How AI is Forcing Startups to Rethink Pricing with Madhavan Ramanujam | Ep. 5 The NFX Podcast

20% of what you build in tech drives, 80% of the willingness to pay. This, I've seen it over and over again.

Why is AI making us rethink pricing generally?

I think products have very increased autonomy and very increased attribution that actually gives you insane monetization potential. And if you don't capture that from the get go, you're training your customers to expect more for less and then that's a slippery slope. Right. So how do you capture that value?

All right, Madvan, so good to have you and so good to see you again. I'm Pete Flint, joined by Anna Pinole from nfx. And so it's such a delight to have you back on the NFX podcast. We go back a long way, way back. So a little bit of context. So we were together like a dozen years ago when you were SKP doing some pricing work at Trulia, and I think you'll get into this. But we at Trulia, we started with a very simplistic pricing idea, like what can we sell and what is easily understood really, really quickly. And we left a lot of value on the table and then we were lucky to be connected with you. And you came in and really helped to improve and add a level of sophistication about pricing and insight, which was not only economically super valuable, but I actually found it fascinating just professionally because it's just this mix of psychology and economics.

Yeah, absolutely. And it was also time date was before the ipo, I think.

Exactly. Yeah. So, yeah, I think it sort of pretty quickly added seven figures to the bottom line. So. And you've. And we stayed in touch over the dozen years and so maybe share a little bit about kind of like what, you know, a little bit of kind of how you spend your time and perhaps what you're doing right now.

Absolutely. And it's super awesome to be back. Thanks for having me. So, I mean, the last 15 years I pretty much spent working with tech companies on helping them with monetization. So I worked with over 250 companies, more than 30 unicorns now, navigating monetization challenges. I did this in a professional services capacity. Skp, or Simon Kutcher and Partners, like he called it. Me and my co GP Josh, who was also at Simon Kutcher, we left last year, towards the end of last year, and we started a venture firm. That's what is keeping me busy now. I mean, combined the two of us have worked with over 500 companies. But the reason we actually pivoted to venture was to Specifically also work with very early stage AI companies and we can unpack a bit of that. But my life right now is invest in companies, roll up my sleeves and work with founders.

And you've written a couple of books. You've written the first one, which is Monetizing Innovation, which I've highly recommended to so many founders. And then the most recent one, which launched just a couple of weeks ago, is Scaling Innovation, which is really thinking about monetization in a AI first environment, which we're in now.

Exactly. I mean, not just building great breakthrough products, but how do you build a great business, especially AI focused.

Great. So I think. So today we really want to pack in. You know, we spend a lot of time, Anna and I and the rest of the firm at nfx thinking about how do founders think about pricing in this crazy AI world where you've got these breakthrough products, but you've also got a world where you're seeing software become increasingly ubiquitous and potentially commoditized as well. So maybe, Anna, you'll kick things off.

Yeah, just to kick things off. So ever since this new wave of AI companies started, it was pretty obvious to all of us that this new technology was unlike anything we had seen before. By investing early in some of these companies, we were quickly exposed to the new possibilities in terms of value creation and as a result emerge in terms of value capture. And we often find ourselves discussing pricing with our founders. As I'm. I mean, and one of the things that I wanted to kick this off with was how have your own conversations around pricing evolved over the last few years in which we've been in the era. And why is AI making us rethink pricing generally?

Yeah, I think the big. There are a couple of big changes that have happened. Right. I mean, if you think about AI monetization or pricing, when should I think about it has changed dramatically? Because in the previous vintage of companies we could still say, let's just grow and figure out monetization when you're running a 80% SaaS margin business to some extent. Right. You can't do that in AI for two reasons. One, there's cost dynamics to navigate from the get go and there's also like value capture, like you rightly said. I mean, if for the first time, I think products have very increased autonomy and very increased attribution, that actually gives you insane monetization potential. And if you don't capture that from the get go, you're training your customers to expect more for less. And then that's a slippery Slope. Right. So how do you capture that value if you're building something as an agentic AI product that taps into labor budgets. Labor budgets are 10x compared to software and IT budgets. So if you use the old playbooks, you'll completely under monetize. So what we are seeing is monetization and GTM is becoming really, really important. Even in the pre seed and seed companies and even in the previous vintage of companies. We could say that if I have a two year coding head start, that is a moat. You can't say that anymore. Right. I mean you can probably code things up overnight. So what is the moat? You need to have some proprietary training, data, network effects and also like GTM becomes a moat and in that perspective your monetization model becomes a moat. And how can you have durable revenue? So all these questions kick in from day one. So what has really changed is the focus starts much early. I mean that's kind of also why I alluded to the fact that that's why we also pivoted to like work very early in a venture setting where we are investing and operating. As opposed to being on a fee for service model ourselves. We changed our own pricing model. If you think about it. Right. It was more on the usage. Now we are on outcome basis.

Yeah, that's awesome. And have you seen this happen before? I guess there's been like prior waves, prior technology waves that have also been known for setting new pricing models.

Yeah, I think with every technology wave there's been a new pricing model innovation that's kicked in. Right. I mean if you take Salesforce back in the day, you know, SaaS pricing was a huge shift from like on prem perpetual licenses. I mean take someone like AWS for instance. The, you know, pay as you go for infrastructure, that was a big wave that actually started. Even companies like Uber started like dynamic pricing as a wave on its own. With AI, what we are seeing is there's probably a move and impetus to be more outcome driven because we are moving from a, you know, buy software for access to buy software for work delivered. And I think that is where the industry is heading the way I see it.

And you are a big proponent. I mean in your book you talk a lot about encouraging founders to have willingness to pay conversations early. A little bit aligned with this idea of like pricing is very important since the get go. What are some tactics that you can share to do a good job at that?

