Good afternoon. Thank you for joining us, day two of the Citi Global TMT Conference. My name is YC Wong. I am part of the software analysts here at Citi. We are happy to have CFO of Amplitude, Andrew Casey, with us. Welcome, Andrew.
Thank you for hosting us.
Look forward to it.
Yeah.
Maybe just give us a quick background about yourself. You have been at the company for, what, two plus years?
And a quick background of the company for those who are new to the name.
YC, I am very old, and I have been in the Valley a long time, so too much background would take too long. You would not have any time to ask questions. But I started my career in the Valley, in Silicon Valley, in 1995, working for Sun Microsystems. I then went to Oracle, not through the acquisition. I was there for a couple of years, then went to Symantec, then I went to HP, and then I went to a company that very few people have ever heard of called ServiceNow.
Yeah.
I was there for a little while and really cut my teeth on building out a scalable enterprise software business.
Right. For sure. Everybody knows Mike Scarpelli over there, right?
Yeah. Mike was a great mentor and great CFO to work for. I aspire to try to live up to his standards. I probably would still be at ServiceNow if they would've made me the CFO, but the board really wanted somebody who'd already been a CFO. At that point in time, John and Bill were in transition, and both of them asked me to stay, but I told them that, "Look, even if Gina leaves within five years, the fact of the matter is, by your criteria and the board's criteria, I still wouldn't be qualified for the job." That's why I left, and I went to a company called WalkMe, helped take them public in 2021. I then left WalkMe before the sale to SAP and went to a company called Lacework, which was in cloud security. We sold that to Fortinet.
The funny story I tell people is that in August of 2024, I was signing the definitive agreement to sell Lacework to Fortinet at noon, and by 2:00 P.M., I was announced as the CFO of Amplitude, and been with the company for two years now.
Wow. Congrats on that journey. That's definitely long tenure in software.
Yes.
Yeah. Maybe going to Amplitude currently is not too shabby to manage a company on this growth trajectory right now, especially in the past few quarters. Can you give us a sense of overview what happened that saw the re-acceleration in the business?
Yeah, I would say it's funny, going back two years, John leads our IR capabilities and corp dev. He had done a synopsis of all of our transcripts and our earnings releases and other things. You go back to that, and there were two major strategies that we were talking about when I first joined Amplitude. First was increasing the amount of business that we do with enterprises. What we mean by enterprises is any company that has 1,000 employees or has $100 million in revenue. Okay? So increasing the business with enterprises, everything's better with enterprises, gross retention, net retention potential, you name it. We wanted to have a greater footprint with enterprise clients. The other thing that we were looking at was really broadening out our product portfolio.
One of the things that Spenser, our CEO, and the other founders had noticed was that there were all these applications that had been created around product analytics, which is what Amplitude at its core is, product analytics. Think about any software that you distribute.
You need to have instrumentation and observability associated with the application, the website, the wearable. You want information and feedback to see how customers are interacting with it. Well, all these other applications had been created around that, around Experiment and Session Replay and Guides and Surveys.
Yeah.
The postulate was that really those didn't need to be separate. They really should be part of a broader platform. So those two strategies were, when I first joined, we were trying to get off the ground. We were trying to get in motion. There was just a host of things that we did on both fronts. Building new products ourselves.
acquiring companies to augment the capabilities, really engineering the applications together so they work very seamlessly in customers' workflows. Changing the go-to-market profile, the process, territory design, pricing and packaging. All these things we were pulling levers on to really go drive those strategies. I would tell you that the acceleration you saw in revenue from when I joined in Q2 2024, it was 6%, all the way up to our most recent
Yeah
announcement around 22. I would say it is kind of boring, but I would go back to those things and say
this is just us doing very well against some of those core strategies, those core actions that were meant to go drive both those strategies. Now, sprinkle that, too, with we have increasingly added more agentic capabilities into our platform. We are releasing more agentic products for fee, and that is broadening our appeal and our
Yeah
and our interest from those classic enterprise clients who are asking the key question of, "How do I adopt some of these new
AI capabilities and get the return on investment that I'm expecting?
Yeah. There's a lot of levers, a lot of products in there that are helping drive the growth.
Since we are more in a CFO discussion, maybe can you talk about what are some of the key metrics that you are seeing that give you the confidence of that 20% trajectory?
