The Generative AI Disruption on SaaS Growth and Churn Rates

Live from Schenectady, New York, it's SaaS Talk with the Metrics Brothers, Growth and CAC.

And I'm Growth, better known as Ray Reich, the founder and CEO of Benchmarkit.

And I'm CAC, better known as Dave Kellogg, independent consultant, EIR at Balderson Capital, and the author of CalBlog.

And together we are the Metrix Brothers.

And we go together like bratwurst and beer.

Okay, food and alcohol again, Dave. Are you in Germany this time?

No, no, I'm up here in Oregon, Ray. I'm up here for a few weeks in Bend.

Wait a minute, the whole beer garden thing is a, is a thing up there in Bend?

Absolutely, Ray. Food trucks, food courts. We just got a new one near our house. Craft beers on tap. It's a very big thing.

Well, maybe I should leave Pennsylvania because my wife just built a new vegetable garden. And I can tell you, I think I would prefer the Ben beer gardens. But let's move on to today's topic. What are we going to talk about today, Dave?

So we're going to discuss growth and churn implications of generative AI on B2B SaaS, inspired by a recent June 2024 report that you in fact found, Ray, done by McKinsey.

Yeah, it's the McKinsey report and it's entitled Navigating the Disruption That Generative AI Will Cause in Software. In fact, the actual report's name is Navigating the Generative AI Disruption in Software. And I'll tell you, those McKinsey people are awful smart, Dave, 'cause I had to read it 2 or 3 times and have you kind of explain a couple things to me. So thank you.

Yeah, okay, well, that was fun. Look, before we do that, we should hear a word from our sponsor.

The Metrics Brothers is proudly sponsored by Benchmarkit, The B2B AI and software industry's largest, broadest, and most contextualized benchmarking index. Dave Kellogg and Ray Groth-Reich have put the Metrics Brothers podcast on hiatus for the remainder of 2026. All episodes we have ever recorded are still available on your favorite podcasting app. You can reach Dave at Kellogg and Ray on LinkedIn at Ray Reich. Thank you so much for listening to the Metrics Brothers podcast. Okay, Dave, let's dive into the McKinsey report and the impact on SaaS growth and churn rates.

Let's do it, Ray.

Okay, so as we said, the report's name is Navigating the Generative AI Disruption in Software by McKinsey, and there's 3 high-level things the report covers. Number one, they have a prediction of generative AI penetration velocity versus SaaS in the early days. Second, they have a prediction for how generative AI will impact the switching from vendor A to vendor B, specifically how it's going to impact traditional software and SaaS vendors. And then third, where the rewards of GenAI will go to, what's the category. So why don't we jump into it real quick? And, you know, at the very top, what jumped out at you, Dave?

Well, let's go to the first thing you talked about, Ray, which is that this is happening fast, that GenAI is going to do in terms of market penetration, GenAI is going to do in 2 years what it took SaaS 4 years to do. I.e., in this particular case, to get to 2% market penetration, which doesn't sound like a big number, but it is. GenAI is going to get there in 2 years and it took SaaS 4. The other way to say this is that GenAI is projected, it's different data, to hit 10% penetration in the next 4 to 5 years, which is roughly 3x the penetration of SaaS at a similar level of maturity. So, so pick whatever Whatever metric you want to use to measure it, but AI is going to penetrate enterprise software much faster than SaaS did.

Well, I am growth, and I want to talk about the growth that generative AI will facilitate. But the first thing that McKinsey did after showing that kind of penetration velocity, and this is on Exhibit 2 of page 3 of the report, they looked at how switching from vendor A to vendor B will be impacted by generative AI solutions. And what's interesting, Dave, kind of at the highest level, what I took away from this slide, which really kind of shows almost an NRR-type waterfall, is that today the revenue switching impact is about 10 points from one vendor to another. But with generative AI, that revenue switch rate will go up to about 15 to 20%. So anywhere from 1.5 to 2x higher. That was really shocking to me.

