了大家的反思:“好吧,等等,原来其实没有人真正搞懂。让我们真正把东西具体化吧。”这帮助他们实现了转型。现在,我不知道你了不了解 Wiz,但他们最终成为了历史上增长最快的初创公司。
Shaun Clowes: 是的。这太不可思议了,对吧?这并不意味着那种情况就是致命的,但提出那个问题并经历那样的面对现实,结果是他们变得更强了。
Lenny Rachitsky: 虽然可怕,但结果往往是最有益的。在我们进入非常激动人心的闪电问答之前,你还有什么想补充的,或者想给听众留下最后的一点真知灼见,一些你认为在结束前可能会有帮助的东西吗?
给产品人的最后建议
Shaun Clowes: 也许有几件不同的事情,我觉得大家有时都懂,但我猜我还是想重复一下,因为它们对我来说非常有价值。第一件是,如果你让日程表统治了你,那么就不会有什么好结果。我知道人们经常谈论这个,但在产品管理领域,人们最终被日程表统治的情况出奇地普遍。
所以这与前面提到的把 80% 的时间花在思考业务外部发生的事情上有关。说来容易做来难,如果你不这么做,没有人会替你做。因此,除非你找到办法强行让这种事发生,否则真的很难取得成功。所以再来重复一点,也有人这么对我说过,我从来没去查证过这句原话,但据说科林·鲍威尔说过,如果你在掌握不到 30% 的可用数据时就做决定,那你就是在犯一个大错。
而如果你在掌握了 70%,或者是 77%——我记不清确切数字了,当你掌握了所有可用数据的 77% 时才去做决定,那你就等得太久了。我一直觉得这番话非常有见地,它也和我们早些时候关于数据的讨论有点关系,但归根结底,我们做产品管理拿的就是做决定的薪水,做好的决定的薪水,拿的是做能带来商业利益的好的决定的薪水。数据太少的决定是致命的。耗时太长、收集了太多数据的决定同样也是致命的。所以这一切,都是为了试图在所有这些不同的事情之间找到平衡,从而试图带来商业优势。
快问快答环节
Lenny Rachitsky: 这是一个很好的方式,把我们刚才谈论的所有事情串联起来。说到这里,我们已经进入了非常令人兴奋的快问快答环节。你准备好了吗?
Shaun Clowes: 准备好了。开始吧。
Lenny Rachitsky: 开始吧。你最常向别人推荐的两三本书是什么?
Shaun Clowes: 是的,都是些经典老书但非常棒,可能算是《精益创业》吧,我依然觉得它真的很好。其中的关键经验我认为对很多人依然非常适用,特别是群组分析那部分,出于某种原因,我依然觉得人们做的群组分析远远不够。所以这就是我的一个小提示。然后是 Marty Kagan 和硅谷产品组写的《启示录:如何打造用户喜爱的产品》。这也是一本经典老书但非常棒。我觉得里面包含了很多产品管理的关键经验,尽管它已经问世很长时间了。
Lenny Rachitsky: 这些都是经典之作。非常棒。你最近有没有什么特别喜欢的电影或电视节目?
Shaun Clowes: 我在看一个节目。我其实没多少时间看电视,大多是在晚上。我喜欢看那种极其轻松的东西,完全不会引发任何压力,而且非常短。所以基本上,短小搞笑的就是我的菜。Netflix 上有个新节目,我觉得叫《底特律人》。
Lenny Rachitsky: 噢,我也在看那个。
Shaun Clowes: 是的,它真的很好笑。我非常喜欢。极其荒诞,但非常搞笑。所以我很喜欢。
Lenny Rachitsky: 那个主角,他太搞笑了。我忘了他的名字。Tim Sweeney 还是叫什么来着。对,他太棒了。好剧。我一直在看,我爱死它了。它非常古怪。我觉得《纽约时报》在上面的引语就是“非常怪异”。
Shaun Clowes: 它真的太怪了。看第一集时我就想,“这是个什么节目?”甚至都不清楚它设定在什么时代,非常怪异。但真的很酷。
Lenny Rachitsky: 是的。很贴切的描述。
最近喜爱的产品
Lenny Rachitsky: 下一个问题,你最近有没有发现什么特别喜欢的产品?