Well, certainly as we go forward and the underpinnings of our guidance where it shows accelerating our revenue and our growth is based upon customers' adoption, our pipeline itself, opportunities we see around moving into new product categories. I think all of those things are giving us a lot of benefit. I would say one of the strategies that we implemented was to go drive longer-term relationships with our clients. When you're driving a consolidation story in the market, which is us going into an enterprise client-
and saying, "Look, you've got Fullstory, and you've got Pendo, and you've got AB Tasty. We can consolidate all of those onto our platform, and we'll charge you less in aggregate fee, and you'll be more efficient." When you do that type of consolidation play, one of the things you realize quickly is that customers are not going to just rip out all those things at the same time. They want consistency in the rollout of how they're replacing that and influencing workflows. That usually means they want a longer-term arrangement. For us, that's exactly what we wanted as well. We want that longer-term arrangement because it not only gives us greater predictability, but when we sign up a client who is doing a multi-phase replacement of applications, they're committing to that.
Yeah.
When they commit to it, that shows it up in RPO. RPO has been growing over 30% for six consecutive quarters.
Right.
Now, why is that important? Well, because it's greater revenue visibility.
It gives us greater revenue predictability, and it reduces the actual amount of renewals that you have to do in any given period, because you've got a broader set of contracts there for a longer
Right
period of time. When I first joined, we had to renew 89% of the install base.
Yeah
we had to renew 72% the next year, 56% the next year, and we're driving that down further and further. I'd much prefer a contract duration which isn't 22 months, which we're currently at, but more closer to 30+ months. That enables me to have an investment in sales and marketing, which is more focused on new logos and expansions
Yeah
rather than just doing renewals. All of these strategies, I would say, are instrument. In fact, I remember it wasn't maybe a year ago that an analyst asked me, "Well, isn't this just luck?" I said, "No, it's not luck.
It's like the COVID luck.
Yeah. Driving the RPO growth was absolutely something we instrumented. We rewrote the comp plans. The incentive structures are there. We taught sales how to do deal constructs which align to the customer's value propositions. We got them into the right framework around give and gets, around getting to a longer contract duration. Pricing and packaging reinforces this as well. All these things are, as you mentioned earlier, the levers we were pulling-
Right
that would generate growth.
Yeah. How does this longer duration contract would have an impact on your net new retention, and then also how is it impacting your net new logo versus expansion drivers?
It's a good question, and I would tell you that just intuitively, if I have to renew less next year, even if I had the same gross retention rate, churn would be down.
Mm-hmm. Yeah.
Churn dollars are down, then net is up, so it is driving growth. I would tell you that the reality is when you execute well on that front, and sales has to spend less time on doing renewals, they can spend more time going after new logos. By the way, we have slowly but surely ratcheted up the focus from the sales team on generating new logos. Every year, we have added things to the comp plan structure or the territory design in which they have to bring in new enterprise logos in order to meet their quotas.
These are all tactics and instruments that we have used
Yeah
to go drive the right behaviors and to achieve the core KPIs that we look at. Obviously, core to us is ARR, and you know this because you have studied us, but the reality is I take a very conservative perspective and definition around ARR. It is exactly what we will recognize over a 12 month period. There are no caveats. There are no exclusions on contracts. There is no termination for convenience. There is no monthly annualization. It is committed contracts, and that is what we hold our sales reps accountable to as well, creating committed contracts.
Yeah. Since you joined, I think enterprise has been a much bigger focus. Is that enterprise driving to have this longer-term contract versus shorter term, or it is more the level that Amplitude is pulling?
I would say it is more Amplitude in the construct, but you lean into the customer motivations. Like I mentioned before, customer wants to replace multiple applications.
Over a period of time. They actually want cost predictability.
Right
Which means they want a longer-term contract.
Yeah
They have that cost predictability. A champion who is doing the work, who is actually doing the replacement of those applications with Amplitude, looks at their success as Amplitude being successful in that environment. They want to have a partner in that, and a way to have a partnership in that is you have that longer-term contract.
It is a multitude of factors. I would say the contract duration expanding is a result of the actions we are taking to align to the customers themselves.
Okay. Yeah, I guess with the contract duration, there is also a pricing impact, which has been a big topic over the past few quarters as a new pricing model.
Could you help us refresh our memory on the old pricing and the new pricing? What are the key changes?
Sure. When I first joined, I gave a report out to the board, and I do a very classic 30, 60, 90 day report out. The first report I gave had a section on it, basically said, "Our pricing is not aligned to our strategy. It's causing churn, it's causing complexities, it's causing a lower adoption rate." The board basically said, "Andrew, why is this the case? Tell me." I went through, basically, the pricing that we had was very much a point product. Every product had its own meter, its own price curve, its own value proposition, its own. In fact, it was a separate price, different SDK. It wasn't reflecting what we were increasingly doing in the platform, bringing products together. Customers were complaining.