Yeah, this exhibit here, Exhibit 2, as Ray mentioned, is basically kind of, they're doing it for the entire software industry, but you could think of it either for a category or a given vendor. It's kind of an NRR analysis that says, hey, if you start out with 100 units on the year, looking only at your install base, what happens to it over the course of a year? And look, they're saying there might be a little bit more revenue decline. Actually, it might go up to double on revenue decline from negative 3 to negative 6. They're saying growth would be a little faster going from 14 to a range of 16 to 20. But the one that does catch your attention is the one Ray just mentioned, which is they said roughly 10% of people a year switched vendors kind of pre-generative AI, i.e., in regular SaaS, but they expect that to double. And that does beg the question why. And I think it's very important for vendors to think about this because ultimately, numbers aside, this is saying people have a new willingness to switch vendors in B2B SaaS, which they did not have before because B2B SaaS has always been a pretty sticky category and people needed a reason to switch. So, and I think AI simply put is giving them that reason to switch. And I think it's going to have certain impacts on vendors, which we can talk about in a minute. But Ray, do you have any other thoughts on like why people are switching or the impacts of switching as we see kind of a more, I almost want to say frothy or churning software business?

Well, as I read this and reread this chart, the one thing I looked at was revenue decline, sometimes known as down sells, and revenue switch, oftentimes will be known as churn, right, to the incumbent, where that adds up to 13 points traditionally. That gives me about an 87% gross revenue retention, Dave, and that's right at median. But in this new world, it says I could have a 19 to 25% ARR churn, which would put me up to a 75% GRR, which is not a good thing. Is that the way you read it? Does that make sense?

Yeah, I think it's a good interpretation of it, Ray, which is GRR will go down to roughly from 85 to 75, I think is the rough numbers if you believe this report. And the reason it's going to go down is people are going to be switching more. And the chart's a little confusing because they show both the negative and positive effects of switching. Because if they're switching more, you're going to get more people switching to you as a way of people people switching as well as people switching from you. But that's what I'm trying to say, it's gonna make the industry a little bit more frothy, a little less sticky, a little more switchy. And I believe that. Ultimately, here's why I believe it, Ray, because I think people feel like AI is gonna reinvent software, and I think they're right, and therefore they're more willing to switch. I think if your product is roughly the same animal but incremental better, People are like, it took me a long time to buy that system, put it in, it works. You get this amount of inertia in B2B SaaS and AI is going to break some of that inertia. So your customers, when a new vendor calls up and say, we've invented, reinvented this category in an AI-first way, I think they're going to be a little bit more willing to listen.

Oh, I totally agree with that. There's going to be some, I'll call it emotional buying. We want to go where the market's going, not where it's been. But the next exhibit, Exhibit 3, for me, Dave, is something I really want to dig into and get your feedback on. So in this Exhibit 3 on page 4, it first of all shows enterprise spending on generative AI. And it shows in 2023, it was around $15 billion. And by 2027, they're estimating $175 to $250 billion. Now, Dave, that says in 3 to 4 years, the generative AI spending is going to be as large as the entire SaaS marketplace is today, which I think was about $200 billion last year. So that was one thing. But then the second thing that really jumped out at me was where, where the enterprise is going to be investing that money. And what it says is that 35 to 45% of the spending will be having an embedded gen AI in a broader software platform. So maybe that's in their ERP or their CRM. So 35 to 45%. What did you think of that?

Well, let me go back to the first point. So the growth from $15 billion to $175 to $250 billion, my read on that is simple. If you ever wondered why all the VC money is flowing into AI, that's why, because people think this market is literally going to explode out from underneath us. I don't even have the CAGR on that. But it's very, very fast growth, right? $15 billion today, $200 billion roughly in 2027. So that's what's driving— these kind of forecasts are what's driving all the VC investment. In terms of where it's going to get spent, Ray, look, they're saying, I'll call it a third, a third, a third. Okay. They make it a little more complicated than that, but just to keep it simple, they're saying, hey, a third of spending is going to be on in-house generative AI solutions, which I found surprising. A third will be on GenAI point solutions, which I don't think either one of us found surprising. That makes sense, right? AI-driven solutions to solve specific problems. And the third, which I'm not sure if you found it surprising or not, but you noted it, was that it'll be on embedded generative AI in broader software platforms. I'm not sure if that's just the incumbents, to be honest. It could be new broader platforms that have embedded AI in them, or it could be incumbents who have, you know, put AI into their platform. The thing I found most striking, Ray, and they talked about this in the report, was the first one, which is AI is gonna finally enable— I've been hearing this since 1985 when I first heard about 4th generation languages, right? Literally for my entire career, we've talked about citizen developers and now, you know, regular humans could build applications. You won't need to be an engineer. Well, we're hearing it again. And this time they might be right that with AI, that kind of regular people, and they're not necessarily saying this is pure end users, but that kind of regular people will be able to build these solutions. So you're not going to buy them. And that's certainly not something you heard a lot of in SaaS. So that one I think is true and noted because it's going to affect how big the market is. Because a lot of point solutions, like for example, Adeo, that sells a kind of roll-your-own CRM system that, you know, I don't think people would have wanted to hear about 10 years ago. They'd be like, what? Well, how would I kind of roll my own CRM? The answer is you can do it today.

