Check out the conversation on Apple, Spotify, and YouTube
Why PMs should own strategy, not just building (1:49)
Aakash: In the era of builder PMs and Claude Code, everybody is focused on how you build with AI. But for my money, a much more effective use of a product manager’s time than trying to become an engineer is to actually influence the strategy. To make leaps on what problems we should be solving, how those impact the business, and how we do things at the edge of what’s technologically possible. All of that comes back to building a great strategy.
That’s why I’ve brought in my friend of over five years, Tim Herbig. I’ve been following his work on product practice on LinkedIn, and he has consistently been putting out the most thoughtful stuff on how you actually do product strategy in the age of AI. So today we are giving you a masterclass filled with anecdotes from his real work with real PMs, so that you can do better with your strategy work.
This is one of those areas we keep hearing about. PMs, don’t focus too much on execution, focus on strategy. But how do you concretely do that? Today, we’re going to walk you through everything you need to know.
Tim, thanks for being on the podcast.
Tim: Thank you so much for having me.
Cheap features make wrong choices expensive (3:06)
Aakash: We’ve been trading emails for years, so it is my pleasure. I want to get right into it. What do PMs need to know about strategy in the AI era?
Tim: I love that question, and I love even more that I’m just going to show you what I think the real deal has to be these days. The question is, how did we get here? Why do we even debate this? What we really have to understand is what happened over the last couple of years.
Years ago, the limiting factor for most of a product manager’s work was that it takes too long to do this. We had to ruthlessly prioritize what we were working on to spend all those precious engineering resources. These days, as everyone probably knows, producing features has gotten kind of cheap, especially for greenfield products. Even established companies can get stuff out the door much, much faster.
But the downside is that when features get cheap, not having chosen the right feature, funnily enough, gets more expensive. Suddenly you’re flooding your product with all this output. All this fluffy stuff, all these great vibe-coded Lovable things, everything you do to circumvent the complex engineering work or to accelerate your engineering department. This comes at a cost, because you still have to make people use the features you produced. That’s where the expense comes in.
So what I’m observing in the work I’m doing is that companies are excited about amping up the output 100 times. But somewhere down the line, they have to ask themselves, did any of that make a difference? You have to look at it from two sides. Did it make a difference for customers? And even more importantly, did it make a difference for the business?
When we talk about product strategy, and especially strategy and AI, these two terms people love so much, I believe we have to take a step back and think about what we actually mean. What it really comes down to is this. Do we have something in place that helps us say no to things? And do we have something in place that allows us to measure whether we succeed? These fundamental practices and core questions get buried under all the shiny object stuff these days.
While preparing for this podcast, I stumbled upon a quote from our mutual friend Ravi Mehta. He says, “Difficulty prioritizing is often a strategy issue, not an execution issue.” More and more people are actually feeling the pain of that these days. I love it because the quote is four or five years old, and it is as timeless and valuable as ever. It’s not about execution. Execution was never the issue.
The one test every strategy has to pass (5:33)
Tim: So if we dive into that, what is actually the point of strategy? Why do we believe strategy is important? I believe we have to think about product strategy like we think about the product. We have to be clear about for whom it is trying to solve a problem, and how we would know it has worked. We’re told to do this for the product work we do. We rarely do it for our way of working.
If you really break it down, your product strategy has to help teams and stakeholders intentionally say yes or no to opportunities over, let’s say, six to 18 months. Three to six, it doesn’t matter, whatever time horizon works for you. And you know your strategy makes sense if the majority of your team’s effort and energy is spent on work connected to the strategy.
This sounds painfully simple. But believe me, I’ve seen so many companies where, first, this understanding is not in place, and second, they definitely don’t operate like that. So whatever you think your strategy is, whatever shiny deck, whatever one-pager, whatever framework, ask yourself whether it helps you say no to things. That’s all that matters. It doesn’t matter if it looks pretty like in the book, or how much time you spent working on it. The point is, did it help you say no to things?
Aakash: So it sounds like you can’t come up with a list of 14 priorities, which was a strategy I was actually once handed when I was a product leader. You need to come up with things that allow you to say no to a lot of things.
Tim: Exactly. And that’s the hard part, because this removes all the fluff. The shiny object stuff, how it looks, what the process is. In the end, it has to withstand the test. Does whatever we’re calling strategy help the teams say no to something? If it doesn’t, it’s not a strategy.
The core questions every strategy has to answer (9:02)
Aakash: What is it then? Is it basically just a plan, or an idea, or goals?
Tim: The way I would summarize it, and we’re going to get to this in a second, is that you have to make somewhat informed choices around the core key questions of your product strategy. The core questions the strategy has to answer. And your answers to these questions have to be opinionated enough that you can actually enforce the no.
That’s what we’re going to look at next, because the problem is that most of the time people look at strategy as this big monolith. This is the strategy, this is the deck, this is the one-pager. But if you really boil it down, if you look at all the smart stuff out there from different strategic thinkers, it all comes down to the same core movements.
You want to have some kind of narrative in place. A narrative that allows you to express why we should do this, and why we should act on it now. Then you want some kind of playing field definition, which is a fancy product strategy term for the market. For whom are you trying to solve what problems, and who else is solving these problems? Your alternatives, your competition. And then you have to decide, within this narrative and this playing field, how do we plan to win? What value do we provide, and how will we differentiate?