It was very much an optimization for a single product rather than an optimization associated with the experience that we wanted our clients to have in having access to all those applications, and us being able to monetize it very easily. All those things that were wrong are things that we've changed. The first thing we did is we started talking to our customers and understanding where do we think that this platform sale is going to go? Does it need a different type of meter? Ever since we've been founded, we've used event volume as the primary meter.
Right.
I would say 86% of our install base had that. When we were talking about all these other products having different meters, if you were a sales rep at Amplitude but prior to making the change, and the customer wanted to adopt the entire platform, you'd have to ask them for five different estimates on how they're going to use the platform. If they were over one of those estimates and under in another, we'd still charge them for the additional amount over.
Okay
One of the individuals. So it was very much a point product, rather than the customers looking at, "Oh, well, I don't really understand exactly what my adoption pattern was going to be. This is what my best estimate is." So we asked them, "Is event volume still the primary way in which you believe you're getting value from Amplitude as you use more?" We came back, and it was still one that was very valid with clients. The industry has adopted event volume as a standard metric. So it wasn't a problem with the meter itself. Now, that doesn't mean you get the price point right. You have to understand that when a customer uses more, they do expect that they're going to get incremental discount associated with the unit price they're paying.
That is just standard.
Yeah. Especially longer duration. They are like, "Okay. What discount are you giving me?
Exactly. Well, we actually did not have that. We did not have a volume-based curve for.
Okay. Yeah.
For quoting. The reps were quoting prices that were all over the place. Customers talk and they go, "How come I got a better price than the other one?" Having a standard framework for volume-based discount is something we had to go
build and develop, and that got changed. Now we can go in and talk to a customer like, "Look, if you want to have 1.5 billion events, it's this price.
Yeah.
If you move up to 2 billion, it'll be this price." Now, obviously, you architect that there's marginal incremental discounts, but as they move from one tier to the other, they are going to pay more in ARR, just at a decreasing rate. All that architecture had to be done. The other thing that customers were very clear about was, "I need greater cost transparency on what I'm going to be paying Amplitude as I adopt more of your platform. I can't have all these different meters. You're just going to push your compliance over to me." What we did is, because we indexed off of event volume and that price point and did the volume-based curve, we then indexed every other application as an uplift on that price.
Now it's very simple for a customer to compute what they're going to pay Amplitude as they adopt more of the platform.
Yeah.
That trend
What is that uplift that you are seeing on the new pricing versus the old pricing?
Each module, it ranges in its, I will call it value attribution to a client. You have Experiment, Guides and Surveys, Activation, Experiment. They all have a range of about 25%-50% uplifts, individually.
Okay. All right.
Then you have other ones that are smaller, like AI Feedback, which is more like 10%, Session Replay, which is more like 20%. But they each add up, and if you did each one of them and you're adopting the full platform, you'd probably pay somewhere between 3x to 4x, which you would if you were adopting product analytics alone. But it's very simple for a customer to compute what that's going to be.
Yeah.
It's very easy for sales reps to quote. A funny thing happens, when sales reps actually quote more often, they have more engagements, they actually drive greater velocity, forecasting improves, pipeline improves. So the underpinnings of driving greater efficiency on the sales side, greater adoption from customers, having them give them confidence and transparency in what their cost is going to be, and driving multi-product adoption. All these things were objectives we had with the new pricing and packaging, and I would tell you that we're doing very well against each one of them. Now, we're rolling it out in a very methodical way
to our sales team. I'll tell you a funny story. When I was in the boardroom and we were talking about the new pricing and packaging and changing it, one of our board members said, "You know, Andrew, CFOs and CEOs have been fired for getting this wrong.
Yeah.
I said, "Well, I guess,
Some risk worth taking.
Yeah. I said, "I guess we shouldn't get it wrong.
Yeah.
We did a lot of testing with clients and got a lot of good feedback. That is why this last quarter, 70% of all transactions,
Yeah
including renewals, were on new pricing and packaging. I expect that we are going to continue to see that ratchet up to where it is 100% of every transaction every quarter. Now we have roughly 28% of ARR that is on new pricing and packaging.
I expect that that's going to get into upwards in the 60% by the time we end the year. The good thing is we're seeing already that average ARR is increasing.
Mm-hmm. Yeah.