Yeah. Well, I'm still a little freaked out by the fact that almost a third of the spending will be on in-house GenAI solutions, because that definitely wasn't the way it happened in business applications in the cloud era or the SaaS era. So I'm not sure that most corporations are gonna have that intellectual capital to do that successfully, Dave.

Yeah, I mean, put it this way, Ray, what drove the massive expansion of application software back in the day was the fact that you didn't wanna build in-house. People were buying relational databases and trying to build their own finance system and manufacturing system and HR system. And vendors came along and said, you know what, we'll do that for you, PeopleSoft, SAP, et cetera. That's what drove their growth, was specifically not to do this. So, we've heard this claim before. This time, I think it might be true, however. So, I think I might disagree with you on that one. But look, I think they'll get further. I don't think you're gonna build your own ERP, but people are kind of— and I don't know if you consider ADO the bottom category of this category. It's probably the bottom category because you're going to use their platform to roll your own CRM. But, uh, I, I do think you'll be able to build stuff more easily now.

Well, that's the other thing. They also on the same chart showed which function we're going to have the greatest, um, enterprise spending for GenAI. And, you know, sales and channels 11%, marketing 14%, customer care, customer service 15%. Okay, those all made sense. But then, Dave, You look at the largest segment, they're saying that 40% of enterprise spend will be in IT, information technology. And I still can't quite make sense of that. Have you thought about that?

So look, I think the first thing is, well, it might actually relate to your first point, which is who's able to use AI to build their own apps. And certainly IT would be able to do that better than other departments, certainly better than marketing or HR. Second, I mean, there is a large— look, I'm a huge believer in internal AI for case resolution. So to the extent IT is doing things like case management, trouble ticketing, they can certainly benefit from using AI-based systems to do that, automating a ton of it. I work with several companies that do that. I have a feeling it's also going to be used for configuring, configuration, those sort of tasks. So I'm trying to look up here, Ray, in the report exactly where they say it's going to get used. I'm having trouble finding it. I don't know if you can.

Oh, But yeah, yeah, in fact, let me dig into that because it's Exhibit 4 on page 7 of the report, and it talks about where will it mostly disrupt the majority by software category. And it has a ranking from very high disruption to high to medium-high down to medium. Is that the one you're thinking about, Dave?

Yeah, yeah, because basically to me this seems to be a warning shot over the bow of the entire kind of data warehouse infrastructure industry, because the first thing they're putting there is data aggregation along with information services. So I'm assuming that's kind of the ETL market and the data integration market of tying together all these different data sources, which AI is quite good at. Information services, I think, is a different definition. So I think they have bound 2 pretty different things together there.

Totally agree. And then the next really big one is content creation. Now, this one really resonated with me. Think about your designers, your Copywriters, your, you know, your marketing communications people. They said that's going to be one of the ones impacted the most. And then of course, customer service.

Yeah. And they have 2 things on this. One is the size of the disruption. The other is the color. The darker the color, the greater the degree. And that's where they have customer service. Absolutely. As a person who ran the Service Cloud at Salesforce for a while, absolutely, AI will disrupt and is disrupting customer service for the same reasons it's disrupting internal help desk, for IT and HR services. The other thing they talk about is data and analytics, particularly data and AI development and data solutions, which I think is another area that's going to be changed pretty radically.

I agree.

The biggest one they have though in dollar volume, it's only a medium, it's a medium kind of impact, but the biggest dollar impact is CRM. And they're calling it ERM, but I think they basically mean ERP. So kind of core systems. So they think they're going to be kind of medium disrupted, medium high. But, but, but on a very, very big space.

