Amplitude 2026 Q3 Earnings Call
Review the key takeaways and the transcript of this earnings call.
- Q3 revenue was $111 million, up 25% year over year and 10% quarter over quarter; total ARR was $431 million, up 24% year over year and $21 million sequentially.
- Net new ARR of $21 million comprised an estimated $15 million from Amplitude and $6 million from StatSig; total remaining performance obligations grew 27% year over year to $497 million.
- Customers with more than $100K in ARR increased 31% year over year to 855, and 50% of customers had multiple products, accounting for 81% of ARR.
- In-period net dollar retention was 105% on a pro forma basis, including StatSig and Amplitude customers; gross margin was 70%, down approximately 6 points year over year and 1 point sequentially.
- Non-GAAP operating income was $5.7 million, or 5% of revenue, the company’s highest quarterly non-GAAP profitability; non-GAAP net income per share was 4 cents on 138.4 million diluted shares.
- Free cash flow was $5.8 million, or 5% of revenue, compared with $3.4 million, or 4% of revenue, in the same period last year; Amplitude ended the quarter with approximately $155 million in cash and investments.
- Amplitude said it added or expanded relationships with T-Mobile, Kalki, Fireworks AI, a leading British airline, F5 Networks, SoFi, Hilton Resorts, Coursera, and Adidas.
STOCKNOW INSIGHTS
Continue with outlook and guidance.
Log in to unlock executive comments and Q&A highlights.
Log in for the full summaryStockNow uses AI to translate and summarize earnings calls. Accuracy and completeness are not guaranteed.
Transcript
Preview the first fifteen paragraphs, organized by speaker.
With that, I will hand the call over to Spencer.
Thank you, John. Good afternoon, everyone. Welcome to Amplitude's third quarter 2026 earnings call. Today, I will cover three things. First, our Q3 results. Second, how we have transformed Amplitude into an AI company and why every company I talk to now wants to learn how they can do the same. Third, a look at our product and a spotlight on our customers. Let me start with the numbers. Q3 revenue was $111 million, up 25% year-over-year. Total annual recurring revenue was $431 million, up 24% year-over-year, and up $21 million from last quarter. Non-GAAP operating income was $5.7 million, our highest quarter of non-GAAP profitability ever. Customers with more than $100 thousand in ARR grew to 855, an increase of 31% year-over-year. Enterprise customers are turning to us as a trusted partner in their AI transformation and are adopting a greater breadth of products from us.
We are growing ARR at the fastest rate since 2023 while also achieving the highest profitability we have ever recorded. Coding has changed dramatically over the last two years with the rise of long-running agents. Output is exploding. The amount of code being generated is growing six times per year. The lengths of tasks AI can handle is growing seven times per year. We are on an incredible exponential. Every CIO I know wants the same thing, though. They are saying that they are spending a fortune on tokens, but is that turning into something that they want? That is why Amplitude is essential as enterprises adopt AI. The bottleneck is no longer producing more software. It is knowing what to build and whether it is working. That is the core of what we do. We help our customers build better products.
When I talk with executives, they all want our help making AI transformation successful in their business. Every one of them is at a different point in that journey. At the start, on the left here, are companies on legacy disconnected point solutions, and at the other end, on the right, are bleeding-edge companies using coding agents and context agents like Wave to enable self-improving products. Most companies today sit on the left side in the first two categories. We have a massive opportunity to help companies change their tech stack from legacy players that were built through M&A over a decade ago. We meet customers where they are at, help them take their businesses to the next step, and so are seeing wins across the life cycle. Our platform is designed to bring customers from wherever they are through an AI transformation.
Amplitude gives customers the deepest understanding of how users use their product. Our agents increasingly do that discovery for them. Statsig gives customers feature flagging and experimentation built on top of the world's most advanced stats engine with an engineering-first view. Our customers can now test and ship what features deliver the best outcome. Wave is the future of product development, self-improving products where we automatically recommend what customers should build next based on signals from users. While we are early here, I am very excited to show you a demo today. Together, these three products close the product development loop, understand what is happening, measure what ships, and ship what matters. That loop is how AI-native businesses build, and it is where every enterprise wants to get their product development process to. Let us start with Statsig. We are now six months into our journey with Statsig and our customers.
It is going better than I expected. We have accelerated our development on the platform and continue to release improvements to what is already a best-in-class experimentation platform. Experimentation is important in this new age of building because it can prove what is working. It can close the loop on self-improving products when paired with best-in-class analytics. This is the power of having the context for what is working and a platform for testing those results. We are working closely with data warehouses to unlock capabilities across experimentation and analytics. I am also thrilled to bring Statsig's founder and CEO, Vijaye Raji, to our board. Vijay founded and led Statsig through its acquisition to OpenAI and is now CTO of applications at OpenAI. Statsig and Amplitude's history are deeply intertwined.
Amplitude was initially inspired by Meta's growth team and their finding that making seven friends in 10 days was a strong predictor of long-term engagement for a new user. We were convinced at the time that everyone would need the same type of insights, and so we started Amplitude in 2012. Vijay saw firsthand the power of Meta's internal tools, including Deltoid, which helped teams measure the impact of their experiments. Meta helped pioneer the modern data science and experimentation stack that is now standard practice across the industry. In 2021, Vijay left Meta to bring those same capabilities to the rest of the world through Statsig. For a number of years, we were competitors. We both had an engineering-led approach, working hands-on with customers and building around their needs. Amplitude focused on its core of analytics, and Statsig focused on experimentation.