Sure. I mean early willingness to pay conversation is really important. I mean let's talk about the why and Then we can talk about the how, the why. Because if you just build a product, slap on a price and you throw it out, you're just hoping, you just don't know. I mean, what we talk about, testing for willingness to pay is like testing for just like product market fit, right? I mean, entrepreneurs know that. I mean, if someone comes and asks me, do you like the sparkling water? I like it. Do you like it for $25? The whole conversation is different, right? So unless you put actually pricing as part of your product market fit, you often hear what you want to hear, right? So willingness to pay is critical. So you can at least identify even before you're launching your product, is this product something that people need value and are they willing to pay for? And then architect the product around it. And if you find there's no willingness to pay, the most important question is to ask why. And then you start hearing all kinds of things that you can actually productize around jobs to be done and unmet needs. So that's the why. It's very critical. It's like test and learn monetization before just slapping on a price and hoping on the how to do it. In the book Monetizing Innovation, we devoted an entire chapter to that. It's chapter four that's the most important chapter to read for the listeners. But I might like, let me just unpack maybe, maybe one specific tactic that actually Rahul Vora used when he came up with his own monetization was superhuman. So what we call is, it's like the acceptable, expensive and prohibitively expensive questioning to understand psychological thresholds. So the way that works is you take your product, your wireframe, blueprints, demos, free trials, whatever. I mean, just put people through the experience of the product. So you're having your same sales and marketing conversation that you'd actually have, you know, and then, and then have the pricing conversation. So once they understand the value, ask them, what do you think is an acceptable price? You know, clock that answer. Then ask them, what do you think is an expensive price? Clock that answer and ask them, what's a prohibitively expensive price? This is a very stylized way of asking. If you just go and ask something like, you know, how much should I charge for this project to charge for this product? You'll probably get garbage back, right? I mean, that's your job. But if you ask it this space, after you pitch the, you know, entire product, the sales, marketing conversation, you start hearing something reasonable. Because acceptable price tends to be the price where people are negotiating with themselves that's the price that they love, not just your product. So they're going to low ball all day long. Expensive pricing tends to be where it is around your value price. And prohibitively expensive tends to be where they laugh you out of the room. Now if you do this a bit statistically with like let's say even 100, 200 people on an online study or however you administer this, you start seeing that these demand curves have cliffs. Like for instance, after 29, if you go to like 31, suddenly 20% of people actually think it's expensive or you know, or 40% don't find it acceptable, et cetera. So then you start seeing these psychological thresholds. That's how you know that, okay, you need to be right around that price point. If you cross it, it's going to be a threshold. So Rahul actually used this for Superhuman and found that, you know, $30 was a great price point for the product that he had. And that's also how we actually launched it at a $30 price point. And he was talking about this in an A16Z podcast. That's where I learned that he read Monetizing Innovation and did this. And we've been great friends since then.

Out of these, out of those three different segments, acceptable, expensive, prohibitively expensive. Like where do you want to anchor?

So if you're in the look, if you're, if you truly want to be price value aligned, it's typically on the expensive. Probably expensive is like, you know, you shouldn't be there. That's like, it's a price premium paradox where you just want to overprice thinking it's good, but you're actually going to hurt yourself. If you're in the acceptable zone, maybe in the growth phase, it's still okay to be there because you can actually then you're going to get a lot more acquisition. As long as you have a land and expand strategy, it might be just fine.

Yeah, okay.

Yeah. But if you want to start off with a value price, then the expensive price is probably more closer to your value price. It's the price where, you know, people don't hate you, they don't love you, they're just neutral. They'll pay you.

Yeah.

So I'm curious thinking about like popular like ChatGPT cursor. There are, there are many claims that they might be not fully capturing the value that they're delivering. Like what do you make?

Yeah, I think it's, there are some self inflicted anchors in that space. Right? I mean, is a $20 price point actually, you know, good, or can it be more? I mean, back in the day, Copilot, GitHub, everyone started at 10. They moved it to like 20, 30. I mean, there's some anchors, right? I really think that for that, you need to really understand what value you're actually bringing to the table. And can you charge a value price based on that? And can you contextualize that, you know, pricing? I mean, we talked about Superhuman, but just to drill down a bit on that, one of the chapters we write in scaling innovation is called beautifully simple pricing. And how do you keep it simple? And when you think about it, when Rahul actually introduced Superhuman, he was competing with free alternates, Gmail and others. Like, why would people even pay money for another email app? Was a question mark. But he was actually delivering core value, which is, you know, I can free up your hours and increase your productivity because you can, you know, log into an outbox, et cetera. But the key there was the beautifully simple pricing. Talked a value story. So he didn't just say it's a $30 price point. He said, it's a dollar a day to get five hours of productivity back in a week. Now that $30 doesn't look too expensive. Right? And that's actually how everything took off because. Okay, will you pay the price of a latte to get five hours back? Absolutely. I would do it. Right, but. So the entire premise was on the value, where many of these coding agents wipe coding, et cetera. While they're emphasizing the fact that you can do things fast, they're not emphasizing necessarily the value. So instead of A$20 per month, if it's A$30,$1 a day, to get incredibly efficient at coding and save you like five hours or 10 hours a day, would you pay for it? You would, but you probably never charge, so why should I? So I think if you start wrong, you kind of end wrong. Then of course, a different strategy. You can say, I want to grow. I mean, you, you see some of, you know, some of these companies having like, very fast ARR, but we can talk about whether that's durable. What are the margins? There are a lot of question marks.

Do you have a point of view on margins? Because it's such a competitive environment out there and any good idea is replicated very quickly, and with the cost of software coming down, it seems the sort of incumbent SaaS businesses are kind of holding onto margins. But are we in a world where you see many of these AR companies launching and their margins are 90 plus percent because they're first mover and they're replicating labor budgets. Whereas your mentally like, well, how sustainable are these margins long term? Do you have a point of view on how things will evolve?

Yeah, I think the key is to focus on both market share and wallet share. Right, that's how I see it. So it's more profitable growth, it's not profits, it's not growth, it's profitable growth. That's also the subtitle of the Scaling Innovation book, How to Architect Profit Profitable growth. And what that means is it doesn't mean that you need to have equal efforts at any given point in time on market share and wallet share, but you need to have equal attention in the sense that even if you gave away on price to actually acquire customers, are you thoughtful about the fact that you can land and expand later and you have a clear vision on how to actually go towards wallet share? And similarly, if you started more on the wallet share side because you think there's a new market that you can start shaping, do you have an alternate to actually create a low end product to actually gain more market share? So it's being thoughtful about it and having equal attention but not necessarily equal efforts because at certain points in the company you might actually want to index more on one level or the other. But the best CEOs have been the ones that can actually think about the interaction effects between acquisition, monetization and retention and then you start building towards profitable growth. So while we talk about some AI companies having high margins, on the flip side some of them have really low margins, especially if you take some of the coding ones that we talked about. There was a TechCrunch article that it's either neutral or negative in terms of margin. And if you have a lot of ARR at negative margin or neutral margin, we can question, is that a great business or do they have a pathway to get to better margins? Would they build their own model? Is there efficiencies that we will see or if it's just hoping that the cost would come down? Hope is not a strategy.

So founders ask us every day just how do you think about monetization strategies and what are some of your frameworks to think about? Monetization strategies for early stage startups?