Contract duration's increasing, multi-product adoption's increasing. All those things benefit all other key metrics like gross retention, net dollar retention, and customer adoption.
Yeah. The uplift that you mentioned, overall uplift of that, especially the 70% in Q2, expecting for 100% going forward, do you see the uplift going up, increasing as we progress?
It's been about 5%-10% based on the customer transactions itself. I actually think that it probably will continue in that way. I don't think it'll be dramatic, though.
The one thing that you've got some customers who are overpriced on some things, and they're being corrected. You had some that are underpriced that we're taking a nice progression with. I think that the biggest lever that we're seeing with the pricing packages as it applies to uplifting ARR, though, is multi-product.
Yeah.
Additional product they're adopting.
Just given the pricing is still at very early stages, right, in some ways, are you worried there are going to be some optimization going forward?
There certainly will be, because some of those price points that our sales reps would quote when they didn't have a volume-based discount card.
Okay
were outsized. There certainly will be. We've addressed those as we've gone forward and, in some cases, you do have contraction if it was egregious. Other cases, we found our way into selling them more product in order to absorb those amounts. I wouldn't say it's been a huge headwind, though. I think it's been a very modest one.
Okay. No, that's definitely good to track, see where that's going. Now, I want to pivot a little bit, just given there is 10 minutes left, and then we still haven't talked about the newest and greatest stuff we have.
That's right.
Statsig acquisition.
Yep.
Love to hear your thoughts on why you acquired them, then what do you see with Statsig in the product lineup?
Yeah. There's a couple things you have to understand that are going on in the market right now. First is that every enterprise probably either has or is talking about ways in which to centralize their data in a data warehouse, through Snowflake or Databricks or another. That's a major trend happening. Two, one thing we definitely know about agentic capabilities is that the ability to create code has never been so fast. You can now have an agent creating code for you in specific areas. Code development, the whole build and ship process of the product development life cycle is exponentially grown, okay? As part of that, there's also an increasing effort to figure out how you fix the next part of the bottleneck in the product development life cycle process around the QA and test and learn.
One of the things that the industry has been struggling with is now we're creating all this code, is it actually working as it's intended to? Are we seeing the benefits associated with it? That's where Statsig had done a really good job about appealing to engineering use cases in the product development life cycle process, Experiment and testing how that code has actually been deployed. They'd done it in a data warehouse native environment. Amplitude had an Experiment product as well, but it was primarily in the cloud-based environments. Okay? We were starting to build a product
to go after this opportunity because those major trends we see happening. When OpenAI approached us about acquiring the technology, it was very much when we looked at as, look, this is the leading technology in this space. They have got a bevy of customers for it, and we believe this is going to be a fundamental addition to Amplitude's efforts longer term around our agent analytics, what we see happening in the product development life cycle process, and the benefit of having agentic capabilities deployed in our latest product Wave, incorporating the Experiment efforts as well. Think about it in terms now, Wave, which we are in beta, and we demoed on our last earnings call. We think of it as an agent that is encapsulating what you would normally have a product engineer, a data scientist, and a developer all working towards creating a product.
Well, you can have an agent actually doing that work for you, recommending what should change next in that product or that process, and bringing in the Experiment capabilities associated with Statsig and doing the experiments and giving the feedback on it all encapsulated with your process.
Yeah, for sure. I wanted to touch on Wave, but staying with Statsig for now, following the acquisition, I know there is only the product side because the whole team technically just remained with OpenAI. What have Amplitude did to integrate that technology? How was the hiring? Because now you are selling to different persona from Statsig business. How has that transition been?
It has been actually better than what we expected. As part of the agreement with OpenAI, we only had a 60 day transition period to learn the code and set up a Google environment to actually host it, because a lot of those things did not come over. Day one, we were hiring new engineers to staff up the team. That has gone very well. We have got a good team now that is working on it, especially on support. The selling side, I think, is an interesting one. It is certainly a more technical sale related to an engineering persona, and our sellers had not been used to that in the past. We have hired what I will call some specialists, but I would not call them overlays. I would call them more augmentation.
They are more like we always had a technical group that was in our sales team that was helping out when we had these highly technical sales, and now there is just some specialists that actually handle the engineering use case and the Statsig Experiment product itself.
Do you need something like an FDE that everyone has been talking about?
