How are you?
Okay, thumbs up. Doing well.
Yeah.
Yeah. Thanks for doing this.
Yeah.
Good afternoon, everybody. Welcome to the 27th Annual KeyBanc Capital Markets Technology Leadership Forum. My name's Jackson Ader, the enterprise Software Analyst here at KeyBanc. We're thrilled to have Andrew here from Amplitude.
Thanks for having me.
Introduce yourself and the company. I have a bunch of questions, and then we will open it up to the audience every so often. But take it away.
Okay. Andrew Casey, I am the CFO of Amplitude. I joined now two years ago, believe it or not. Maybe the best way to start was, when I joined, Amplitude had two major strategies it was focused on. First was building out a set of capabilities, applications that surrounded our core product analytics business, and the other major strategy was increasingly selling more to enterprise clients. Both strategies had not made a whole lot of progress, and when I joined, that was the remit from Spenser, our CEO, and others was like: We need to start making progress on that front. The engineering team did a great job of increasingly identifying areas where other businesses had been created around experimentation capabilities or Session Replay capabilities.
There were just all these other companies that had created things around product analytics that Spenser had postulated, "Look, all these things should really be part of one platform." The engineering team did a great job on building some, and then we acquired a few, but I think the real linchpin was they brought them together in a very cohesive way where workflows became much more streamlined, data taxonomy became much easier, and we had increasingly started talking to customers about a consolidation play that they really should drive more of their requirements onto the Amplitude platform. Equally at that time, there was an ongoing change out of the selling organization to drive more and more enterprise sellers and focus more on enterprise clients.
If you look back, if you decomposed our ARR at the time, we were in right around 60% of that we would consider enterprise. We've raised that now up to 69%. A lot of that is related right back to success in our capabilities selling to larger and larger enterprises. Both fronts have made substantial progress. I would tell you that if you look at our growth over the last two years, when I joined, we were growing like 6%. Now with Statsig included, we'd say from an ARR perspective, we grew 22% last quarter. The reality is, I would go back to those two key strategies as being the fundamental areas where we got it right on the product side, and we got it right on the enterprise sales motion side.
There was a number of things I can mention, like our pricing and packaging changes, how we're thinking about investments, the acquisitions we've done that have helped augment it. Even if you took out all the things that Amplitude's doing today in AI and said, "Would you still be growing?" I'd say we are, we would, because back to those two core strategies. Now, you throw in there-
Don't give it all away now.
What's that?
We'll hit on all these things.
Okay.
Don't give it all away.
Oh, yeah. Right.
Yeah.
I would tell you this last quarter, maybe I'll just stop there and then we can go into questions. I was very gratified. I think actually, you weren't wrong when on the Q1 press you said, "Hey, there's a lot going on.
Yeah.
We had AI transformation. We had the Statsig acquisition. We had changes in go-to-market, changes in price to packaging. I'd say, yeah, you're right, there was a lot going on.
Yeah.
Looking back, I would say I wouldn't change a thing about taking on those challenges. I'm not saying we executed perfectly, we didn't. We've certainly executed well on those fronts, and it's made all the difference.
Okay. This most recent quarter, you want to just lay the foundation, another quarter of organic acceleration. Like anything in the second quarter that worth highlighting before we get into some of the higher level stuff?
I'm particularly proud, or maybe it's just because I'm responsible for it, but I'm particularly proud of the progress we've made in our pricing and packaging. One of the things that I highlighted to the board, literally my first 30 days, was that our pricing and packaging strategy was wrong. It was organized for a point product architecture. It was causing churn. It was causing friction in the selling process. It was causing friction with customers adopting more of our capabilities. We went through an extensive process of talking with customers and working through different structures. We tested our core meters, and we rolled it out in Q4, did a lot of testing. I think I told you this, too. It was like at a board meeting.
And literally, one of our board members turns to me and says, "You know, Andrew, this all looks good, but CEOs and CFOs have been fired when they get this wrong.
Right.
I thought, "I'm not going to get it wrong." We did a lot of testing and rolled it out in certain test cases for Q4. We did a broader exposure for new customers in Q1, and I would tell you in Q2, we really started seeing pickup, to the extent of 70% of transactions we did were on new pricing and packaging.
Just staying on this, do you mind just telling us again, there were customers were disincentived to really use the platform.
Yeah.
To add modules sometimes was not done in a clean way. Can you just lay the foundation of what was wrong about the pricing before, and now what specifically is the new pricing?
