The State of SaaS Go-to-Market l Theory Ventures General Partner Tomasz Tunguz #saas #b2bsaas

There's actually no difference in performance if you look at ultimate bookings of a company or their ultimate lead conversion rate. Irrespective of whether or not you use AI, it doesn't matter. So today there's no change in ultimate conversion rate as a function of using AI or not. Was fortunate enough to spend dinner with a CRO of a publicly traded company, and I was asking her, are you using an AI sdr? And she said, yes, I am. And I said, how's it going? And she said, you would not believe the amount of pipeline we've been able to generate with an AI sdr. And I asked her, how long? How long have you been using it? And she said, nine months, and it's great. Then I asked her, how much business have you closed? And she said, zero. So it's not to say that AI won't work. It's to say many of us do not know how to use AI. Thank you very much. Pleasure to be with you all today. Thank you very much for coming. It's been about 17 years since I started as a venture capitalist, and in that time I've been fortunate to work in the world of software. There's been a tremendous transformation. When Marc Benioff started Salesforce, he codified the sales playbook, and he talks about it in his book, where there are 115 plays that he run. Mark Roberge, who was a CRO at HubSpot, he then wrote the next book, the Sales Acceleration Formula, which shared deep insights into the quota structures and the evolution of that organization as it went through a high churn model to a low churn model. And in the subsequent 10 years since then, we in the software world have instrumented and optimized and measured every little bit about the way that we sell software. But today, it's all different. And it's all different because the kinds of software that we sell isn't the same. It's not predictable. With Salesforce, if I update a lead from a prospect to a closed loss, it will work the same way every time. But if I use ChatGPT or OpenAI or a call recording software to determine what's going on with a lead or customer prospect, state of the art says it'll work somewhere between 80 to 85% of the time. That has radical implications for the way that people buy software. To give you a sense of this, at theory, we host events with different software buyers. Six months ago, we were asking them, what is the number one prohibition, the thing that inhibits you from buying AI six months Ago the answer was security. Today it's roi. People are struggling to find exactly the value that AI systems are building. This change is happening within a macroeconomic backdrop of significant inflation. So as we were looking at this inflation, I wrote a blog post about two years ago where I said I would expect software companies quotas to increase at the rate of inflation. If inflation is going up 4% per year, if the cost of bread is increasing, shouldn't sales quotas shouldn't sell software? What we've seen in the last six or seven months, most notably with Klarna now cutting Salesforce and other major software products as a way of reducing costs, is in fact the opposite. There's been a real compression in the use of overall software, AI notwithstanding, as part of that AI is ushered in a new era of building, or at least thinking about how to build software companies. I was chatting with this former CEO of a publicly traded CRM company and we were, I asked him what do you think happens to sales development rep counts over the next 12 to 24 months? And he said I think 2/3 of them are out of a job. The entire fundamental unit, the AE plus ASCR plus csm, does that model change? These are big and important questions that we don't really yet have the answers to. We're in the midst of this transition. We hosted an event two, two days ago with the CRO of Hebia, which is a unicorn AI business, and Nala. He was previously at Scale AI, which is another AI unicorn. And we had three founders on stage. We asked them how are you learning to sell in this new ecosystem? And there were lots of different answers. Some companies were sales led, some were product led, some companies were selling value, some companies were selling seats, some had matrix hybrid business models. And the answer is we're figuring it out. The playbook that we've built over the last 20 years basically no longer applies. And we need to reinstrument and redetermine what this new world looks like to track that progress. What we do every year at Theory we run a go to market survey where we survey several hundred startups and large companies publicly traded and we ask them about the health of their underlying business. So here I will share with you the 10 observations from that data. So we'll go through the demographics here pretty quickly. It's about 68% early stage companies. 