We do have FDEs. I think that it is not necessarily with Statsig, although that can be the case. I think FDEs are where customers are coming to us and asking: How do I get that? I really do not even know where to start. If that is the case, then you deploy Forward Deployed Engineers to actually show customers how they could change their business process and become or integrate more agentic capabilities in order to make those processes better. Maybe it is through cost efficiencies they are looking at or optimize their revenue funnels.
Okay. I guess just given in general, looking as capital allocation, you have Statsig that is being opportunistic. How do you view the market opportunity that in the next 6-12 months on your capital allocation needs?
Yeah. I think the next 6 - 12 months, I think there's a couple things that are really interesting to me. I think every enterprise is starting to deploy, in some fashion,
agentic capabilities, whether that's customer service related, fraud detection, business process optimization. There's just a host of areas where enterprises, as they digitally engage with their clients, are looking to augment with agents. That's a surface and an area where we believe we can sell agent analytics around because every customer is going to want to know, is the agent actually doing what was intended to do? If not, why not? What would you change? I think it's a huge opportunity for us to go sell into that. We're already dealing with a number of clients in the financial services sector, in the telecommunications and media sector, with the exact intent. Even if we're not selling the agent, we're simply surrounding the agent and making those interactions better. So we think it's a huge opportunity.
Yeah. As you expand on, especially new agent use cases, anything that you see on the next frontier that is interesting?
Yeah. I think this whole move towards these router optimization companies associated
Yeah
with the usage of models is an interesting area as well because Amplitude could add its own behavioral heuristics around it to help optimize for the outcomes. Think about in terms of most of these companies are approaching model usage based on the individual model capabilities and the cost structure, not necessarily engineering it for the outcome they are trying to achieve. I think that is an area where we could play as well.
Yeah. Maybe just guess on the Wave product that you talked about earlier, right? That is kind of an agent that you can ask to do things. Would that have any cannibalization risk as to the core analytics or Experiment product? How do we think about it as that agent continue to improve or even the models continue to improve? We have Astra last week, that is just an exponential improvement.
Yeah. Look, I think, Wave is an interesting product for us because it, like Statsig, it is a whole new potential land for us. You actually do not need to be an Amplitude customer to get benefit associated with Wave. However, if you are an Amplitude customer, it is a whole set of dataset that you can leverage in order to make that agent, as it is deployed either against your website or your mobile application or your product development life cycle.
All of that is more richer data for that process to be optimized. So it is not an either/or, it is you deploy Wave along with your Amplitude infrastructure and datasets and your connections makes it that much better as you are actually deploying it. Pick your favorite use case. But it is one that it is not working in against Amplitude, it is actually working with Amplitude, and you can use it outside of Amplitude as well.
You could literally deploy Wave against one of our competitor's products as well. Use that as a data source.
Yeah. As Wave scales usage across Amplitude, do you anticipate any margin compression? How does it work versus your core analytics product and Wave?
Well, Wave will be its own price point, its own meter. We will price it certainly to optimize for value, as much as possible. We will have agentic capabilities embedded within it. I see it as a little different from our core analytics. Think about it like our core analytics products are capturing event data. Let's use a marketing use case for an e-commerce site. It is capturing tons of event data on the e-commerce site, and you can deploy Wave to actually optimize and personalize every interaction that people are having on that e-commerce site, because it is recommending and taking action based upon parameters you have actually specified for it.
Do you see it as an option to do something like outcome-based pricing where a lot more people are talking about?
Outcome-based pricing is interesting thing. I think what you do in your pricing structures is you optimize for the customer achieving the outcome. The thing we think we can do for Wave is that we can give very detailed level of feedback that as customers spend money on tokens for a specific thing, that they can actually see what the return is associated with those tokens so they can get to, that is the outcome.
Yeah.
I don't believe that you take every customer's unique outcome and then try to embed that within a unique pricing model. If you did that, you'd have a unique contract every time, because customers have different desires.
Yeah.
We have financial services clients who are trying to use or testing Wave to do loan optimizations.
You've got telecommunications and media that's trying to optimize our marketing use cases around promotions. If I started pricing on both those things, suddenly I've got a lot of unique instances and it's very difficult to scale that. The architecture in pricing is when you've got to get to a meter, which a big rule in pricing construct is you should have a meter which you anticipate is constantly going to increase. The perception of that meter has got to be one that customers look at it and say, "I'm getting greater value as I use more." If those exist, then you can get scale in your pricing architecture and still achieve those value-based outcomes based on the price point you actually charge.
Okay. No, I think that's a good way to start since pricing has been one of the bigger lever for Amplitude to continue their growth. Well, Andrew, thank you so much for joining us.
Pleasure