Especially these last two bits are the most affected by AI in my experience, and we’re going to look at this closer later on.
If we take this to heart and say these are the core dimensions every strategy should answer, it hopefully clicks a bit more for you. It means it doesn’t matter what form or visual you use to answer these questions. What matters is that you answer them. Then you can rearrange them in whatever shape you enjoy, and pick whatever component you need. How do I want to describe my strategic narrative? How do I want to describe the playing field? How do I want to describe the winning moves? It doesn’t matter. It matters that you answer them in a way, as we just talked about, that allows you to say no to things. That’s the important test things have to pass.
Frameworks are windows, not the strategy (11:16)
Tim: If we take this to heart, it gets a bit clearer. Look at all these fancy things. The business model canvas, a lean canvas, a strategy kernel, whatever pitch deck or glorified PowerPoint structure your boss loves. All of these are basically just windows into the actual work of strategy.
That’s the difference. Everyone who has worked on strategy will tell you it’s messy chaos. And it’s supposed to be. The pretty, structured, neat stuff you see at the end is just a very curated snapshot into the actual work.
So don’t start with a polished structure. Embrace the messiness. Work through these questions whatever way you have to, and in the end, I promise you, you will always be able to assemble the actual strategy work into whatever shape your boss or your frameworks require.
The fundamental questions don’t change. Strategy creation stays nonlinear and messy. These canvases allow you to look at it, but they are not the strategy itself. So when you are asked about product strategy, don’t fall into the trap of starting with the framework or the template. Start with the core questions every strategy needs to answer, in a way that enforces a no. Then it’s just painting by numbers. You assemble.
Why AI is not a value proposition (12:39)
Tim: You can look at those core questions in more detail. Take everyone’s favorite thing, the business model canvas. I love this post from David Bland, because the question now is, okay Tim, we have all these dimensions. How does AI fit in there? Where does AI influence things the most? I said parts of value proposition and differentiation are the most affected by it.
What I love here is the point that AI by itself in your strategy is not a value proposition. Almost no technology in history has been a value proposition on its own. Instead, technology like AI becomes a resource you can use. A new resource you can leverage to hopefully amplify the core value of why your customers come to you.
They don’t come to you to see the 15th glorified ChatGPT window embedded in a product. Everyone can spin these up. It’s not a differentiator. Everyone has access to the same frontier models or open source models. You can’t compete on that technology alone. You have to make sure other things remain, and we’re going to look later at which differentiators actually remain.
Can AI ever be the value proposition? (13:38)
Aakash: If I just play devil’s advocate here, would there be some companies where AI could be the value proposition? Levels.io has this company called Photo AI that does 83K revenue per month for him as a solo operator, and I guess it’s just about generating photos that look like you. Or maybe ChatPRD. Would you say AI is the value proposition for those tools?
Tim: I would challenge that. What’s important is what is layered on top of AI as a technology. Take the Levels.io product. I think his name is Pieter, the operator who runs it. He has this incredible personal brand. He’s this incredibly well-branded solopreneur operator. People love rooting for him, and people love the products he’s building. If you are above a certain quality threshold, your brand can carry you all the way. By being this solopreneur, he has created his own blue ocean. He’s not compared with the big companies in that space. That’s how I would see it.
Or you mentioned Claire’s ChatPRD.
Aakash: ChatPRD. I was just talking to the president of HP and asked him what AI tools they use for their PM team. He said ChatPRD. I was like, wow.
Tim: My bet is that’s the part. Why do people choose ChatPRD? My bet is they choose it because of Claire’s product experience, which is leveraged through AI in there. In theory, you can write PRDs anywhere. But Claire adds the value of her CPO background to really amplify AI as a technical resource.
Aakash: So it’s still really a resource those people are tapping into. And the strategy, or the business model canvas they’re filling in around it, is their brand, their alpha, their distribution. Those are the things that allow it to succeed.
Tim: I think there was a great podcast, maybe a year ago, from Brian Balfour from Reforge. He chatted with, I think, Ravi Mehta, and they talked about the remaining differentiators in this wild, wild west. They said, essentially, it’s the same differentiators as always. It’s brand and it’s data, and they just get so amplified right now because nothing else is a moat anymore.
I think that shows more and more. You can be an incredibly well-branded company like Notion, or all these other favorite tools we have out there. The brand is amazing, and they have a gazillion data points of history about you to leverage AI into a fantastic product experience.
Using AI to build strategy vs AI inside your strategy (16:18)
Aakash: Very interesting. So where do we go from here?
Tim: Before we dive into the case study for today, I want to lay out a few things for people to understand. What are the connection points between AI and strategy? On the one hand, of course, you can use AI as a tool to create your product strategy. The other part is where AI appears in your product strategy, which is what we talked about.
To point out some of the obvious things about using AI to create your product strategy. Researching markets and competitors is an obvious one. Synthesizing customer evidence. Generating and comparing scenarios. Pressure testing strategic scenarios. These should be part of your workflow. It would be stupid not to do it. So don’t put on your blinders. Be aware of these tools being available to you for strategy creation.
The thing you can’t replace, which people often fall for, is that you will still be held accountable for the consequences of your strategy. You can’t say, “ChatGPT told me this is a great strategic bet, it’s not my problem that we lost 10 million in the market.” You can’t do that these days.