Yeah, I think there are two questions that come up and maybe I'll be curious to see if it comes up in your conversations with founders and if those are the right ones, we can unpack each of those. The first question that comes up is, you know, how do I charge for this Product, like, what's the pricing model? Because often we say how you charge is way more important than how much should I be on consumption base? Should I be on seed? I saw someone else doing outcomes. Should I do that? So, like, what should I do? Right. I mean, that's closely tied to your operating your business, that it comes up and that's a sort of choice that you take earlier on. And the other question that comes up, especially with B2B AI companies is, hey, how do I navigate PoCs? I'm getting into these commercial agreements. My buyer wants to see whether these products deliver value, how do I charge for it, how do I navigate these big deals? So those two questions keep coming up in the pre seed and seed stage.

And so let's just.

Is that same with your company?

Yeah, they're like, okay, we have a breakthrough product idea. We know that this is. I mean, I think there's a. In consumer, it's, you know that in some ways the superhuman stories as a reference point for some of these consumer applications. But in B2B, it's kind of more complicated because there's a sort of deeper engagement. So maybe just walk us through how you might, how founders might think about kind of navigating the price discovery with customers who are. Which we see today, like enterprises and SMBs are so open to AI, they've kind of experiencing it in their daily life. They say, I know this can be helpful. So they're very open and receptive to conversations and founders are building amazing tools, but they are not clear how to navigate this dynamic between what is the value I'm delivering and what is the price I'm able to charge.

Yeah, no, totally. I think let's probably then talk about navigating those big deals first and then we can come back to the pricing models later. So these POCs have become critical because like you said, there is a lot of curiosity on the buyer side. Right. And also a lot of budgets to experiment. They want to see if AI can actually help their internal efficiencies and things like that, but they don't know how. So they keep asking these. The AI startups and companies saying prove the value. So now the classic mistake that a founder makes is approaching the POC as a technical validation. Right. Because if you're just approaching it, as will my tech work in the environment of your customer, you're not really proving out any business case at all. Right. So we actually talk about, when you think about POCs is to frame a POC as a business case validation exercise. Tech Validation is part of that. So which means that you try to co create an ROI model with your customer. So what that means is you start the POC and say, okay, it's a finite amount of poc, maybe a month, two months, three months, whatever, not more than three months typically. And then you say, okay, for that period we are going to jointly create a business case. Why is this important? Because the buyer on the other side now actually participates in that co creation and becomes the smarter person in their organization where they can actually shepherd it after that, you know, two to three months and say, hey, this product will actually unlock X millions for us and we should buy this, right? I mean, so you also inevitably make them smarter. And the other reason to do this is if you just work on something and show up with an ROI model, no one is going to believe you. If they work with you on the inputs, they will believe the outputs. As simple as that. So framing the POC as a business case exercise, where you're building a business case saying is there value generated from that AI and being on the same page with customers, this is the key thing. And if you don't do that, it's a mistake. And when you actually do that, there are various things that you need to think about in terms of how do you value sell, how do you create an ROI model, how do you negotiate. All of those things become critical. But you postpone the pricing conversation to after the poc. So the POC is typically a fixed engagement only to build the business case. Then you tell the customer that, okay, look, commercial discussions will follow because you want to clearly upfront state that. Otherwise your POC becomes, you know, POC price becomes the anchor for your pricing. They just multiply by 12 times.

How do you handle like one of these customers being like, no, I don't want to, I don't want to postpone this pricing conversation because it's important for me to know now.

Absolutely. So that's a great question. If you're asked for price during a POC and pushed, what would you do? I would say you should talk about price, but in a more, let's say, strategic manner. So there are a couple of ways to deflect that question and that's the most important thing, right? So you can say, hey, look, no, the goal of the POC is to actually create that value case and we can talk about what portion of that we deserve. Let's assume the buyer on the other side says, that looks theoretical, I still want a price. Right? So another way to deflect this would be, you can say, look, customers like yours have been able to unlock tens and millions of values and we are typically on A1 is to 10x in terms of your investment in US and the ROI that you get out of it. So you basically just communicated that you're probably a million dollar deal when actually it comes to it, but without actually saying it. Some customers will be satisfied by that because they're like, yeah, 1 is to 10 sounds fair, we will deal with it, we'll finish the POC and we'll come to it. If they push you even more, then don't give just a price point because that'll be the most artificial thing you can do because you're also guessing. Right? So in that case, I would give a range. I would say, look, it can be anywhere from 500k to a million for the first year. Where we would be will depend on how much value unlock we can actually do for you. And that is why we are bringing, that is why we need to do the value case. So in all of these conversations, you're bringing the focus back to the POC is a business case exercise and how do we co create value? But I just answered that, okay, it's a range. I gave you predictability. Beyond that you don't need to. And where you are depends on the business case. And if you unlock 10 million, you can be on the right side. If you're unlocking three, you're probably on the left side. Right. So I think that's also how you can navigate this more. The best negotiators or the best, you know, founders who are selling hardly reveal their price because if you don't understand value, what is price?

Are there any tactics to communicate value? Because I think it's like you might be able to, it may be that, I don't know, there is the VP of sales which is using the product, but the CFO may deeply understand the kind of value story. And so just in incorporating into the product the demonstration, the value. Have you seen any companies that do a really good job of demonstrating to constituents the value that they create?

Yes, I think absolutely. And the lesson learned in those situations is being very structured and systematic in terms of how you communicate value. Random statements of I'm actually saving you this or that, it's kind of lost. It has to be systematic. So you're almost training your buyer to repeat the message that you're giving them internally and they become internal champions. So the best ones actually say, hey, look, there are three categories of value that we actually provide. The first one is incremental on your top line based on your business KPIs. So that could be. We generate more revenue for you, we reduce churn. So these are all incremental. Right. So that. How are we participating in that? If you are communicate that that's separate from every other value element. So we actually improve your top line. The second one is we actually save costs. That could be other license cost savings. It could be headcount reduction. It could be, you know, some tangible cost savings that just happened because of AI. Right. So these are the cost savings which a CFO would actually like. Right. And the third one is opportunity costs, which usually a CEO like. Right. Because you said, I'm also freeing up 10 to 20 hours for the team on a weekly basis. That can also be quantified. What would they do with that time? What is that actually worth for the organization? So when you put it into these three buckets, the incremental top line, which is really important, everyone cares about it. The cost savings, typically the CFO function and others care about it. The opportunity cost, definitely the management team actually cares about it. Then you can start putting your value story across these categories and then train the buyer to actually go paraphrase that wherever they are in the organization.

And when you say trained, you mean prompts within the product? I mean not prompts in terms of AI prompts, but just kind of visual prompts in terms of just reinforcing the value that's been created at every interaction.

Correct. And I would even, you know, in a negotiation, when I tell founders this, pause and ask them the questions. Hey, this is what I said. You know, how we will add value. What do you think? What did you get out of it? Like, if you had to repeat that back to me, what would you say? I mean, this is very like, I just want to know you understood what we are saying, right? I mean, like, like how do you think about it?