20, notably 25% of the companies that replied are remote. Most of the respondents actually have more than 36 months of Runway. If we think about one of the previous surveys we ran was about Capital efficiency. During the 2020 era, most companies were burning $5 for every new dollar of bookings. When the capital markets retrenched and the Fed raised rates 550 basis points, companies went from burning $5 for every dollar marginal dollar of ARR to burning 3. Now we're seeing that this and this incredible move to efficiency in, in an even bigger way. I talked to you a little bit about pricing models. So the seat based pricing model had been completely dominant for maybe the first 15 to 20 years of software. In the infrastructure layer, usage based pricing started to take off and it was dominant. Look at snowflakes. Net dollar retention at the time of IPO, 177, 175, 171% customer accounts were basically doubling each year. That in addition to the pressure of the CFO on the seat based model, how much software are we actually using? Has pushed even at the software layer companies to look at usage based pricing. And today it's actually the the dominant across the respondent set. And we'll look at the net dollar retention across these pricing models because there's actually a statistically significant difference in the underlying performance of these businesses as a function of their pricing model irrespective of the layer in the stack in terms of who they sell to. They sell to every buyer imaginable. So pretty good representation. And it's predominantly a mid market group, 37%, around 10 to 50K. So let's dig into the data. Let's look. What are the 10 observations from 2024? The first question we ask is your outlook. The average founder's outlook on 2024 is 6.7. That's up from 6.1 at the height in 2022. So founders are really excited, right? There's been a statistically significant increase in the overall positivity of founders within the ecosystem. The biggest challenge to that outlook, the thing that damps the outlook the most, is an increase in sales cycle. So why did sales cycles increase first? We transitioned from a physical selling universe to a virtual selling universe that elongated sales cycles. When you have elongated sales cycles, it creates pipeline supply shocks. What does that mean? If my pipeline, if my sales cycle doubles, my bookings less than half. And so all of a sudden I have a massive lack in predictability. I don't know how to spend my money. My account executives are not making money, they're not making commission. My business is destabilized. So if your sales cycle lengthens, it's a big impact on cash. I was working with one company, we had A doubling in sales cycle, we tripled burn for that quarter as a result of a 50 or a 60% increase in sales cycle. So no surprise. That sales cycle should ultimately be the biggest driver of optimism or negativity within the ecosystem. And what there's a 20% correlate to funding valuation change. This is remarkable because when. So when we started theory, I modeled our funds and I modeled Series A's at about 10 and a half million entry ticket size. The idea was that at the time, average series A tickets to lead around was about 10, 11, $12 million. And as I mentioned before, the Fed had to raise rates 550 basis points. Now, in every other part of the capital markets, if the Fed raises rates, capital becomes expensive, companies raise less, there's less money to go around. And not in startups. In fact, valuations, series A ticket sizes have increased 42% in the last two years. So there's been massive inflation. And this was a surprise to me. I thought founders, especially the feedback from venture capitalists would have been this is a much harder capital markets environment. US venture funding has gone from 8 to 300 over 15 years. It's now falling to about 150 because 81% of the dollars invested in venture capital at the height of the boom came from non traditional venture capital firms who are all very likely to leave. But what we see within the founder universe is actually people are pretty positive on the fund, on the fundraising ecosystem. And even as the public market multiples have collapsed, you've seen Snowflake go from about 25 to trading at 8 times forward multiple in the last 18 months. Private markets have been resolute and resolved to keep their positivity. We talk a little bit about sales cycles. So half of all respondents, irrespective of the segment of the market that they sell, to see an increase in sales cycle. It's most notable within the mid market. This is actually something that came up within that AI dinner that I mentioned to you a little bit before. We see most AI agentic software companies, people who are selling software that replaces labor. They either sell at the bottom PLG bottoms up or they sell at the very top mid market is much harder. Why is this? Our rationale is that the office of the CFO is now tightening the purse strings, eliminating discretionary budget at the mid market, at the mid manager level. And so CEOs, C levels, CISOs, CTOs, VPs of engineering who come in and say I really want to buy this enterprise software product are able to bypass procurement. They're Able to say we need this, this is the reason why I know what the process is, but we're going to go around it. At the bottom of the market, an individual user still can sign up for a product with a credit card. $9 a month, $15 a month, $20 a month. Both of those parts of the market remain relatively healthy. The mid market is where it's challenged. If you imagine you have 40% over provisioning of seats, there's a friend of mine who is part of a cost cutting exercise at a major publicly traded company. They had signed up a three year