And when we look at AI as part of your product strategy, as we talked about before, you have to come to some core realizations. AI as a technical resource on its own is not a moat. The resource is widely available. It’s not differentiable. You have to understand how you can use AI as a resource to amplify existing value propositions, and to become appealing to new audiences if that’s what you want to do.
And you have to understand that as easy as it is to spin up AI and use it as part of a product strategy, because it’s so widely available, the only sustainable differentiators in product strategy are the ones that take the longest to build. We just talked about brand and data. You cannot speed those up. You cannot speed up brand and reputation. You can destroy it incredibly quickly, but you cannot build it up quickly.
The Roger Martin test (18:41)
Tim: From here, we can dive deeper into how to work through product strategy. One of my favorite things I use to pressure test whatever people put in front of me is a quote from Roger Martin. Honestly, it’s one of those light bulb moments. Whoever I show it to will never look at strategy the same way afterwards. You just can’t.
He says the question is, is the opposite of your strategy choice stupid on its face? If the opposite is stupid, you haven’t made a real choice. He uses examples like this. If you say, “Our strategy is to make our salespeople more efficient,” what’s the opposite of that? To make salespeople slower. That’s a pretty stupid idea.
Aakash: So then you haven’t really made a strategic choice, because anything could fall in the bucket of making salespeople more efficient. You need to find something where the opposite might not be stupid. Like, we are trying to make salespeople spend less time in software and more time with people. I guess the opposite of that is also stupid, though. Spend more time in software and less time with people. So it’s not easy to generate off the top of your head.
Tim: Yeah, but that’s the thing. If it were easy, strategy would be easy, and we all know strategy is not easy. I get what you’re saying. You can easily tweak this. You can say we want to make salespeople more efficient, but then the question is how? That’s what strategy should answer.
You could say we want to make salespeople more efficient by shortening the preparation time for meetings. That could be something, if you believe it’s the best you have. An alternative would be making salespeople more efficient by sending more AI bots to sales calls. Who knows if it’s the right one, but it’s one you could do. Or you could even say we want to make salespeople more efficient by having every sales rep focus on fewer accounts. How do you do this? That’s the part strategy should be able to answer.
Aakash: So it’s almost like having a point of view about not just what your product is going to accomplish, but how you’re going to accomplish it.
Tim: Exactly. How do you want to do this? And we all know strategy is not just this one sentence. A strategy is never just make salespeople more efficient. It’s much more complex, and we look at the individual building blocks of strategy. So don’t get frustrated about your product strategy being stupid on its face. Pick out an individual component of your strategy you can work on.
The stupidity flip (21:20)
Tim: One person I work with, when I showed this to her, started calling it the stupidity flip. I love this. It’s one of my favorite exercises. She basically decomposes the strategy. For example, part of your strategy is making choices about your audience, or choices about how you want to win the playing field we talked about.
She worked on a meal kit product. She said, I can make a choice by saying the audience I focus on is people who regularly cook at home but struggle with inspiration. I could flip this. I could also focus on people who never cook at home, or people who are extremely proficient home cooks. Legitimate alternatives. No one knows if it’s right, but I have made a choice that makes sense.
She could also do it for the winning moves. My value prop will be end-to-end home cooking, from inspiration to inventory. You could also say, I’m going down the HelloFresh route and I’m going to be a meal delivery service. Or I’m going super high-end, I deliver complex recipes with specialty ingredients, and that’s it.
So you can pick individual parts of a strategy, flip them, and ask whether you can make one part more tangible. Can I make one part of my strategy more no-enabling, so to speak, so that I can then work with it?
Aakash: Just to clarify, because I didn’t totally follow. Where is the flipping happening here?
Tim: The flip happens like this. One part of your strategy is that you have to make a choice about the audience you’re focusing on. You want to say for whom you’re solving a problem. That’s a strategy component you would focus on. So you make the choice, say, I’m going to focus on people who regularly cook at home. What could be flips of that? I could flip it into people who never cook, or people who are extremely proficient.
In this case, the stupidity flip worked. These are actual choices. They’re not stupid choices. You could focus on them, but you chose to focus on these people. So that’s the part of the strategy you’re flipping. You pick an individual component and you flip it.
Does a strategy have to cover everything? (23:25)
Aakash: Just to play devil’s advocate on this now. I also felt this sometimes as a product leader implementing a strategy. One strategy I had at Apollo from the CEO was, let’s only focus on paid users. I think he was going off a similar sort of reasoning, that there needs to be a stupidity flip. But the strategy kind of ignored free users. As a product leader, I can’t just ignore free users. So I would have to make decisions that don’t just nerf the free users. Is there a component where strategy should be giving you myopic focus on certain segments of users? And is that good?
Tim: I would say it’s two things. The first thing people have to understand about strategy is that it does not have to cover everything you’re doing. It’s okay that, as you said, the focus might be paid users. That does not mean you’re turning off the product for free users. That’s the first thing.
The second part is that it has to fit the strategic narrative we talked about before. What is the reasoning for focusing so strongly on paid users? Is it because the company has a retention problem with paid users, and increasing retention is the biggest bet for improving revenue? What’s the bigger thinking behind it? The choice of a segment has to match the narrative. These components have to form a coherent net. They have to make sense towards each other and reinforce each other. They don’t have to be right, because no one knows what’s right anyway. But they have to make sense.
So if you told me the company’s biggest strategic priority is to grow revenue by improving retention for paid users, and therefore we’re going to focus on acquiring more free users, I’d say, well, I don’t know how the math on that checks out.