They say the best products are those that get people promoted internally. Right?

Absolutely.

Get the champions of. Yeah. Promoted internally.

Exactly. And, and this is. I mean, if you ask it in the most innocent possible way, it's a great test. We talked about beautifully simple pricing. One of the tests for whether you have a beautifully simple pricing is Monday morning. Go back and ask one of your customers to play back your pricing strategy to you. If they cannot, you don't have a beautifully simple pricing. It's just complex as hell. And you probably somehow sold it and they don't understand it. If they didn't understand how they were charged, they don't understand how the value is created, that's a churn situation waiting to happen at some point. But if they can articulate and say, hey, your pricing makes sense because we get this, the price value realization is aligned.

So we've seen the evolution of SaaS in this AI world move from this per seat licensing to a lot of the exciting stuff is on outcome based pricing. And you've got this two by two matrix, maybe break that down for us and kind of what types of companies are kind of able to capture most of the value?

For sure. And I think the two by two, it goes back to the same two dimensions I talked about in the beginning, autonomy and attribution. Right? That's really what's changed with AI. So if you think about a two by two with autonomy on the Y axis and attribution on the X axis and low, high, low, high, if you take the bottom left quadrant where autonomy is low and attribution is low, when we say autonomy is low, that means you still need a human in the loop. You're operating as a copilot. Attribution is low in the sense that your product probably is good, but it's not directly tying into any kind of business metrics that your customers are tracking. So if you're in that quadrant, you're kind of relegated to a perceived model. Right? I mean, even if you take a company like Slack for instance, you can say, you know, I use Slack and my productivity went up, but you can't measure it, you can't meter it, you can't monitor it, you can't charge based on it. Right? So the attribution is low and it's autonomous. Autonomy is also low, it's copilot. So that's necessary in a seed based licensing model. But if you take the bottom right quadrant where attribution becomes higher and autonomy is too low, you're still in copilot, but you're adding a lot of value. This is also where some of the coding platforms have started moving. Like Cursor for instance is here, or Clay is here for instance. Right? Where it's a hybrid model makes sense. It's still a seat, but then you also overlay a usage component. So like certain packages might come with certain number of AI credits and if you use it after that you can buy AI credits on top. So there is a blend between a seat and a usage model. That makes sense because it's still copilot, but the more people actually use it, the more value they're getting. So I want to monetize on that, right? If you're on the top left quadrant, autonomy is high but attribution is low. What that means is there's no human in the loop needed for these AI products. They kind of live in the background and do their stuff, but they're not directly influencing the KPI of your businesses. So they are more background infrastructure, tools, platforms. So they have to be purely on a usage based basis because that's your best proxy for value. There's no seats involved so there's no point and it'll be bad. But a usage is the best proxy of a value. Like in a company like even the classic Twilio would be probably in that kind of quadrant, right? If you take the top right quadrant, that is where autonomy is high and attribution is high. That's the kind of exciting quadrant if you can be on. What that means is your AI is fully autonomous and can also do stuff that is highly attributable, then you can be on an outcome based model. That's the holy grail of AI monetization. It's not for everyone because many products you can't show attribution. I mean even if you show it, do people believe it? Is the loop closed? Is it fully autonomous? So I think there's still work to be done. We are seeing about you know, less than 5%, probably around 5% of companies that are right now in AI on outcome based. We do predict that this will go to like 25 to 30% in the next three years. And I think that's kind of where people are moving and we can talk about that. But a good example of a company here would be like if you take Intercom, they develop fin AI. So the way it actually works is if the AI agent is able to like you know, autonomously resolve a customer support ticket without any human intervention, then they charge for it. If a human intervention is required, then they don't charge for it. And there is self attribution loops built in here because the customer might say I'm satisfied with this ticket. They're actually closed with an agent. No human in the loop is clear outcome, right? So they charge for that. Or there are companies that actually have demonstrable savings. Like for instance a company that, you know, recoup's chargebacks, they do a 25% of the recoup money. So that actually is directly, you know, because they measure it, they, they understand it, they take a portion of it, right? So if you're and it's autonomous. So if you're highly autonomous, highly attribute, attribute attributable in the, on the top right quadrant, you can be an outcome base. So I think this is the key thing. The, the, the thing is to really understand, you know, what is the right pricing model or archetype for you based on your product in terms of like how autonomous and attribute it is. Where I see founders making a mistake is chasing shiny objects like they just heard from some networking event that they went to that outcome based is cool. Like I need to be an outcome based. And actually talking about outcome based pricing when the attribution of the product is incredibly unclear, that's setting yourself up for failure. So like understanding where you are in the squadron is super important, then you can build pathways to also move around if you need to. But if you stop wasting time in trying to like just copycat and like see really what is the right situation for you.

It's, it's also the, the outcome based pricing also creates an alignment internally between the product efficacy and, and, and tools and also customer expectations. And so that alignment is just terrific because it, rather than focusing on like okay, how do we sell more seats? It's like, okay, does the product actually deliver? But as you say, I think it's more challenging.

Yeah, but that's a great point because for instance, if you take these, you know, customer support ticket resolution, like, right. I mean if, when I start, if for instance only 20% of them got done by AI without any human intervention, then everyone in the company is now incentivized on how the hell do I make that, you know, 60 or 80%. Because that's when the product is actually delivering on its value. So you're completely tied in destiny in terms of like your vision, everything else with the customer and you're partaking on that. That is where it becomes a very beautiful model where you switch to like, you know, charging for like work delivered as opposed to access to software.

You talk about wallet share and market share and you can see that outcome based pricing can be critical because it or capture a huge amount of value. But at the same time the concern is that unless there's a real commitment, which is like okay, you're paying whatever a particular flat fee, people are less invested in the product, they'll just use it episodically or, or as an option against many other channels. How do you see that in practice as a real challenge or do you see strategies for companies to kind of mitigate this sort of one of many outcome based companies.

No, that's an excellent point because there's also some characteristics around vendors. Outcome based model make sense of a company and a market and also when do the other ones make sense? So we actually talk a lot about this in Scaling innovation. But you talked about episodic. For instance, if my value delivered is episodic, you don't want to be on an outcome based model because that almost looks like holding someone hostage. Like when the value was delivered, you're actually coming, the bill collector is actually showing up. Think about this. In the previous vintage of LifeLock, it's operating in my background, checking my credit, identity theft, everything else. And then you say, okay, I found an identity theft breach, I'm going to charge you. It's like episodic. Like it just happens once in a while, right? That actually that. But at that time it's insane value. You don't want to monetize on that because that actually looks like a really wrong thing to do. But you're actually paying for the, you know, peace of mind and it's like an insurance. So in those kind of cases you want to actually have a, you know, either a fixed fee or a recurring revenue basis. So like whenever it happens, it happens. But of course you need to choose your level of pricing commensurate with value. But if it's like a more, let's say more frequent deliverables outcome kind of makes sense. Like a ticket resolution. There is a ticket, it's resolved, you get monetized on it and then it adds up. Right. So I think there are some real characteristics on when each of these apply. If you're an episodic one, probably not.