Shaun Clowes: 有的,这个产品可能你的一些听众已经在用了,就是 Glean,它现在是一家相当知名的初创公司了。他们最近融了一大笔钱。我们在 Confluent 已经使用 Glean 很长时间了,它简直令人惊叹。就是令人惊叹。我无法形容它有多好用。我这么说并不轻率,因为我认为搜索,比如企业搜索,可能是计算领域最难的问题之一。真正把它做对是计算领域最难的问题之一。太棒了。我很少使用一个产品时会想,“这东西比以前出现过的任何东西都要好十倍。”对我来说它就是其中之一。
Lenny Rachitsky: 用最简单的话来说,它是如何为你提供帮助的?
Shaun Clowes: 它搜索我们组织的所有知识。所以就像你刚才说的那样,你会问,“AST 是什么意思?”如果我在会议中遇到这个问题,我只要打开一个新标签页,它会自动接管我的新标签页,或者我就直接输入,“AST 是什么意思?”然后它会向我总结 AST 是什么意思,并给我一个链接,指向我们公司内部所有提到 AST 含义的文档,然后它会告诉我谁是公司里的 AST 专家。这就像拥有了第二个大脑。它是一个极其酷的企业搜索工具。
人生座右铭
Lenny Rachitsky: 很棒的提示。好的,还有两个问题。你有没有什么最喜欢的人生座右铭,是你经常跟别人分享、觉得很有用并且在生活中践行过的?
Shaun Clowes: 这个我想了很多。当我刚开始职业生涯时,我是一个典型的工程师中的工程师。我以前非常看重技术上的正确性,看重计算机能做什么,以及技术上的正义,也就是正确答案,觉得世上只有一个正确答案云云。说这么多其实是为了表达,我经常想起这样一句话,那就是人们并不在乎你知道什么,除非他们知道你在乎他们。因此我意识到,真正能够影响别人,其实与你是对是错无关。说到底,这首先是关于信任,关于关系,关于关心彼此的结果是什么,他们的动机是什么,所有美好的事物都建立在此基础之上。一旦你有了这种基础,你就可以建立真正好的合作关系,而好的进展正是来源于此。
Lenny Rachitsky: 哇,这话说得太好了。它和《彻底坦诚》的理念相连,都是关于关心的。人们需要感觉到你深深地在乎他们,然后才会接受你的建议。这也和我正在看的一本叫《倾听》的育儿书有关,是以前的一位嘉宾推荐的,这本书全都在讲当孩子觉得你与他们的联系薄弱时,他们就会出现问题。所以解决方法就是建立更强的联系,让他们知道你深深地在乎他们。所以这真的,把我最近读到的很多东西都联系起来了。
Shaun Clowes: 是的,完全正确。
悉尼旅行建议
Lenny Rachitsky: 太棒了。最后一个问题。你出生在悉尼,大家也许能从你的口音中猜出来。如果有人去悉尼玩,你有什么建议吗,有什么你觉得他们应该去看看的,你在悉尼最喜欢的是什么?
Shaun Clowes: 好的,悉尼真的是一座非常美丽的城市,它以海滩闻名,基本上是一个大都市。你去参观时可能会非常惊讶。它是一座非常大的城市,非常都市化,有点像纽约,但如果你愿意这么想的话,它是有着非常美丽海滩的纽约,这有点疯狂。但实际上在悉尼周围有大量非常酷的自然景观和美丽的事物。所以如果你想做一些不走寻常路的事情,你其实可以去一个叫蓝山的地方,从悉尼开车大概一个半小时,你可以沿着瀑布速降,嗯,实际上首先你要在有水的峡谷里进行溪降,然后最后从瀑布上速降下来。如果你想在距离一个特大城市一小时多一点的地方,寻找一种真正美丽、有趣的冒险体验,那那里就是我的快乐之地。真正美丽的户外活动,同时又紧邻一座美丽的城市。
Lenny Rachitsky: 你刚才说你航行,从瀑布上怎么航行?