Aakash: 100%. Makes sense. And that actually was the reasoning. They wanted to focus on retention as the biggest lever. So it’s interesting that just by knowing paid users were the focus, you were able to figure out the larger narrative.
Tim: Exactly, and that’s the beauty of strategy. It’s not a top-down thing. You can start anywhere. Someone could throw at you that the focus is on a segment, and all you have to ask is, why this segment? Or someone might ask, are we really trying to differentiate through in-store experiences? Okay, great. Who will we reach through in-store experiences, and is that the segment we want to focus on right now? Because all of these things are a web, you can start anywhere.
Running the flip on real choices (30:11)
Tim: To make this all a bit more tangible. We already ran through some of it, so I don’t have to repeat it. The choice might be, make salespeople more efficient. The opposite would be make salespeople more inefficient. Obviously a stupid idea.
Now say, make salespeople more efficient by being in person with clients. The opposite would be making salespeople more efficient by being in virtual contact with clients. That’s not obviously stupid. It could be a legitimate alternative we could focus on.
Aakash: Yeah. So now you’ve made a real choice.
Tim: So you flip it, and then you ask, is it stupid? That’s the simple way to validate.
Aakash: That’s the stupidity flip.
Case study – Fair Mile, a used car marketplace (30:59)
Tim: Yes. And to make this more tangible, we’re going to work through a fictional case that represents lots of the clients I’m working with. It’s a fictional marketplace product called Fair Mile, and it’s essentially a used car marketplace.
What we have here is your traditional marketplace situation. You have sellers and dealers, you have Fair Mile as the company in between, and you have buyers. Sellers and dealers want what everyone wants. They want qualified buyers, they want to sell the car quickly, and they need low-friction listings and verification of whatever they’re putting up there. We, as Fair Mile, have to facilitate all this. Inspection, creating trust, connecting supply and demand.
On the buyer side, we have people who come through search, through Google or AI search. There might be first-time buyers who are very uncertain. And of course, it can be difficult to purchase a car sight unseen online. They don’t trust the quality of the car enough to commit remotely. So that’s a bit of the landscape.
They have a few interesting challenges ahead. They could do what everyone says. Throw AI everywhere into the product, expand to all the countries and all the categories, accelerate growth, try to build everything. We talked about this before. This is an all-yes strategy. It doesn’t let you say no to anything. If there’s a suggestion, do it.
Aakash: You just say no to…
Tim: Yeah, exactly. Just say no to sleep, say no to capital, say no to margin, say no to quality.
Then you could make a specific focus. We could pick first-time buyers. This would be another example of the stupidity flip working. You could say we focus on first-time buyers. You could flip it and focus on second-time-and-beyond buyers, so experienced car buyers. It’s a choice. So the question for Fair Mile is, what do we do, and what do we choose not to do? For example, if we chose trust, to make sure we actually focus on these first-time buyers. That’s the case I want to use as an example to work through.
Stress-testing Fair Mile’s strategy statement (33:21)
Tim: You talked a bit about the questions we could work through. Take some of the choices Fair Mile has already made. They might say the strategy is to become Europe’s most trusted used car marketplace by using AI to deliver a seamless buying experience, delighting buyers, blah blah blah. And you could flip all of that, and it’s obviously stupid. You don’t want to be mistrusted. AI or not doesn’t really matter. You don’t want to create a hard-to-use buying experience. You don’t want to not delight buyers. It’s nothing. It’s not a strategy. It doesn’t give you anything.
Europe’s most trusted car marketplace. Flip it, and it’s Europe’s least trusted used car marketplace. You shouldn’t do that. Obviously stupid. Buyers and sellers alike. Well, of course, without buyers or sellers it doesn’t make sense. No buyers, no sellers. Also obviously stupid.
Aakash: Yeah, exactly. That’s not a marketplace.
Tim: But you could look at some of these other things. Accelerating buyer growth through automated AI verification. You could flip it to taking our time on buyer verification through human call center verification. You could do it.
Aakash: Doesn’t seem super stupid.
Tim: Yeah, it could be. It’s not obviously stupid, because if you really want to become trusted, it might help you with your trust play. So this might actually be a bet we could try out.
Improving the recommendation algorithm through an AI-enhanced start page feed for first-time SEO visitors. Well, flip it to making the first-time start page feed experience worse for SEO visitors. That could be one flip, but maybe that’s a bit too much devil’s advocate. You could also say, provide a static start page feed experience for first-time SEO visitors compared to an AI-optimized one. Somewhat possible.
Audience trees (35:43)
Tim: So there’s something in there. Now let’s take a look at some of the components of your product strategy and what you can do to break things down. We talked about audiences and segments a lot, and one of my favorite ways to work on them is an exercise called audience trees.
Audience trees borrow from an idea from Andy Johns, who’s a former head of growth at Facebook. He says your target audience has to be comically narrow. Almost ridiculous how narrow you are. And I think you have to land somewhere between very broad and comically narrow.
Take the Fair Mile example. All potential car buyers in Europe. That’s very broad, probably a bit too broad. We could narrow it down to first-time buyers of vintage Mercedes cars in southern Germany. Okay, that’s oddly specific. That’s probably comically narrow.
What you want to do with an audience tree is figure out the dimensions by which you could break things down. An obvious one is region. You could do something like income. For cars, car value might be important, people looking for high-end cars or low-end cars. We already established first-time versus repeat buying as another category.