Do you see any consumer based pricing model, consumer applications which have outcome based pricing. You can imagine a fitness app which is like, okay, if you get feed.

Yeah, yeah.

If you lose 10 pounds or kind of like do you see anything happening there? Or the customer is, is perhaps not sophisticated or too much noise.

No, I mean we've actually seen these kind of models. Like that's more of a gamification for me. I don't know if it's an outcome based. You could claim it's like driving outcomes. Like if you lose weight then we like you start with $100 a month, for instance, just making it up. And then if you lose weight then we give you like, you know, 20 bucks back. Or like if you study well in a course and you actually get a good grade. We've seen some of these things, we get some credit back because then you can actually have tie ins where people Will submit their scores and everything else and you know, that the product actually improved their performance. It actually works well, it's a, it's more of a gamification. I think an outcome based model in the consumer side probably has not taken off as much as in the, you know, B2B side. Probably more to come.

One of the key characteristics for outcome based pricing, apart from autonomy, is being able to prove, I mean there needs to be a key metric that you impact and the company needs to have some sense of like what's their current cost without AI or without your product, what, what percentage of do these AI companies are charging? How much are they capturing? Because obviously if you charge the same as their current cost, they're going to say no. But I wonder what's the discount?

That is common for sure. Like you said, first of all, attribution is super key and the fact that you can attribute it and it's also closed loop in the sense that at the end of the day your customer independently can say I got this incremental, you know, revenue or cost savings or opportunity costs because of the company. And then you know, the question then really is what is if you can show that attribution and you unlock value, what can you charge? Right in the, in the historical, like I mean in the previous vintage we used to say if you're one is to 10x it's good pricing in SaaS. In AI companies we're actually seeing that you could even capture 25 to 50% of that attributable revenue because it's true, incremental and autonomous. Right. So 25 to 50% is the benchmark that we have for that we are seeing actually some companies actually do it.

And do you think this is going to sustain or are there any. Because another thing that makes me think like these outcome based pricing is that it's so simple and clear that is also very easy to potentially switch providers. Right. Like if someone comes and is able to resolve that ticket as you know, the current vendor and it's not charging 50% or 25% but is willing to go to 10 like the switching costs. Yeah.

Makes me think, I think switching cost probably can be built internally with the product and as opposed to like having your pricing model be the switching cost. In fact I would argue that if your pricing is complex, that is when people actually want to leave you to like someone else. If your pricing is simple and they understand why they're paying and it's based on outcome, it's actually hard to like for someone to Leave because you're putting skin in the game. Unless someone is going to charge lower for outcome. But then that's a different question because then your product needs to have some switching costs in the sense that, you know, it's very well integrated with things. It's hard to rip and replace. It'll take some another AI agent to like do the training all over again. And you put some guardrails and you've built some new, let's say enterprise RL models that actually can solve things. Then of course it's not easy to rip and replace because then there are models that you're building, there's training data, people get used to it, they understand the user interface, they understand how the agents work. Then it's not that easy. I've seen actually your pricing is complex. That's when people invite other people in the room. Right? Because then it's like, okay, I want to simplify my pricing, but if it's outcome based, itself is a moat is what I'm trying to say.

Yeah, everything's aligned and you certainly have also just this is easy to communicate and easy to track and measure. And how do you see folks? Do you see folks moving around the map or revisiting pricing? Because in this we see a number of AI startups which are entering the market with an outcome based pricing and then building these compound startups where they're like, we're tacking on a bunch of other value after this very attractive initial wedge. How do you see, how should founders think about that evolution and revisiting pricing?

Yeah, I think, like I said, it is. When we, when you start based on your product level of autonomy and attribution, you need to pick the right quadrant, but you also need to build pathways to like move around the quadrants. If that is something that you can actually aspire to do. And this has actually happened in many industries already. Maybe we can take an example to unpack that. If you take me, we talked about coding agents and coding assistants. So if you think about the classic GitHub copilot, for instance, they started in the bottom left quadrant. Attribution was low, autonomy was low. So they were a per seat model. Back in the day, GitHub was per user. Right. I mean we all know that when you take the cursors of this world, they actually move to the bottom right, which is, you know, attributions higher. You're saving a lot of, you know, time for the developer. You're creating code that can be almost reused as code attribution. But it's still on auto, you know, copilot mode. So they are in a hybrid of seats and usage model. But you already hear companies talking about, I'm building an entire autonomous coding platform. Like, I'm going to hire you a QA person who can actually do qa and you don't need Jim or Jen doing qa. You can have my AI do it and I can actually show you how many bugs we've fixed. Everything else autonomy and attribution is increasing in that space that you could actually say, okay, then I might want to actually charge based on outcome. How many bugs did you resolve and how's that compared to like a human? And should I charge for that agent differently? Then you get into more interesting conversations. But the key is to understand that where you are and what do you need to do? Like, how can you build more things in your product that will demonstrate more attribution over a period of time? And how can you build stuff in such a way that your product can get more autonomous over a period of time? And for attribution, some simple hacks are things like, okay, you actually want to have some kind of dashboard where I log in as a buyer and I can actually see some charts saying, how is this AI actually increased my top line, my cost savings, my opportunity cost, and you know, sort of make that the front page. I mean, this, hey, we have actually saved these things. Or it's a report that's commonly available. We are training people, the attribution exists. And how do you build pathways towards that before unlocking outcome based pricing, et cetera? So I think you can move around, but you need to plan for it.

The pace of AI is getting so much better so quickly. And we're. Super intelligence is a tough definition, but as you kind of roll this forward a couple of years, do you see outcome based pricing being. You mentioned 25%, but do you see it perhaps kind of more prevalent than that over time as sort of artificial superintelligence becomes kind of more pervasive? Because I think we have to be building for this world because it's getting better at such an exponential rate?