But you know what's interesting to me, Dave, because I've been trying to do a lot of research, talking to a lot of people, and what— there's 2 competitive moats that I think are pretty good for a legacy SaaS vendor to have versus new point solutions, and that's data and it's distribution. So when I saw this CRM one, of course I went to Salesforce and I'm like, well, Salesforce has so much data and such large distribution, i.e., number of customers, that if anybody can really fend off these point solutions, it should be a company like Salesforce. Does that resonate with you?

Yeah, it does, Ray.

Yeah, and it's interesting, I was just— we're talking about growth rates, and of course I think for the first time the guidance for Salesforce is down a little bit. So I think with the addition of Einstein and all their AI, I actually think we could see reacceleration of Salesforce as their enterprise customers start looking at deploying generative AI more and more for their sales, marketing, and customer service processes.

Yeah, I mean, I believe that Salesforce has done a good job inoculating against AI-driven startups. So I'm not— I didn't follow their most recent earnings calls or guidance or anything, but clearly over the last several years, they've been marketing Einstein pretty hard. And they certainly have not been asleep at the switch on AI as some other incumbent vendors may have been in other categories. So I think they're doing it correctly in terms of both where to deploy resources and how to message to the market. Basically, I don't think Salesforce customers feel like if they just stick with Salesforce, they're going to get left behind. They may not be at the cutting edge on every point solution. Got it, right? I don't know if Salesforce has an AI SDR yet like Qualified does. Right. So you may not have all the latest and greatest, but I don't think you feel like you're going to get left behind. And that's the big challenge for the incumbent vendor in my mind.

Yeah, I totally agree. And then if you look at, you know, what McKinsey said is the cat— one of the categories that they've listed here is having only medium disruption risk. It was system infrastructure. So I'm still trying to reconcile this. So maybe you can help me, Dave. So I understand why they're not going to be switching out entire system infrastructure. But if you're building a lot of your internal generative AI apps, aren't you possibly going to be implementing different hyperscaler infrastructures to do that or using some of the foundational models as part of your system infrastructure on a go-forward basis?

Yeah, I mean, their description of this, Ray, is they say that smaller impacts to operating systems, computing, and IT services due to their essential nature, but key innovations are likely to continue. So I think they're saying it's a big space. It's going to— and by the way, the lowest end of their scale is medium. That's probably the answer to your question, right?

Right.

It's like olives. The smallest you can buy is jumbo, right? So the smallest level of disruption on their scale is still medium. So they're not saying this stuff is going to be unchanged. They're just saying relative to, say, CRM or relative to, for example, content creation, where the impact is listed as very high, it's not going to change that much. But yes, it's going to change.

Now, we've only got a couple minutes left, so I kind of jumped ahead to our next episode, Dave, where we're going to be covering the Bessemer Ventures State of the Cloud Report, and they're talking about the AI cloud. So I was thinking about, boy, what can traditional SaaS vendors do to prepare their product and even their go-to-market strategy for this generative AI disruption? And one of the things I was thinking about, man, if you don't have some AI generative AI on your product roadmap today, or something you're going to be releasing in the next 6 months or so, you better over-index on that because I think your risk of churn is going to be much higher. And what Bessemer is saying is only 43% of Bessemer portfolio companies— or I should say their investors— say that all their portfolio companies actually have generative AI on their roadmap. And in fact, almost 32% saying less than 25% of their portfolio companies have generative AI on their roadmap, which is a scary proposition for me.

Got it. So I think first, that's gonna be our next episode, right? So you're gun jumping.

I am.

The charts you're referring to, well, look, one of the interesting things, we'll do a preview. I'll join you on the preview, right? is that the Bessemer 2024 State of the Cloud, they list a bunch of issues, and then they gotta do an internal survey of Bessemer staff. Because unlike, look, Ray, the original Bessemer State of the Cloud report had one author on the byline, Byron Dieter, and it was 2015. And this year, there are 13 authors on the byline. So I think one of the reasons they have these surveys is they're going, I don't think, I think they're serving more broadly than the 13 authors. But I think they're going inside Bessemer to just say, basically surveying people. So they're kind of saying, here's the Bessemer point of view on this issue. And in this particular question you're talking about, it's what percent of your companies have AI-driven features on the roadmap? And you're right, 43% are all of them, 0% are none of them. So most people have it, everybody has something on their roadmap, but only 10%, well, Actually, I'm having trouble reading this chart now, Ray. Um, what percent of your companies say about 10%? 2% say about 10%. This is confusing. Yeah.