What I recommend companies do is take the broadest one and maybe a comically narrow one. Then think about the aspects by which you could slice your audiences, and rearrange things.
So let’s say Europe. Europe is great, but maybe we love Germany because they’re crazy about cars. So we have Germany, maybe the UK, and then the rest of Europe. That could be a segmentation you play with. Region is out of the way, so we put it up top and say we’re going for Germany.
Within Germany, what could I do? I could split between first-time buying and repeat buying, because Germans love cars and often buy new ones. That could be another segmentation question to break things down. I might say I don’t want to deal with the hassle and inexperience of first-time buyers. I want to go for repeat buyers. They might be easier to convince to buy online.
Then you might say I also want to make a big margin on each transaction. I don’t want to play the volume game. I want quality over quantity. So by car value, say above 100K and below 100K, we focus on above 100K, because then every transaction is really worth facilitating.
The beauty is that you can rearrange these things however you want. You could start with car value and then do region. It doesn’t really matter. It leads you to the same place. In the end, what you land on for Fair Mile is that we focus on repeat buyers of cars worth more than 100K in Germany. Now you have something you can say no to. Something that gives you focus. And if it fits your narrative, hopefully you have a narrative for why you made these choices.
This is also where AI is really helpful, to assess these criteria. How big is the market? What’s the difference in trust issues for repeat versus first-time buyers? How many cars do competitors list above or below 100K? You can speed up all this research. But in the end, you land on a segment you can actually work with.
Sizing the segment with AI (39:42)
Aakash: And how do you know that segment is big enough, and not comically narrow?
Tim: That’s what I’m saying. Of course you have to do this. I don’t want to call it a sanity check, that sounds too easy. But you still have to ask, is Germany a good market? Or would you say, let’s do Luxembourg? I don’t want to do Luxembourg an injustice, but there are maybe 10,000 people there. It doesn’t make sense. So you still have to go through what it means.
That’s where the acceleration of this breakdown can happen. You can easily get access to how many car owners there are in Germany, how often a German buys a new car, and the average value of the cars they’re buying. You can get that data very quickly. You can run this decision through some AI-assisted process to back up the data and make sure. Okay, this is a market of, I don’t know, 50,000 people we could reach every year at this expected transaction size and the margin we’re taking.
Then you see whether the numbers make sense. Is this a business model, or do we need to broaden it? Maybe it can’t just be Germany. It has to be Germany, Austria and Switzerland. You can tweak the dial, so to speak, to make the choice.
Aakash: So AI really helps in the market sizing component.
Tim: Yeah, exactly. And again, what I’ve been banging the drum on, you now have something you can say no to. You can flip this and the opposite would still make sense. We could go for first-time buyers of lower-value cars in different markets. It allows us to say no to things.
Getting your audience in place has so many downstream consequences. If you have your audience dialed in, the problem space you focus on becomes very focused, because you can only solve problems for the audience you’ve chosen.
In the marketplace scenario we have another interesting component, because we also have to say we focus on sellers of cars worth more than 100K. Maybe not only in Germany. We could sell across Europe. But then we have the buyer and the seller side clarified at the same time. We have this lock of two strategy components we can use for the team working on the buyer side and the team focusing on the seller side. We know what we focus on in acquisition and how we will measure success.
That’s why I recommend so many people actually get the audiences right. Don’t get lost in a comically narrow one. Don’t go for vintage Mercedes cars in southern Germany. But do the run-through, do the slicing, and focus.
Differentiation and the 7 Powers in the AI era (42:16)
Tim: Another strategy part, and I teased this already, is winning moves and differentiation. I believe it’s the trickiest one and the one most influenced by AI. I love going back to the original concepts out there. I don’t believe anyone needs to reinvent strategy work. What we do have to do is come back to the fundamentals and stay true to them.
One of the most timeless concepts in differentiation is Hamilton Helmer’s 7 Powers. If you haven’t heard of it, I recommend you go read it. Timeless stuff, really good. If you boil it down, he says there are seven things you can do to differentiate from your competition. You have to understand what each of these powers does in general, whether it’s feasible for your company to execute, and most importantly, whether it even helps you differentiate from your competition.
To give you some color on what some of these mean. Switching costs is probably the most painful one for everyone. The blue bubble problem. You can’t leave the iOS ecosystem, you don’t want to leave all your iOS data behind. Or switching from Spotify to somewhere else, with all the playlists and all the algorithm stuff.
Process power. What kind of internal efficiency do you actually have, and can you have that in place?
Counter-positioning is a very interesting one. My favorite example across history is Netflix versus Blockbuster. Blockbuster made a lot of money on the late fees for rented movies. When Netflix waived late fees, Blockbuster couldn’t compete, because their business model relied on them.
And of course we have brand, and we have network effects.
So what I said before comes true. The only sustainable differentiators these days are the ones that take the longest to build. On the one hand, as we talked about, you have brand. Hard to build. You have network effects, because you can’t copy and paste network effects. Whether that’s offline distribution or the number of buyers and sellers on the marketplace. This would really work for Fair Mile.
Switching costs are probably not a good one for a car marketplace.
Process power. How easily do your internal wheels turn? Companies can catch up to that. It might take them a while, but it’s copyable.