Yeah, I think that in the next three years, also based on studies that we have seen, we have done ourselves, stocks, et cetera, where the buyers are actually getting more comfortable. If you look at CIOs and organizations, even the ones who used to say, I want predictable budgets, they're like, you know what? I'm actually willing to see if outcome would work because our incentives are aligned. So that's also how we get to that 25 to 30% in the next three years. If I take a more forward look of, let's say five, 10 years, superintelligence, AGI, we can debate what our worlds will look like. But I actually think intuitively outcome based pricing models should make a lot of sense. There's. Right, because I could even envision a world where AIs talk to AIs and they reconcile how much to pay each other and they basically, you know, it's based on the outcome and there's actually no bias like humans. Right. I mean it can actually do that and it has to be an outcome. If you take a world with less bias, outcome is the right model. Yeah, yeah. There's no debates.

Yeah. There's no, you know, it's like, what is the value to me and what is my willingness to pay? And, and it's a pure like individual economic decision which lends itself to, you know, some sort of quasi auction type situation in a, in a, in a real time basis.

Yeah, exactly. My agent hired a, you know, recruiter agent because those models are actually better. I'm a general purpose AI hire them and they do the work and I pay them based on the outcome. Yeah. And it all happens in the background. Right. It's a bit interesting to see how the world will evolve, but it's anyone's guess.

Yeah, but then it's in that environment, I think the founders need to think about what is the true defensibility beyond the sort of the AI product that they're building, which is a lot of data, network effects, workflow, et cetera. So if you're a startup competing against an incumbent and you talk, you know, when we spend time together a dozen years ago, there's a lot of thinking about bundling and kind of pricing and different kind of components. You can see an environment where you have a sort of an incumbent which arguably has a sort of bloated product set, adding an AI component and they're looking to kind of okay this sort of like machine of like, okay, we need to kind of ratchet up the kind of acv. And then you have a startup which is saying, okay, I'm going to like monetize on an outcome based on my kind of high value AI product and then give everything else away for free. Like, what are, what are some of the merits of kind of that approach and any tactics for, have you seen that working? Have you seen, and have you seen tactics for founders to execute on that?

I mean, like, look, when, when you have a incumbent in town and you want to compete as a startup, there are probably a few ways that you can actually do it. You can say I would try to be a low cost provider compared to the incumbent and that's the reason to switch. But guess what, the incumbent probably has a lot more capital and they can drop the price faster than you can drop in some way. So that's a losing proposition. It's always the other person who started the price war. I mean, so that's one way you could think about it. But the more smarter way actually to think about it is to say are there some strategic things that I can do in terms of how I charge my pricing model? Can I actually make it different? Are there some core pain points with the current incumbents that I'm actually solving and how can I leverage that? Even back in the day, when you think about Netflix, Blockbuster used to have all these movies and you have price based on movies, late fees, everything else. But a subscription price on a Netflix to keep your DVD however long you wanted was a different pricing model that actually worked. I mean, and the rest is history. So this has been done in various vintages like where a pricing model conversation itself is the reason for disrupting an existing incumbent. So if an incumbent, because think about it, a really big incumbent with multi product, many different teams just building AI into their existing tech architecture, which does not have AI, what are the chances they will move into an outcome based world? Zero. They're going to have 20 debates, consensus and they will shut it down and say no, let's just preserve our margins for our current business. And they won't do it. But if you're, if you're coming new to the market, you have nothing to lose. If you say look, I win, win, you win. And my model is outcome based. Great. Maybe that's how you start stealing accounts. Right. So I think thinking about those things would be key. And I've seen it work. Not just now, we're not just seeing it now, we have actually seen it.

Historically and we're seeing this, this trend. Right. Like with AI at the end delivering work. Right. It's reducing the, the need for so many people in a company. And, and like a seed based pricing thrives when a company adds more headcount. Right.

So yeah, it's like catching a falling knife if you're on seed based pricing.

Exactly. Yeah. Yeah. It's kind of like the, the classic innovators dilemma.

Yeah.

So, so in the book you talk about these, these founder archetypes and perhaps failure modes and Kind of things that folks don't do so well. Can you talk through a little bit of the failure modes and archetypes?

Sure. I think it ties back to the same market share and wallet share. Right. I mean, so I love two by two, so I got to use one more.

We love two by twos. Like frameworks.

Yeah, exactly. So when you think about market share on the Y axis and wallet share on the X axis. Right. And let's take the top left quadrant first. High focus on market share, as in a leadership archetype or a person who has high focus on market share but low focus on wallet share. You know, we, in the book we call them the disruptors, right. These are people who say I'll grow at all costs and I'll figure it out in terms of monetization. Right. If you take the bottom right quadrant where wallet share is the big focus, but market share is not that big focus for the CEO or the leadership. We call them the money makers. These people actually think about from day one how to build a great commercial business, but not focusing more on the market share. And then you have this bottom left where you're neither focusing on market share nor wallet share. We call them the community builders. They're focusing on a core set of customers and they want to do right by them and just keep working with them. And then if you take each of these archetypes, I'm sure all of us have seen many of these. Let's take the disruptor one. Focusing more on market share but not paying attention to wallet share. Not equal efforts, equal attention even, not even paying equal attention to wallet share. And you're basically selling $80 cents literally. That's what's happening. So you fall into two traps. You know, you're probably landing without expanding in the sense that you gave the farm away in your entry product and then you're chasing your tails to like build something and hoping to monetize. That often does not, you know, sort of translate. And that happens with disruptors all day long. And the other one is making the mistake of thinking that market share won is market share earned. What that means is you're so focused on acquisition that you keep thinking about getting new customers. You're not focusing on keeping those customers retaining their, adding more products value to them. And retention is not the focus, acquisition is the focus. So you won market share, but it's not durable. Right. That happens with the disruptors. If you take the money makers, you know, focuses more on making, you know, money and, or thinking about that before the market share, they fall into a couple of traps. One is the price premium trap, which is, hey, you know, I'm. I want to charge a premium because I learned that premium price means that it's value and, you know, I'm the next eve. And a $200 wine is like, value and whatever. I mean, they've learned stuff. I mean, while there are obvious connotations that if price is signal of value, the price premium paradox is when you charge it so high that you become irrelevant to your audience, right? So that actually happens all the time. Like when you start charging for a Juicero machine at like, whatever, 700 bucks, when you can probably squeeze the packets and you have the same result, you're in a price premium paradox. I mean, so that's different. Or you also fall into a, you know, nickel and diming kind of trap, which is you're so focused on the wallet share that you have these. Your pricing is incredibly complex money, maybe even different, you know, elements of the fee structure, things like that. Because you're only thinking about that now, the market share. If you're the community builder, this is an interesting one because there's also some literature around. Let's just focus on a few customers, build a product and scale it. CEOs who have actually done that have still been very thoughtful about wallet share and market share and paid attention, even though they started community builders. But the community builders who don't pay attention to these, they fall into traps like, you know, giving away too much for too less because they're so, you know, want to please their customers in their community. They keep giving. They keep giving and they train their best customers to expect, like, insane value for less. And they can never undo it because once those customers become reference customers, they'll be like, yeah, I got that for like $10. Are you kidding me? Like, it just doesn't make sense because you just train someone to, like, get insane value, low price. They're not even going to be good reference customers because they're going to say what price they actually got it at. And that's going to be the reference point for everything that you actually do. That's a trap that you fall in. Or the other one is, you know, you fall into this trap of, you know, you're solving for, you know, current, let's say base, but you're missing the frontier, which means you're so focused on the current customers that you're not even thinking, are there adjacent markets or others who look different to my loyal base that I should be building towards, so you miss adjacent markets, you know, opportunities to acquire new types of customer segments. So the best quadrant to be in is the top. Right, which is the profitable growth architect is what I call it, where you're focusing both on market share and wallet share. Like I said, not equal efforts at all point, but equal attention. So a profitable growth architect is actually a disruptor, a community builder and a money maker all at the same time. How do you do that? And that's really the thesis of the book, as in, you know, we talk about nine strategies to build towards profitable growth and, you know, demonstrate that you could become a profitable growth architect if you follow those strategies. Four of them for the, you know, zero to one startup phase and five of them for one to five when you're scaling up kind of thing. That's the whole thesis of the book. We actually have this thing called axioms. I want to unpack that too, if you guys are open for.