Sure. Let's start with just the construct itself. Amplitude had all these different products, and each one of them had their own meter. Their own price curve, their own value propositions. They were oftentimes not well articulated together. You would have customers who had to give five different meters back to a sales rep in order to quote the full platform. There were paywalls around each product. If a customer wanted to try experimentation, immediately they were met with, "Well, call your rep." They could not even get a taster in it.
Believe it or not, we did not have a standard volume based discount curve. Customers would go, "Well, what if I decide to add 30%? What would be my discount?" The sales rep, "Well, there really is not." "What?" Those things that you would think are just like standards, they just were not in place. All of them have been implemented now, where you have tasters on the modules. They get to a point where the customer is, "Oh, I really like this. Okay, so now I want to know, what is the pricing look like?" Rather than us directing them to, "Hey, it is a different meter, and it is a different price curve," we basically standardized our pricing. By the way, we did a lot of testing to determine whether the event-based meter was the right proxy on value for clients
Okay
and realized that it is. We stuck with that as our primary meter, which 86% of our installed base was already on that, so that was another factor, that they were very used to paying us based on the event volume data and data ingested in the platform. We kept that. We did add a volume-based discount curve. If a customer was paying us for that and they wanted experimentation, we made it very simple to do a 30% uplift on that value. Customers go, "Wait a minute. If I add all the modules together, is it an uplift every time?" We said, "Yes." Now, you might say, "Well, how did that align back to cost?" The biggest cost for us is with analytics. It is that initial data ingestion into the platform.
Experimentation, Session Replay, Guides and Surveys, if we add more modules for customers get more and more profitable for us. It was a natural way to make simplification of the pricing, simplification of the quoting process for sellers. We broke down those paywalls so that customers could get tastes, and we gave greater transparency to customers as they adopted more. All those things incented greater adoption, greater multi-product lands, easier expansions, greater contract durations. Go back to all the things that we've been trying to go drive improvements in our financials, and they have their basis back in a number of different areas, our coverage, our sales incentives, but also in the pricing and packaging now.
Yeah.
That's incenting the right behavior.
You changed quota or sales incentives in order to
Absolutely
drive the, yeah.
Absolutely. Look, I was talking about this earlier. Someone said, "Well, what's your optimal breakdown between new logos and expansions?" I said, "Look, here's the reality. Expansions can be five to 20 times what the initial land is." I was at a sales meeting recently, I told them, "Look, I would much rather have 10 new Global 2000 at $100,000 each than one at $1 million." The reason is because we go prove value, those expansions could go very dramatic. I do target our team to a more balanced approach. We were 30% new logo, 70% expansion, given the multi-product initiatives we had. I really, we've shifted based on the incentives. If the reps don't make their new logo count, then they're not going to make as much as they could-
Yeah
on the incentives. I'd really like to be more 50/50. Right now we're more 40/60, that's an improvement.
Is the marginal cost, is it truly ingest once, sell many products on that ingestion, so the marginal gross margin on the additional modules or products is nearly 99%? Is it that?
If the use case was that one only. Let's say you deploy Amplitude on your mobile application. That was the only footprint that you were going to do. As you're adding more modules, that would be the case.
Okay.
Now, what we're often seeing, though, is customers will add more data sets as well.
Right. Okay. And then cost to ingest the additional data set. Okay.
Yeah, exactly.
All right. Let's talk about AI. You're accelerating. You've made some acquisitions in the AI space. I know that Spenser is very involved in the AI community, so I am sure that there was some internal AI capabilities that you already had. What is the company currently doing in artificial intelligence, and how is it an accelerant?
I would say a couple things. Maybe a little spectrum.
Yep.
First, when we started talking about injecting AI into our products and our capabilities, it was a lot about efficiencies, how you can use our product more effectively. There was a lot of discussion around, should we be charging incrementally for that additional capability? I would say I was on the side, and there were a number of other executives within Amplitude, that our biggest problem was not initially trying to monetize for AI. It was actually driving broader adoption of Amplitude across different business use cases. If we could get to a point where, through agentic capabilities in the platform, we had broader users and a natural language way of interfacing with Amplitude, we would lower that barrier to adoption.
That was what we did initially with our Global Chat and with our MCP capabilities. We are broadening out the use cases for Amplitude. That certainly has increased the inference costs-
Yeah
especially as customers have adopted more. The good thing, though, is we are getting past that initial worry that this was not going to work to drive adoption. There were pricing barriers in place. What we are seeing is that customers are adopting more, and the postulate was always that that would generate greater data ingestion and greater usage of overall platform, because one of the things that the Global Agent does, it exemplifies the capabilities across the platform of all the different modules. Each one of the modules has agentic capabilities as well, which we charge for, right?