And something like cornered resource, which is an old-school way of using technology. Not sustainable. We talked about this before. Everyone can tap into the same models, so this is difficult.
So if we look at Fair Mile again, it’s brand, and it’s network effects, really getting the buyer and seller sides in place and locked in. Again, I don’t see switching costs. They shouldn’t focus on building switching costs for a car marketplace, because it doesn’t really make sense. An interesting one for them could be economies of scale, maybe by working with dealerships in Europe. Can you get dealerships in place with multiple offerings and multiple locations that really blow up your inventory? That could be a really interesting one.
And cornered resource, just going crazy about being the car marketplace with the AI search, is probably not sustainable. So looking at the winning moves and the seven powers, I think for them it’s about brand, network effects, and maybe getting the dealerships in place. They shouldn’t get lost trying to build a fancy AI search. They might do it, it’s not painful, but it won’t really help them differentiate. No one will buy a car with you because you have an AI search on the marketplace.
Is there a sequence to strategy work? (46:31)
Aakash: Does 7 Powers come after audience? Is there a sequence to this that makes the most sense when you’re building your strategy?
Tim: I would say there isn’t. If you look at the core questions we talked about, you can start anywhere. Joe from marketing might come in and tell you all about how we now differentiate on price. Okay, thanks, that’s a bit hard, but okay. The big question is, do we focus on a price-sensitive audience? What’s the link? Is that a link in our strategy that actually makes sense? That points to who our audience is, and even further.
Once you understand what makes your audience choose a solution these days, you can take it further and ask, why do customers choose our competitors? If it’s price, I would say get out of that market, because price is one of those things you cannot win on.
There are outliers, of course. These days we see Google, Microsoft, Anthropic and OpenAI competing on price for a while by cutting costs and not charging you the real AI cost. They can, because they can swallow it and subsidize it. But no one else has a business that can subsidize these incredible costs you don’t have to pay for.
So to come back to your question, it doesn’t have to be a sequence. You might start with a wild differentiation guess and link it back to all the other strategy components you should be able to answer. Does that make sense?
I dove a bit deeper into audience and differentiation because, on the one hand, audience is where you can do a lot of AI-assisted work to really nail things down. Differentiation is the one most affected by the question of whether you can actually use AI to differentiate your product anymore, and what that means for the other differentiation possibilities you have as a fundamental layer.
The toolbox for narrative, playing field and winning moves (49:00)
Tim: We talked about how you link all these things. Let’s talk a bit more about what you can do to really articulate them.
For your narrative, you can use things like your vision, your mission, your business model, your company strategy if you have one in place, or a North Star metric. These are components you can use to describe the strategic narrative you’re going after. You don’t have to use all of them. Please don’t use all of them. But you can.
When we describe the playing field, we have to be clear on buyers and users, especially in B2B where buyers and users differ. We have to be clear on the jobs, the buyer jobs and the user jobs. And we have to be really clear on the alternatives. What are users and buyers using to get the job done today? These are things we need in place to articulate our playing field.
For the winning moves, you can talk about things like the value prop, distribution channels and differentiators.
These are the things in your toolbox you should be able to pull out to articulate the winning moves, the playing field and the strategic narrative. As I mentioned before, all of these are inevitably linked. You can’t make a choice about distribution channels without being clear on the users and buyers you’re serving. You can’t make a choice about differentiators without knowing who you’re competing against. All of these things are incredibly interlinked. You might start anywhere, but be prepared for the messy work of figuring all of it out together.
Why the North Star metric gets misused (50:55)
Aakash: When you were talking about the strategic narrative, you said don’t use all of vision, mission, business model, North Star metric and company strategy. Why not?
Tim: I believe you can get lost in them so often. My problem is that people, and especially companies, love these terms, but they’re rarely used the way they are intended. Take the North Star metric. It’s one of those things people put on the wall, and they say EBITDA is the North Star metric. Well, it’s not.
Aakash: It’s an outcome, I guess. A lagging metric, right?
Tim: Yeah, exactly. That’s the problem. Again, I love going back to the inventors of concepts. Sean Ellis, who came up with the North Star metric, had a very clear definition. A North Star metric is the metric that represents the value your customers experience from your product. That’s what it is. Of course, an increase in this metric has to lead to an increase in the business metrics. But the business metric itself is not the North Star metric. That’s so easy to forget.
So coming back to your point. People love to debate. Is this a vision, is this a mission, what is it exactly? It’s not written correctly, and so on. I don’t really care what you use as long as you can tell me why we should act now and why it’s us who should act. Whatever you use up here, only the North Star metric, only the vision, I don’t care. Your users don’t care about your vision or your strategy. Your colleagues and your teams care about your strategy. So if you can inspire them to think, yes, this is why we should act now, just with a vision, be my guest. I’m not going to police your strategy framework.
Aakash: Makes sense.
How to know if your strategy worked (52:48)
Tim: We’ve talked a lot about strategy, and now we have to talk about whether the strategy has worked. I mentioned this honeymoon period many of my clients are in. Let’s have so much output, everything is happening now, everyone is more efficient. But I have so many clients who told me a year ago they have issues with feature adoption. They have issues getting customers to use the features. So tell me exactly how 10 or 100 times more features helps you with adoption. Honestly, I haven’t heard a good answer to that. You might throw more things out, but you don’t really know what happens.