Yeah. What we're going to do is we're going to entice our audience to join us for the next episode and we'll go into all the detail. But I think suffice it to say for right now, if you're leading a B2B SaaS company and you don't have a generative AI capability leveraging your data on your roadmap, It's probably a good time to accelerate that investment. That's my perspective.

Yeah. And the takeaway from the 2 charts you presented, which are not Bessemer's best work in my mind, they're on different scales and other stuff. We'll talk about these charts next week. But I'll tell you my takeaway. 43% of them say, basically 43% of the people they surveyed said nearly everybody's got AI on their roadmap today. And what percent of your companies plan to have AI-driven features in the roadmap in 3 years? 77%. So, so it sounds like nearly half the people who work at— Ray, I'm fumbling. I don't know what the hell it says. What does it say, Ray?

Well, that's what it says. 43% of the investors at Bessemer are saying the majority of their portfolio companies have generative AI in their roadmap.

That's what it says.

And by the way, if I compared this to A VC who had enterprise software company, when I say enterprise, on-premise software in 2002, I will guarantee you not 43% of them were thinking about going to the cloud and selling it as a subscription. Think, you know, Siebel, Oracle. They made that transition much slower.

Yeah, I see what you're saying there, Ray. That's a good point. So, we'll come back to the best of our thing in our next episode, but that point's a really good one, Ray. Which is, I think, cloud, the cloud disruption came with a kind of the, what is it, Elisabeth Kübler-Ross, 5 stages of death and dying, right? Denial. I think cloud started with denial. You had Larry Ellison telling people it was snake oil. You had a lot. So the first reaction to cloud was denial, and AI has bypassed that step. I don't think there's a single big software vendor that goes, nah, AI is a fad, you don't need it. So that is an interesting observation.

Yeah. So if I was to summarize, you know, my takeaway from today's podcast, McKinsey report, it is integrate generative AI into your roadmap quickly, introduce it to your customer base quickly, or be prepared for something more into 75% to 80% GRR versus that traditional 86%, 87%. That's my big takeaway.

And mine would be, if you've got a point solution that wasn't possible before, market the heck out of it. Like AI SDRs, they didn't exist before. If you're a new vendor reinventing a category, drive that message really hard. And if you're an incumbent vendor, go inoculate against that message because there's going to be people out there saying, we've reinvented your kind of staid old category and you need— and your customers are worried they're going to get left behind. So you need to convince them that you're kind of AI-embracing enough that that's not going to happen.

I agree with all that. And the last thing, since I am growth, if you see a market going from $15 billion to $175 billion over the next 4 years, I want to be in that 100% CAGR. That's all I know.

Well, if you don't grow fast enough, you get left behind in hypergrowth markets. That's clear. Um, it's, it's one of the reasons, by the way, why all the VC is flooding in. And why for the AI companies I look at, when people have me review their financial plans, it's like from another era that the numbers these guys are, how fast they're growing and they plan to grow. But you need to, because if the market is exploding and you don't plan to explode, you get left behind, which is not a good outcome.

But Dave, as you know, no longer is the technology industry going to have growth at any cost. Oh, maybe that's another episode.

Yeah, except, see asterisk, except for hypergrowth markets. So yeah, I don't think there's gonna be a lot of concerns about CAC and AI in the next couple of years, but we shall see.

Totally agree. Hey Dave, thank you so much. This was a fun topic for me.

Yep, thanks, Ray.

Bye-bye.

The Metrics Brothers AI, SaaS, and More is a production of The Metrics Brothers Growth and CAC and a member of the Benchmarkit Podcast Network. By accessing this podcast, you acknowledge that The Metrics Brothers make no warranty, guarantee, or representation as to the accuracy or sufficiency of the information provided Or the humor content of the jokes told. Ray, information, opinions, and recommendations presented are, according to our spouses, probably wrong and provided for information purposes only. This podcast should not be considered professional or, for that matter, unprofessional advice. We disclaim any and all liability for any direct, indirect, undirect, misdirect, incidental, special, ordinary, consequential, inconsequential, or other damages arising out of any use of or God help you reliance upon. Upon the information presented here. Ray Grothreich is based in New York City and available on X at Ray Reich. Dave Kellogg is based in Silicon Valley and available on X at Kellogg. Schedectity, which is French for unspellable, is not our actual production location. Thanks for listening.