Yeah, that was going to be our next question.

Oh, great. I mean, so the axiom related to this thought is what I call the 2080 axiom. This is my favorite one. I've seen this over and over again with tech companies. I don't know if you'll agree with it, but let's see how you'll agree with it. Right? 20% of what you build in tech drives 80% of the willingness to pay. This I've seen it over and over again, right? I mean, any product that you say, okay, if you ask people recall on, like, why do you buy this product? What are you paying? It's always like 20% of like, what someone has built. 20% of what you build drives 80% of willingness to pay. The biggest irony in tech is that this 20% is the easiest thing to build. So what happens if you're in that disruptor mindset? You will build that 20%. You'll say, okay, you know, that end is the fastest thing to build. You'll say, okay, let's build it, put it out on the market. Let's call it mvp, give it away for free. So basically, what have you done? You've basically given away the farm and now you're gonna build 80% of ridiculously hard stuff. That's only driving 20% of willingness to pay. Right? So you're basically disrupted, but you have no pathway to get to profitable growth or monetization. You just gave away and you're hoping that something would happen. The right person would say, okay, this is the 20% that's driving willingness to pay. How do I Probably get it in such a way that maybe. Is there some usage things. I want to give it away for free, but after a certain point you have to monetize on that. So can I actually compensate on a land and expand and you'll have that equal attention kick in. And then you'll actually put it in such a way that you will, you know, cross both market share and wallet share, but if you didn't, you'll just go on leaning on one side. Right. I mean, so that we call it the, the 2080 axiom, which is so true every time people will give it away and. And then they're just chasing their tails.

Yeah. And it's surprising how founders, like have such a hard time sometimes being afraid to charge a lot for their.

I mean, in this case, they don't even charge, they just give it. I think we should stop, stop calling things minimum viable product. We should call it most valuable product. This whole MVP definition needs to change. I think it's, it's. I also heard another definition, most lovable product. That's a good one. Actually, I think that's a better definition. But this MVP, which is the 20% that you can build, is the core of the willingness to pay. If you gave that free, tough luck.

What are other axioms that are your favorite?

Yeah, I think there is the axiom that I talk about on the price increase axiom. So usually, I mean, the price increase axiom is that to do a price increase, often the reluctance to do it is internal and emotional. It's not external and logical. And we unpack that axiom in the sense that. I think Warren Buffett said this. Well, he said if you need to have, you know, for doing a 10% price increase in a company, if you need to have a prayer session, you have a terrible business. Right. I mean, so how do you do a price increase if it's a 10%? And we talk about that like, you know, how to actually achieve that. But the axiom there is that every single time we found a price increase being more internal and emotional as opposed to external and logical, there's always a way to do it because, you know, the price of like everything that you consume goes up year over year. Your software or AI cannot be on the same price for the last five years. There's something off. I mean, it's internal emotional, it's not external logical. So that's one that I.

That's why we, that's why we love network effects. Right. Because it's like if you, as you, as you Scale the business without necessarily providing more features. From a software perspective, the value of the product increases over time. And so if you, and you know, you and then actually sort of pricing and increasing pricing over time is sort of like a natural part of the service you provide or increasing value within the product you're providing.

Oh, totally. And every person who brings in another person makes it valuable for the whole network. Right. I love network businesses too.

Another axiom that I like, it says something like attract customers who won't leave.

Oh, that's a good one. Yeah.

And of choosing your customers, I mean always, but also especially early on.

Yeah, that goes to the, I mean, promotion example that I was giving. Right. I mean, if you acquired customers to three promotions who are going to leave, that's not robust revenue. Right. I mean, that's just revenue. It's not durable. The best way to stop Churn is to acquire customers who won't leave. That is the axiom. Right. Because most people try to stop Churn at the time that someone says I'm going, it's too late. You can prolong it by a few more months, but that person is going to leave. Right. And it's reactive. The proactive way of understanding Churn is say, look at my customer base and say, who are the types of customers who tend to stay longer? Who are the ones who actually tend to engage with me more? The usage, buying more products over a period of time? Who are they? How can I find more of them? Can I transfer all my acquisition dollars to find more of them? Yeah. Then you're stopping Churn because you acquired the right set of customers. Yeah.

It's interesting how sometimes like a retention problem is really an acquisition problem.

There you go. It is totally right. Nice way to phrase it.

And last question on this one. How would you relate this to what we were discussing before about POCs? It's very critical, Right.

For sure. All of these tie back because if you choose customers, I mean, first of all, you know, you need to separate the, you know, tire kickers from the people who really want to use your product. For that, you need to charge for a poc. That itself is a lead qualification mechanism. Like many founders ask, should I charge for a poc? I'm like, hell yes. Because otherwise you're going to get some curious buyer who will take you down a three to six months pathway. Might never buy. They're just curious to see the AI. How do you know they have budget or they want to do something? Right. So charging for a POC and at least identifying Some customers who want to partake in that because they have to put some effort for it. And also like how do you frame it as a business case so that your monetization is part of the value delivery that becomes key so that you're not giving the farm away. So all of these principles actually apply there. So what we talked.

So if you're so let's just say you're an early stage founder, you raise your pre seed or seed round led by NFX and 49 palms.

That's great.

And you're building an AI product. Like what are, like what is perhaps three pricing tips that they should do in the first 90 days.