The problem is people then say, let’s write more OKRs, and they get lost in the metrics frameworks. And just like with every framework, it’s important to go back to the basics. If we go back to the basics, the question is, how do we know the strategy has worked?
I don’t tell people to write OKRs or North Star metrics, because then people focus on the framework. What I have them do is say it in plain English first. Tell me what you want to do differently in the next quarter or six months. Just tell me what you will do differently, which is the essence of strategy. Don’t talk to me about business as usual. Don’t talk to me about bug fixing. What will you do differently?
Aakash: Differently from what you have been doing.
Tim: Yeah. What will change? You cannot expect a revenue increase of 20% by continuing to do the thing you did before. You probably have to do something differently. This applies to B2C, B2B, customer-facing and internal teams. It doesn’t matter. Everyone needs an answer to the question, what will I do differently in order to achieve the business goals I’ve been given?
Once you’re able to say that to me in plain English, then we can talk more. Now tell me how you will know it has worked. How will you measure that what you did differently has worked? Once we’ve had that conversation, we can translate it into whatever metrics framework your boss wants you to fill out. A dashboard, OKRs, a balanced scorecard, I don’t care. It’s like the strategy template. Filling out the template doesn’t make it a strategy unless you get the fundamentals right. For metrics and OKRs, it’s always what’s different, how you will measure it, and then translate it.
Tokens used is not a value exchange (55:22)
Tim: Teams struggle with that, especially in the AI context these days. Companies get easily lost in very machine-centric metrics. They tell me about tokens used, language model efficiency, blah blah blah. The point is, that’s no value exchange. Tokens used is not a value exchange for your customers. It’s a value exchange for Anthropic. They make more money from you using their tokens.
So you still have to go back to the core question. We’re building a product ultimately for humans, for now. Maybe agents in the future. But explain to me whose behavior you want to change. Whose behavior has to change to tell you that this idea worked, and by how much? That should be the core question.
Even if you tell me we want to increase AI feature output by 20x next quarter. Okay, what should change for whom as a result? Why do you want 20 times more? Is it about an internal metric, like people being more AI fluent? Or do you actually want customers to be more effective, more efficient, or faster at solving their problems? Make that the actual goal, how you measure the success of your strategy, and not a random AI output metric.
Leading vs lagging metrics (56:38)
Aakash: So is it better to have a metric that’s more of a leading product usage metric?
Tim: That’s a good question. There’s this great quote, I don’t know who said it first. There are no solutions, only trade-offs. There is no one perfect metric. Ultimately you have to decide. Let’s say you pick the most customer-centric metric, but it takes you nine months to figure out if it has changed. That’s not a good metric. So you have to reverse engineer. You go from lagging to leading indicator. What is the earliest signal I can detect?
It’s a little bit like what we did with the audience tree. You might have a comically narrow metric, like number of tokens used per day. It’s super leading. You can track it every day. But it doesn’t tell you anything about the value you’ve produced. That’s the balance we have to strike. On one end of the scale you have very lagging customer behavior change, and on the other end you have this very leading work output indicator.
There might be a time when you have to work with the leading work output indicator. But honestly, my biggest pain is always this. Why do we do all this AI stuff if it doesn’t get products into the hands of customers faster, and therefore let us measure things faster? If it still takes you nine months to measure customer behavior changes, but you’re burning through all these tokens, what exactly is the point of that?
Building the strategy and metrics loop (58:30)
Aakash: So how do you create the right strategy and metrics loop then?
Tim: We’re going to package things up and revisit some of the things we talked about before. We have this fundamental idea of what we do, and we can go through the full picture. This is what our strategy should answer. It needs to give us context on why us and why now, for whom, which problem, and so on.
Let’s skip the narrative for now. We’ll use the segment choice from before as the playing field part of our strategy. The core question to answer is which metrics will tell us that this strategy choice has worked.
Of course, just having the segment doesn’t make it a strategy. We still need to make a choice about how we will serve this market and win it. Let’s take a bold bet. What’s the value we provide? We provide white-glove on-site inspection and car delivery services. Maybe this is the winning move we believe will help us win this market. Who knows if it’s right.
The last thing we have to put in place is why we want to focus on repeat buyers of cars. Coming back to the example you gave, the narrative in this case might be to grow company revenue without losing margin by focusing on high-end markets. So maybe these are the three things we have in place, and we have a somewhat coherent strategy.
Now, what kind of metrics could we set up to tell us whether this strategy has worked? Something simple, like share of cars sold above 100K. That would be an obvious one. Average profit margin per transaction from German buyers could be another. This should go up as well. We don’t just want to sell more cars, we want to sell them at a certain profit. And to prove the white-glove part has worked, a recommendation score above eight among German buyers of cars above 100K. That’s how we know whether we were successful in the high-end market. These are some metrics we could put in place to see whether this strategy has worked.
What we want to avoid is generic metrics. We don’t want to say increase in revenue or increase in inventory. Those are great KPIs, but they’re not great OKRs for executing the strategy, because they could change through all kinds of things. You want metrics specifically tied to the strategy choices you made. The choice to focus on repeat buyers. The choice to provide white-glove on-site inspection and car delivery. If these metrics don’t go up, then you made the wrong strategy choices, you have to course correct, and the game begins again.
Aakash: And that’s why it’s a loop. After you come up with your strategy, you figure out the right metrics, and you use those metrics to feed back into your strategy.