It started to introduce other products, like AI Feedback, Agentic Analytics, and Wave, which we demoed on our latest earnings call, all of which we will charge incrementally for. Okay? It opens up a whole new set of revenue opportunities that we just did not have before.
If I am a Global Agent, if I deploy a Global Agent customer that then needs to talk to all the sub-agents, do I need to be a paying customer of all those underlying modules?
You do. After the taster.
If I was paying $100 yesterday, I wasn't an AI customer, and now I'm basically taking the entire platform. What does that $100 look like?
Let's just say in the classic. I won't talk about the whole new products yet, because some of them we haven't priced out yet.
Okay.
The core modules, you'd be paying upwards of $250.
Yeah. Okay. All right. I have to talk about the 20%, right? That is Spenser's floor. I bring it up every time we talk. We are edging closer to it. You are the CFO, right? You are in charge of the numbers. I love the ambition from Spenser, but I guess, can you tell us why does he feel like 20% is the floor? He talks about it in terms of the market itself, not just Amplitude.
Yeah.
This is the moment for analytics, and this is the moment for experimentation, and that this should be, again, not just Amplitude's moment, that it is a market thing.
Yeah. I think, one, his view is somewhat different than mine, and I will explain mine, too. His view is that, look, every customer he talks to needs an observability and instrumentation layer to make whatever they are doing better. Whether you think of that in terms of The Economist who is trying to drive better user engagement, or a fast casual restaurant that is trying to do a loyalty program, or software providers just trying to understand how well their new capabilities are being adopted and implemented in managing security. Everybody needs this and yet there's a whole group of customers that just don't have it. His view is this should at least be this, because there's so much opportunity out there.
Right.
I take a different view in that I don't disagree with Spenser, but I also would say, look, one of the things that's really good is we're increasingly driving contracts with greater number of footprints, with greater contract duration. Our RPO is now growing over 30% for six consecutive quarters, and that's a prelude to
Yeah
future revenue growth. Absent a major falloff on our ARR capabilities and growth in net new ARR, I would say that we're going to continue to drive acceleration. I agree with Spenser, but I would say I root my view in the data and what we're driving. It was in my guidance. People ask me, "Well, how confident are you in your guidance on 21% growth?" I am pretty confident.
Yeah.
The reason is because RPO has been so strong. I do not base guidance on hope and wish. I base it on what I can see and know. From a software business that is generating increasing durability, you get increasing confidence in your ability to go drive growth.
One of the other things that we have liked about the Amplitude story, or the stock story, I hate to put it like that, but is the product diversification, which we have talked about. But also the diversification of the pool of buying
Yeah
within a customer. I don't know. It's like, where does the Chief Product Officer get their dough, right? What budget do they pull from? You don't necessarily talk to enterprise customers where it's like, "Well, we have a Chief Product Officer budget." Right? So some of it can come from marketing, some can come from CIO.
Yeah.
You've been diversifying there, so can you just tell us how that is going in terms of the buying pools within an enterprise?
Yeah, I think we've always had this postulate, at least as long as I've been at Amplitude. We have this postulate that these individual personas across the CIO, the CTO, the Chief Data Officer, marketing, even the CRO, that they all have aspects of where they could be using Amplitude in different use cases. We think that you can use Amplitude in those use cases, and we try to appeal to those users. I would say probably the most tangible one is the chief marketing officer. If you ask me, "Who is our biggest competitors that we go after?" I would say Google and Adobe are right up there.
Yeah.
Then in the more engineering use cases or whether it is chief technology officer, engineering, or chief products officer, depending on the size of the organization, that may be one person. Right? You are appealing to them in how they are building products and instrumentation around that. I would say increasingly, we are getting better and better footprint with those types of customers as well. One of the reasons I was so excited about Statsig is because there is this whole wave in changing how product is developed. Part of that is an increasing mentality that you drive continuous experimentation about what you are developing, and that is exactly what Statsig does. They have had a customer base which is very much centered around the engineering use cases, and ones that have engineering use cases that are specifically related to data warehouse-native environments, which is where we were weak.
Okay.
You are right. I think we appeal to a lot of different data sets, at a lot of different use cases. You pick a different customer in a different industry, and I would tell you that we are appealing to a different persona, potentially. It is the job of our sellers and our SEs to really help customers understand how they could use our product to optimize for their own outcome.
Part of the diversification story is also a diversification away from the concentration in tech.
Yeah.
Right? You are digital natives, and now AI natives are coming to Amplitude. Should investors be worried that it is like, oh, we are getting right back into that spot where we were in, where it is like we are really exposed to this one area of the market? Or are you broadening out into more verticals than you have before?