We were both making sophisticated measurement tools accessible, and we pushed each other to get better. Earlier this year, Vijay was looking for the right home for Statsig's product, brand, and customers. We agreed that combining with Amplitude was the best way to continue Statsig's journey. I am excited to work with someone who has thought so deeply about the same mission. It is an honor to continue that work together with Vijay joining our board of directors. Now I want to talk about some of the newer products we have released. First, Agent Analytics. Every company is trying to understand how they can deploy AI agents to serve their customers, but they are often stuck asking if those agents are working. Most observability tools can tell you whether an agent is running and how many tokens it is consuming.
They can't tell you whether the responses are any good or if they're improving customer outcomes. That's why we built Agent Analytics. Every agent conversation is automatically scored. Did it complete the task? Was the response good? Was the user frustrated? Then we connect those scores to everything else we know about the customer in Amplitude. You measure your agents the same way you measure the rest of your product. Agent Analytics is about improving agents. Product teams can define evals, identify the signals behind good and bad experiences, and see where their agents consistently succeeded or failed. It gives the people building AI products the kind of feedback loop AI engineers use to evaluate and improve their agents, connected to their users' behavior and to business outcomes. Let's go ahead and take a look at how this works.
This is a list of all agent sessions for a local car dealership called SF Bay Auto. Their agent is their chatbot on the homepage. I can click into individual sessions and see standard AI observability features like cost, token usage, and duration. Agent Analytics also has out-of-the-box evals called signals that judge qualities like task completion, response helpfulness, and session safety. Customers can create bespoke evals based on the needs of their business. Agent Analytics can also connect that agent behavior to business outcomes like conversion and retention. Here, we can see the conversion rate of people who visited the page, interacted with an agent, and went on to submit a price inquiry on a car. We can compare people who interacted with an agent with those who didn't and see how that agent interaction impacted inquiry rates.
We can also compare how using different models impacts rates of how many people submit an inquiry. This understanding is crucial to deploying, maintaining, and improving agents as they become the future front door for customers. We launched Agent Analytics to all customers in August. In just over a month, over 125 customers are using it. We helped HYBRD, an AI-driven fitness coach, connect its agent, Brain, to the rest of their customer journey data to understand if their Brain agent was helping their athletes. HYBRD built an evaluator with Amplitude's Global Agent to flag conversations where someone had to ask twice for a tool to run and better understand where the agent was failing. They discovered that 25% of conversations in the prior week were resulting in frustrated users.
The HYBRD team was able to tweak the prompts to the Brain agent and ship a fix within a day, and the rate dropped significantly. They also discovered that athletes who engaged with the Brain agent retained four times higher than those who didn't. They found this by connecting agent conversations with retention and adoption metrics in Amplitude. Now they're focusing their resources on building better onboarding to the Brain agent and building out more capabilities to keep athletes engaged. This is the power of our platform combined with agent understanding that only Amplitude can do. We're now deploying Agent Analytics at scale. One of our largest customers is currently running millions of agent conversations through it to help refine their own agents and improve their conversion funnels at scale. We use it ourselves as well. We used Agent Analytics to test an open-weight model in our own global agent.
We found where it was failing, fixed it, and matched the quality of a leading frontier model at about a third of the cost. Now let's go on to the highlight of the product section today. I'm really excited to show you Wave. Wave is our agentic product team. It looks across everything that Amplitude knows about how customers use a product 24/7. Analytics, Session Replays, experiments, feedback, agent conversations, and more. It finds opportunities worth acting on, explains its evidence, and proposes a change. Then it hands that change to a person or to a coding agent to deploy, and then it measures whether that change worked. Wave tells you what's worth building and whether it made an impact on the business, because it has the full context from the rest of the user journey.
This is the next level of what Amplitude has always done, help teams understand what users do and decide what to build next. Wave makes that whole process proactive using the behavioral data you've collected and product best practices to surface the right opportunities and actions automatically for product teams. Software factories are increasingly becoming a big part of software development. Fleets of coding agents write and ship code continuously, but they don't have a context layer that can inform them what to build. Wave is that product behavior context layer. Here's a real-world example of how we used Wave to improve Amplitude itself. Wave is constantly looking for improvements across all areas of Amplitude, including on our blog. Here, it found a problem.
Our blog gets thousands and thousands of views a week, but the conversion rate to signing up from Amplitude after reading a blog post is artificially low. It realizes it should put in a topic-matched call to action in the text of the blog post that tells a user to sign up. So, for example, if a blog post is about churn, it says, "Sign up with Amplitude to do a churn analysis now." Wave figured this out on its own without any human intervention. It created a visualization of the problem, it proposed a redesigned solution, and here's what it specifically directs us to do: add calls to action that matches the article's topic and give readers a useful next step. When we shipped it helped us write the PR and create the experiment as well.
FULL TRANSCRIPT
Continue the full translated transcript in StockNow.
Log in to unlock every statement, the English original, and speaker-by-speaker history.
Log in for the full transcriptCall participants
14 people spoke on this call — only 2 are shown here.
PARTICIPANT LIST
View participant details in StockNow.
Log in to see executives and analysts, their roles, and complete speaking history.
Log in to view all participantsKeep exploring