Tim: Exactly. And you mentioned leading and lagging. That’s exactly the point. If I only look at generic metrics like increase in revenue, there’s too much noise and too much delay for them to tell me whether this particular strategy has worked. That’s the difference between something like OKRs and something as generic as KPIs.
Aakash: Makes sense.
Bringing it full circle (1:02:25)
Tim: So to bring things full circle. We talked about the point of strategy, and why it’s important to be able to say no to things as a result of it. We talked about the core questions you have to answer, and the components and tools you can use to answer each part of your strategy. How easy it then is to distribute them into whatever strategy frameworks and templates you want to use. How vital it is to flip them, to do the stupidity flip, to make sure they allow you to say no to things. How you can break things down, like audiences or differentiators, using very simple but incredibly effective tools. And ultimately, how you have to make sure strategy execution happens through specific metrics tied to the concrete choices of your strategy.
What to do on Monday (1:03:14)
Aakash: What should a PM who has to define the strategy for the next half in their product area go do on Monday after this episode?
Tim: If they have an existing strategy, the first thing I would encourage them to do is the stupidity flip on whatever they have in place. What are my choices about audience, value proposition and differentiation? Are the opposites obviously stupid? Have I essentially not made a choice? That would be my first thing.
Then, depending on the result, think about whether you can break down the audiences much more. Can I provoke focus through the audiences in my product strategy, and how does that influence value proposition and differentiation?
The third thing I would recommend. If my current strategy has me believing that people come to me because of AI, please think again. People won’t come to your product just for AI. What is it that actually makes people choose your product, and how can you use AI to leverage that reason much, much more, so people keep coming back?
What you inherit vs what you create (1:04:33)
Aakash: Let’s say we’re at Fair Mile, and there are people working on strategy at different levels. There’s the CEO, probably defining a lot of the 7 Powers and core user groups. Then there’s the head of product, VP of product or chief product officer, defining some of the more nuanced elements. Then maybe a group product manager on buyer, and an IC PM within the buyer experience who’s focused on the delivery experience. These are four different people who probably can’t all make these choices. If you’re just the PM on buyer experience, how do you think about building your strategy? What do you inherit versus what do you create yourself?
Tim: I love that. A great concept I keep coming back to for the multi-level nature of product strategy, also from Ravi Mehta, is the product strategy stack. At the top you have your mission and your company strategy. In between you have your product strategy. Below that you have your roadmap and your goals.
I love that you used the word inherit. When you look at these levels, think of each level of strategy like a funnel. You have to narrow down at each level. You might have the company level, a product strategy, maybe a level below. For every choice from the level above, ask whether you can just inherit it as it is. If the company says we’re betting big on brand or SEO, you can probably inherit that very easily. If the company just says car people in Europe, and you’re the head of the buyer side, you have to break it down. What is the biggest leverage on the buyer side? What’s the biggest segment we should focus on?
That’s the core message I would give. Look at the choice you have from the level above. Does it allow you to say no yet, or do you have to break it down one level further?
The best concept I’ve heard on this comes from a friend who works at Personio, which is Europe’s largest HR software operating system. When I talked to him for my book, he described an idea at Personio. When people think about product strategy, they’re encouraged to think about something they call evergreen benefits. Evergreen benefits are essentially what the company is already incredibly good at, and how you will leverage that for your product.
It’s a great example of how things have to carry through the different levels. What is the company already pretty good at, and how do you leverage it? You don’t have to reinvent the wheel at the product or feature level if your company has mastered a certain distribution channel, or has tech or integrations available.
Aakash: So it sounds like we go back to that first fill-in-the-blank statement slide you had. You look at what you’ve inherited already, where people have already made decisions. Then you apply the stupidity flip and ask whether it allows you to say no for your area. If you need more specificity, those are the places you work in to figure out your strategy.
Tim: Exactly. And you can always run it back. Does it help me say no? If it doesn’t, I have to narrow it down further and make a more opinionated choice. Once you have an opinionated choice that really works, you translate it into metrics to ensure it will actually be executed.
Mapping roadmap items back to strategy (1:08:22)
Aakash: So there’s some component of strategy work where you might want an idea of your roadmap choices, so you can sort through and ask whether this allows you to say no to certain items?
Tim: We all know it’s an imperfect world. But ideally, once you’re clear on what your strategy looks like and you’ve translated it into metrics, the question you should be able to answer is which solutions and roadmap items will help you achieve these goals. You should be able to map whatever is on your roadmap to the metrics that resulted from your strategy.
Say I’m going to build a new detail page with a larger photo view of cars. You might ask, does this help me get people on board who want to sell expensive luxury cars? Then the question is, what’s the evidence for that? Do I have evidence that this is a problem for sellers of cars above 100K? The metrics are your proxy link from your roadmap items back to your strategy. They make sure you have a narrative for whether the work you’re doing fits the strategy you set out to achieve.
Where to find Tim Herbig (1:09:54)
Aakash: Amazing, Tim. We could probably go another two hours on this topic. This has been so insightful. If people want to get more of you, where should they go online?
Tim: The easiest way is to say hi on LinkedIn. You’ll find me there as Tim Herbig. Alternatively, you can check out my weekly newsletter at herbig.co/newsletter, where you get all the good stuff every week in your inbox.
Aakash: It’s one of the few newsletters I have been reading for years and years. I highly recommend it. Tim, thanks for being here.
Tim: Thank you so much for having me. It was a real pleasure.