hundred million dollars infrastructure deal with less than 3% utilization. Where did that come from? It likely came from a mid manager. And so as a result, the lengthened sales cycles within the mid market, that's where things are the most challenged. And this is true across all of software and AI Payback periods. So how do we define a payback period? A payback period is the number of months required to recoup the cost of customer acquisition outlayed to acquire a customer. We've seen payback periods increase about 12% on average. This is basically linear. It's in line with the increase in sales cycles to offset slower sales cycles and to offset worse payback periods. Quotas actually move linear with them. So the average respondents company increased quota 14% year over year, which is higher than inflation. Inflation is an averaging, call it 4% per year. And so you're looking at more than a doubling of overall inflation in sales quotas. We talked a bit about the Fed raising rates, the increased cost of capital. We talked about how founders are more optimistic about their fundraising expectations, which is complete surprise to me. You look at the IPO market basically non existent. The M and A market's very difficult to sell a business. The cost of capital has gone up throughout all that. Founders remain positive. They expect their valuations to be the Same as in 2023 and the same as in 2022. This data is confounding. So we've seen a lengthening in sales cycle. We've seen buyers be much more scrutineering about what they buy, but when they do buy, there's a higher qualification rate. Okay, 10% increase in sales qualified conversion. Why could this be? If you have a really big pipeline as a sales leader and you're not hitting your number, what are you going to do? You're going to redefine your icp? You're going to redefine the qualification criteria and narrow down the funnel, really focus on that core and it May take you longer to sell to those customers, but they'll likely be much higher quality. And so that's our hypothesis for why we're seeing a roughly 10% increase on the average. And then the 70, the second standard, the first standard deviation now, which is that line on the right is actually a 40% increase in sales qualified conversion. So that means the top third of all businesses actually saw a 40% increase in lead conversion. We talked a little bit about the world of AI and sales AI SCRs, there's a huge continuum. I can have a sales co pilot, I can have a full SDR automation. We asked our respondents, how much are you using AI? What are the impacts? We ran the correlation to understand what does that ultimately do to the sales funnel. And here's the data. And so what you would look at this data and you would say if you are using AI, if you're not using AI, you actually saw a 21% increase in conversion. If you're not sure if you're using AI, which is about 25% of the respondent base, if you're using homegrown solutions, in other words, you built something on ChatGPT or anthropic or others, or you bought something off the shelf. There's this distribution of conversion rates, but if we, we ran the t, we ran the statistical test, there's actually no difference in performance. If you look at ultimate bookings of a company or their ultimate lead conversion rate, irrespective of whether or not you use AI, it doesn't matter. So today there's no change in ultimate conversion rate as a function of using AI or not. Going back to that initial point, that dinner that we had with a bunch of different buyers where they're asking about roi, this is the reason for their question. They're not seeing the roi. I was fortunate enough to spend dinner with a CRO of a publicly traded company and I was asking her, are you using an AI sdr? And she said, yes, I am. And I said, how's it going? And she said, you would not believe the amount of pipeline we've been able to generate with an AI sdr. And I asked her, how long, how long have you been using it? And she said, nine months and it's great. And then I asked her, how much business have you closed? And she said, zero. So it's not to say that AI won't work, it's to say many of us do not know how to use AI. I wrote a blog post a couple days ago. It was called the AI Shibboleth. The shibboleth is a word that you use in secret in order to let people pass. One of the problems that we face as investors is trying to understand what is the unusual, what is the outlier, what is exceptional about a business and try to back those companies. And we've been fortunate enough to back a few of them. And that's what I'm looking for in AI. I'm looking to upload a whole bunch of customer transcripts about the way that people use next generation sales software. And I don't want to hear the average. I don't want to hear that Salesforce is great or HubSpot is great. I want to hear the one voice in the room, the one person back there who is saying, I, I really. This next generation started because it does these three things. I'm looking for the inverse of the average. And I don't know how to do that with an AI. Very few people actually do. When I wrote that blog post, There was a PhD from a great university who wrote back and told me exactly what to plug in. But the reality is we're all learning how to use these