Shaun Clowes: 是速降。你可能把它理解为绳降。绳降,我想是的。对,用绳子把自己放下去之类的……
Lenny Rachitsky: 明白了。因为当我听到航行时,我以为是一艘船直接从瀑布上冲下去了。
Shaun Clowes: 噢,不是,是速降,我觉得在美国你们管它叫绳降。
Lenny Rachitsky: 绳降,对。哇。非常酷。Shaun,你太棒了。这次对话极其精彩。非常感谢你的到来。最后两个问题。如果大家想联系你,可以在网上哪里找到你?另外也向大家指路一下你创建的那些 Reforge 课程吧。最后一个问题,听众们能怎么帮到你?
课程推荐与联系方式
Shaun Clowes: 当然。是的,我的 Reforge 课程,你可以在 reforge.com 上查看全部,正如你所提到的,留存、参与课程以及产品经理数据课程,希望能看到大家从中获得一些价值。已经有很多人学过这些课程了,我也真的从中获得了很多价值,因为就像我说的,我的目标之一就是帮助我们所有人成为更好的产品人。我认为我们的杠杆作用可能是巨大的。
如果你想联系我,可以在 LinkedIn 上找到我,当然也可以在 X 上找 ShaunMClowes。至于怎么帮到我,我是说,广义上讲,我总是乐于接受新想法。如果人们有关于如何更好地做 B2B、产品驱动增长(PLG),例如更好的 B2B 产品驱动销售的想法,或者在企业内部更好地进行分发、产品驱动销售和产品驱动增长的方法,嘿,我本人也乐于学习。我们都在同一段学习如何把这件事做得更好的伟大旅程中。
结束语
Lenny Rachitsky: 确实如此。Shaun,非常感谢你的到来。
Shaun Clowes: 太棒了,非常感谢你,Lenny。这次交流很棒。
Lenny Rachitsky: 大家再见。非常感谢大家的收听。如果你觉得这期内容有价值,可以在 Apple Podcasts、Spotify 或你最喜欢的播客应用上订阅本节目。另外,也请考虑给我们打分或留下评论,这真的能帮助其他听众找到这个播客。你可以在 LennysPodcast.com 找到往期所有节目或了解更多关于本节目的信息。我们下期再见。
术语表
| 原文 | 中文 |
|---|---|
| activation | 激活 |
| agentic workflows | 智能体工作流 |
| agents | 智能体 |
| ASP | 平均售价(ASP) |
| ATS | 求职者追踪系统(ATS) |
| availability bias | 可用性偏差 |
| B2B | B2B |
| confirmation bias | 确认偏误 |
| Confluent | Confluent |
| CPO | 首席产品官(CPO) |
| cross product expansion | 跨产品扩展 |
| Elena Verna | Elena Verna |
| engagement | 参与 |
| Feedback River | 反馈之河(Feedback River) |
| ground game | 地面推广 |
| growth hacking | 增长黑客 |
| HCM | 人力资本管理(HCM) |
| Lenny Rachitsky | Lenny Rachitsky |
| LLM | 大语言模型(LLM) |
| low hanging fruit | 低垂的果实 |
| Metromile | Metromile |
| MRR | 月度经常性收入(MRR) |
| MuleSoft | MuleSoft |
| Nielsen number | 尼尔森数字 |
| NPS | 净推荐值(NPS) |
| onboarding | 新手引导 |
| paid acquisition | 付费获客 |
| PLG | 产品驱动增长(PLG) |
| Raaz | Raaz |
| random walk | 随机游走 |
| Reforge | Reforge |
| retention | 留存 |
| RFP | 建议邀请书(RFP) |
| Sachin Rekhi | Sachin Rekhi |
| Shaun Clowes | Shaun Clowes |
| Steve Blanken | Steve Blanken |
| system of record | 记录系统 |
| Tom Kennedy | Tom Kennedy |
| upsells | 追加销售 |
| Wiz | Wiz |
此文档由 AI 分片翻译(translate_long_document)