So the first thing that I would probably, you know, look at is, you know, how do they charge or how do they plan to monetize? Because that's a more strategic and fundamental question as opposed to how much, how much can wait. Right. So that's a 90 day question, like.

Should I the tactics of specifics, specifics around how they will charge, whether that's outcomes, whether that's.

And that's unpacking their entire product, seeing what value they actually offer to their customers, how the customers realize value. Do you have some value elements? For instance, you know, can you charge based on a resolution, tickets, whatever? Or is it like consumption on tokens? Is it like usage based? Is it hybrid? All of that stuff can be, you know, ironed out pretty fast based on the archetype and everything else that we talk about. So that's the first thing that I would do. The second thing that I would do in the first 90 days is to actively start framing PoC conversations as business case conversations. So that to your point, Anna, before that I start selecting customers who actually want to invest time to jointly investigate with these founders that there is value in this AI product as opposed to some just technical diligence. So I would probably, you know, coach them on, on, on that. Right. And that'll be the second thing and.

The third and by business case, you really, it's really deeply understanding the, the needs of the customer in terms of their expectations, their cost structure, business model, et cetera.

Exactly. And the third one is also like preparing the founders on how, how to have these conversations with the, the customer. Right. When there's some. Because what I often find is founders who show up with a product and demonstrate what they can actually do from a technical standpoint. But like are they actually having the right value messages? Are they actually talking through that? So like you know, coaching founders on how to articulate value Create needs, not just discover them. And how do you value, sell and negotiate becomes key. So actually like for instance, the options that we talk about. Right. I mean, things of that nature that you can actually start being more strategic in terms of how you negotiate. And I think that's the other training that we can actually do. I'd love to have a situation of a joint captain with you.

I'm sure, I'm sure, I'm sure. Is there anything else that we missed as founders thinking about pricing in this new world?

One thing that probably I keep getting asked and we sort of unpack that, but if there's interest we can unpack it was all these AI companies are getting to revenue pretty fast. Is that good? Have they cracked the pricing code? Is this durable? I think that's an interesting topic. We kind of talked about the negative margin or neutral margin based on the TechCrunch article. I mean, so that's one aspect of that. But there are several aspects of it which is very interesting, I think, which even investors need to know. There's, there's also like, you know, where am I getting that, let's say revenue from? Is it contracted revenue or is it POC revenue? So there's a lot of, you know, things that you actually see, okay, I hit a 10 million error and you know, I start unpacking it. It's probably more POC revenue than contract. I got contracts from like Tesla, Apple, Google and other companies like, okay, it's a 90 day POC, that's not a contract, right. So I think so let's be all careful about like what these reported revenues mean. That's one aspect. And then there's also, you know, the other aspect of like, is this revenue durable? That's also important question because there's a lot of curiosity on the buyer side to actually, you know, look at these AI products. Is it working for me or not? So there's a lot of curious buying, which means are they going to be there after three to six months? Question mark. So you cannot take something and extrapolate it to 12 months when you've only been there for like three months. You don't even know how Churn is actually going to affect you. Is that durable revenue or not? Key question for us to think about, right? I mean, is this, do they have foundational moats? Is it truly delivering value that people will actually have an ongoing monetization conversation?

Yeah, you see it all the time where you got the board of directors pushing down into their kind of CEOs and VPs saying, what's that AI strategy? And so they go out and sign a bunch of PoCs or experimental budgets to try sort of things to tick the box and then they, and then they move on. Is there any. And just from a founder perspective, is that really just looking at, you know, does the engagement and retention increase over time? Is it, is it as simple as like watching core usage and core monetization within. On a per customer basis on specific cohorts?

Yeah, I would nuance that to say on a segment basis. Like what are the types of segments of customers I'm attracting? What is their usage intensity? Who are the ones that are using? Are they going to tend to stay? So like doing cohort analysis. Exactly. That's the key because if you just look at it on average, averages will always lie. And there's also all these pressures on like what is the ARR? I think we also need to, especially in AI, think about is ARR the only right metric? No, there are probably other things that we also need to look at for the health of the business. And even ARR, how is it even reported? I mean some of the things that we saw. I don't know what you see. Like we met a founder who was actually, we talked about seasonal products, for instance. Right? Seasonal products, seasonal product. I mean it didn't come up in the conversation because it was seasonal. We understood that it was seasonal after talking. But if you look at the financials in terms of how the ARR is done, it's like take the maximum in a five month period and multiply by 12. I'm like, no, that was the peak month. What happens the other month? You can't just do that to extrapolate your ARR. So how is the ARR even computed? Is a question mark, right? I mean like if you don't account for like Xian, how do you do it? So I think being thoughtful about this as a community of like founders, investors and others, like what KPIs are we looking at? Usage retention? You know, I think the seasonal one.

Is interesting one because you could look at, let's just take auditing or taxes, like actually, you know, to many organizations, like I don't want to like staff up, up for a three month period and then staff down again. I like if there's an AI that can do that for me, then it's terrific. And so actually so your kind of fixed labor becomes variable and so you might look at seasonality as negative. But I think we see a lot of companies doing very well because they're just purely seasonal because they don't need to go through the headache of hiring and firing.

Yeah this one was like usage exists in other months but in certain months the usage peaks. Let me leave it at that. Which is a bit different case than what you're saying but yeah it could be negative or positive.

And then thinking through the icps, we see there are a number of small companies intrigued by AI but they don't necessarily have the kind of economic sophistication to think through. Okay if I substitute the with this action with this piece of software then I'm able to see an ROI where you see a lot of the kind of mid market companies that are perhaps more aggressive lorpa freshly run that actually see a lot of adoption for these types of products.

At the same time it used to be that when in B2B in particular you started selling to mid market more and enterprise would come later and now we're seeing all these founders being able to close deals builds very fast with. With enterprise. But. But to your point I completely agree that it's a little given like how much curiosity it is and it's also kind of our jobs to you know, learn to assess also like the quality of that revenue and also founders as well in this process.

That's, that's the right.

Because sometimes you know we don't have time to wait for a renewal. Right. Like to.

Yeah, for sure, for sure. But I think that's also where I mean especially in the businesses that you're probably investing in that network effects itself become a moat and is that strong enough? Is something I'm sure is part of your thesis and is it durable and is it quality revenue? I think those are the diligence that needs to be done as investors but also as founders. Being thoughtful about is my own revenue durable or am I just reporting something that. Do you actually believe it? Ask yourself that question or in the mirror if you can sleep well. Okay. If you can't, what do you need to be answering to actually see if it's durable or not and then work towards it.

Wonderful. Well thanks so much for joining. Great conversation, great insights and excited to do some co investing together.

That would be awesome. Sam.