tools for the first time. One of the reasons it's so hard to sell is there are no experts because we're inventing it as we go. And so to that point, we still don't know how to use these AI software tools to sell effectively enough to be able to move the conversion metrics. This is an interesting chart which is, if you use AI, is your growth rate meaningfully faster than you thought it would be year over year? And the answer again is no. So there's no impact in overall growth rate change as a function of using AI. All of the respondent populations are effectively identical. This next chart is how much. So the question we asked is how much did you grow? How efficient were you in that growth? And then we asked respondents to estimate how much more efficient did AI make you. So you have this one outlier in the top where one one person said, it makes us about 5x more efficient. Most people were about somewhere between 25 to 75% more efficient as a result of that. That's the Y axis, that's the vertical axis on the x axis. We actually looked at how much more efficient they were year over year and those boxes are between negative 4 and 4%. So there's a dominant belief within the ecosystem that AI is making these processes much more efficient and it's not showing up in the data. So this was the biggest conclusion for me and this has actually been independently substantiated by others, which is if you look at. So there's pricing theory. I love pricing theory, love writing about it. There's an academic paper from the University of Minnesota. A brilliant researcher ran Monte Carlo simulations on how to optimize pricing structures. The dominant pricing structure in both B2C and B2B applications is a cell phone plan. There's a base number of minutes that you get for a particular price. You might buy 500 extra minutes at say a cent per minute. And then if you want any additional minutes above that, you might pay two and a half cents for that minute. That's called a three part tariff. That in game theory is actually the most efficient, most effective pricing strategy, whether or not you're an incumbent, whether you're an up and comer, whether the market is expanding or not. And for a long time this is what many SaaS companies did. They had a base platform fee, 10t platform fee and then you charge four users and they deployed a two part tariff. The idea was that the base platform fee would be used to offset some meaningful fraction of the initial cost of customer acquisition. 40, 60, 70% and then the contribution margin of that individual contract was generated through the seats. As I mentioned before, there's been a broad shift towards usage based pricing or some combination thereof. And that's happened over the last five years. If we look at the net dollar retention change. So how much if customers spent $100 today? Last year the customer spent $100 last year. How much are they spending with us this year? Top quartile used to be around 130%, now it's about 120%. So if we look at net dollar retention, if you're using a combination pricing model, in other words seats plus usage 3x the growth in net dollar retention as, as somebody who's using any other form of pricing. So that three part tariff, that theory that was created in 2012 or 2013 out of the University of Minnesota is actually true. We, we see it in the data. And so at that event last night or two nights ago where we were talking about, we were talking about different pricing models for AI, we asked a number of people, about a third of the people are charging for seats, about another third of the people are charging for usage. And then this, the CRO, Dave who I mentioned before, he raised his hand and asked how many people are charging for value. And not a single person raised their hand. But this is the idea, the idea with a three part tariff is that you're able to charge for value, you generate $10 million worth of marginal ROI and you're able to capture 15, maybe 20, maybe 25% of it. And the combination pricing strategy of a base platform fee plus a usage based component is what's driving that in the market today. So overall, in terms of what we've noticed, founders are more positive today than in 2022. Despite the fact that sales cycles are about 12% longer. That lengthening sales cycle increases the payback period. To offset that payback period, startup founders and CEOs have increased quotas linearly. And in addition, as a result of focusing the icp, all the math basically works and we're back to where we were two years ago. Although the quotas are higher, AI to date has no impact on conversion rate, has no impact on ARR growth, even though people who buy AI perceive it as a meaningful contributor to their overall efficiency gains. And then the last data point is, if you're going to be thinking about changing your pricing model, particularly in this new era, it's really important to think about how you can create a hybrid or a matrix kind of pricing model, because you'll grow, you'll have 6 percentage points more of net dollar retention. And as Einstein said, compounding interest is the most valuable force in the world. And so if you can have those 6%, why wouldn't you? Thank you very much for your time and attention. I appreciate it. Have a wonderful rest of your conference.