I think we are broadening out more verticals than we have before. If you look at a classic NAICS code structure in our software by customer and industry, our composition has always bounced in the software area, and I would throw digital natives into that as well, or AI natives into that. It has always been 30% any given quarter. You have some software that is disappearing and some are taking up positions. But then you have companies like Domino's Pizza and Paramount and Hertz Auto. All those customers, I would not call them software related, they are using software to go generate their business.
Right.
There is an increasing footprint that we are developing there, which is a great set of logos.
Is that growing faster than your tech cohort? Like the non-tech versus tech?
I would say no. There's certainly a tech cohort that's been going slightly faster recently.
Yeah.
But it's not that different. I think there's some really big Global 2000s we have
Okay
in the contract base.
I forgot to ping the audience, and we only have two minutes left. Does anybody have any questions? I am sorry about that.
I can stay longer too, if you want.
Yeah, we started a little late.
Sorry.
Yeah. While I ask about margins, you guys can think of a question if you have one. But I think it would be nice if the margins were higher. I understand you are accelerating growth, so you will get half a pass for that. But we have seen incremental margins at other software companies, at other peers, be much stronger than they have been at Amplitude.
Yeah.
Why have they not been?
I think there's a couple things for us. One, Statsig business we took on was not-
Before Statsig, though.
Okay.
This is, you know.
All right. Then you have increasing data ingestion rates. Those are growing 30%. We just talked about that being a big cost component. I think that is a great future predictor of revenue, there is always going to be a lag between when that happens and you actually monetize it. The fact that that is growing at a rapid rate, we are at this point where the amount that we have got contracted, the entitlements, we are seeing that customers are using over 80% of that entitlement, is a great position to be in. When I first joined, we were more in the 60s, that was a predictor of churn and contraction and poor deal structures. It is a great sign that we are seeing customers ingest more and more data into the platform.
So longer term, I would say, hey, we are going to see monetization. In fact, I have already seen some of it. The other side is we took a very specific strategic position to address broader adoption with agentic capabilities and not charge for our Global Agent and our MCP. I think that too, over time, will drive increasing cost of data ingestion, but we will monetize that. It is just the lag. I know what Spenser talked about and said, "Hey, everybody has got to get over it because this is the way AI companies work.
Right.
I think that over the short term, that is true. But I also believe that there are fundamental things happening in the market and our own actions, which we will optimize as much as we can in that cost structure. Everything from using open-weight models to figuring out how you drive better economics through contract structures, optimizations between the models themselves, that is not compromising necessarily what Spenser suggested, which is for those engineering use cases, you want to have the bleeding edge, they want to use it, so we are going to have it. But to say that it is going to continue on in infinitum, I do not agree with that.
I think the bigger issue, frankly, is what is the rate of data ingestion we are seeing and how fast can we do it? Because if that rate continues to accelerate, I guarantee you we are going to have a sideways, but that is going to drive greater revenue growth.
Are per unit costs of data ingestion also going up, or is it just the volume?
Volume.
Just volume.
Yeah.
Okay.
Because you want to remember, outside Statsig, our Amazon Web Services contract, I renegotiated that a year ago, and we're just now getting the benefits of some of those new economies.
Right.
That too will help over a period of time. But yeah, it's the volume that's driving it.
Yeah. All right. Any quick ones from the audience? Sure.
What is the long-term vision for the company that Spenser has? Is it to be analytics-only company going forward?
No, not at all.
Are there other pillars?
No.
Just for the webcast, what is the long-term vision of the company that Spenser has?
Oh, sorry.
No, that's all right.
I would say that we've evolved from that core product analytics to what I would describe as the observability instrumentation layer for every piece of software that's out there. Whether you think about in terms of a product you're developing, a website, a mobile application, a kiosk, there's increasingly more footprints for software, and every one of them needs an instrumentation layer where you're getting feedback on how that software is actually performing. Okay? That includes a whole new set of agentic capabilities and observability, which we're starting to get into. The corollary I would give you is, remember when Datadog first came out, they were talking about observability of infrastructure, then they moved it into application. But it's still a very, I would say, IT specific use case of observability.
We're doing it for the application layer, and over time, maybe those things get closer and closer, but right now, in our use cases, they don't really play. I would say the same type of instrumentation requirement is there. That's the way I think about it, and even if we're evolving into everybody's going to have an agent in some form as part of their digital engagement with clients, we think we can play a key and critical role to making that agentic interaction better.
Okay. Great. Andrew, thank you very much.
Sure.
Got started late. We went over a little bit. That is all right. We fit it all in.
Thank you. Apologize for that.
That is okay.