Metronome - the billing platform powering OpenAI & Databricks 🔋

CV Deep Dive

Today, we’re talking with Scott Woody, Founder and CEO of Metronome.

Metronome is the leading usage-based billing platform built for modern software companies. With a focus on flexibility and scalability, Metronome helps companies launch and iterate pricing faster, give customers better control over spending, and drive strategic decisions with real-time insights.

Metronome is already powering some of the world’s fastest-growing companies, from tech giants like OpenAI and Databricks to startups just getting off the ground. By simplifying billing for companies of all sizes, Metronome is proving to be an essential tool for businesses looking to scale without getting bogged down by billing infrastructure.

In this conversation, Scott shares the story behind Metronome, the technical challenges they’ve overcome to serve high-scale customers, and what’s next as they continue to refine their platform to meet the needs of smaller businesses.

Let’s dive in ⚡️

Read time: 8 mins

Our Chat with Scott 💬

Scott - welcome to Cerebral Valley! First off, give us a bit about your background and what led you to start Metronome? 

Hey there! My name is Scott and I’m the Founder and CEO of Metronome. A bit of background: I was previously at Dropbox for about six and a half years, where I worked as an engineering lead on commercialization and monetization. My role involved new product initiatives, pricing strategies, and overseeing the first 30 days of the user experience. It was essentially a growth-focused role, trying to find ways to monetize and grow the platform.

What was fascinating about Dropbox is that we had a massive amount of traffic, so we could run A/B tests on pricing and see statistically significant results in as little as 12 to 48 hours. But every time we tried to implement a pricing change, we ran into serious bottlenecks with the billing system. The system was home-built and had about 70 engineers maintaining it. So, making even minor pricing changes could take anywhere from three to twelve months. In many cases, it wasn’t even possible to execute some of the commercial models we wanted to try.

Then there was the issue of customer confusion. The billing system was completely disconnected from the product experience, so customers often had no idea what they were paying for. And as a growth team, we should have been learning from our launches, but all the data was buried in spreadsheets and general ledgers, making it nearly impossible to analyze in a timely manner. It could take six months of financial analysis to determine that a product we spent three months building had little to no impact.

Metronome was built to solve all those issues. It's a flexible pricing engine built on top of a real-time data streaming system, kind of like Datadog. We make that data accessible through APIs, alerts, and monitors, and we integrate it back into dashboards for customers, providing tools like cost control and explorers. It's deeply embedded into the workflows of our customers to eliminate the friction that previously existed.

Give us a top level overview of Metronome - how would you describe the startup to those who are maybe less familiar with you? 

Metronome is essentially designed to be a data platform that accelerates the growth of your business. The core concept is to stream all the telemetry data showing how your customers are using your product. Then, within Metronome, you configure your pricing and packaging. From that point forward, Metronome continuously computes invoices on your behalf and makes them available to your end customers. This allows for a richer product experience, as customers can consistently understand how they’re using your product and how much they’re spending. The result is that users feel more confident that they’re being charged fairly, and ultimately, they end up buying more of your product.

Talk to us about your users today. You have some very notable companies in your customer base - which areas of the market finding the most value in what you’re building with Metronome? 

There are two segments I'm particularly excited about. With Metronome, we went against conventional wisdom by starting with the highest complexity enterprise customers, like Databricks and NVIDIA. The goal was to prove we could operate at the highest possible scale, positioning ourselves as a forever billing platform.

More recently, especially with the rise of AI, we’re seeing a lot of traction in earlier-stage markets. There are companies, often just two people in a garage starting an AI business, who look at Databricks, OpenAI, or Anthropic and think, "I want that business model and pricing structure." We provide them with the same platform that these big names use. From day one, they can have a fully built-out pricing and packaging solution that can confidently scale with them all the way to the end. Our goal is to be a universal solution, scalable from a two-person startup to a multibillion-dollar public unicorn company.

How do you measure the impact that Metronome is having on your key customers? Any use-cases that you’d like to share?  

There are three main ways I think about the value Metronome provides. 

First, take OpenAI as an example. When we started working with them, a pricing change on their internally-built platform took six to eight weeks to roll out into production. Now, with Metronome, that same pricing change takes about one minute and can be done completely through an API or our UI. It’s completely self-serve and can even be scheduled months in advance. Importantly, when it took six to eight weeks, they had only thousands of customers. Now, with tens or hundreds of millions, it’s a completely different game for them.

Second, consider the team size required to handle billing. When we first started with OpenAI, they had half an engineer managing billing. Even after massive growth—including the launch of ChatGPT—they still only have one to one and a half people managing billing on Metronome. To put that in perspective, at a similar revenue scale, Dropbox had 50 to 60 engineers focused on billing. So a big part of the value Metronome provides is removing the need for a huge billing infrastructure team.

Third, Metronome accelerates go-to-market speed and saves headcount, but the data we provide can be used in other strategic ways as well. For instance, large language model companies can use our data to understand per-customer margins on a per-model basis. They can analyze costs by customer, model, and discount scheme to make better financial decisions. In addition to this, we help with fraud prevention and provide data that improves strategic decision-making.

Those are a few examples of how we think about the value Metronome delivers.

How have you thought about incorporating generative AI into Metronome’s core product? Any anecdotes from your daily workflows that have improved? 

There are two main ways I personally interact with GenAI in the context of Metronome. First, we have around 45 engineers, and the ones at the top end of the productivity scale are heavily leveraging GenAI to completely change how they approach coding. The days of writing some complex script and testing it manually in the terminal feel outdated now. With coding LLMs, the process is much faster and more efficient, especially when we need to write one-off scripts for our complex product with large customers.

Second, I’ve found GenAI incredibly useful for my own writing. I spend a lot of time writing, and LLMs have really transformed that process for me. Now, I can dump a bunch of rough ideas or content, feed it to an LLM that’s been fine-tuned on my writing style, and it instantly cleans up the text. This saves me hours by turning something basic into something pretty solid, and then I can take it that last mile to make it really good. It’s dramatically increased my throughput for both internal and external communications.

As for bringing LLMs into the Metronome product itself, we haven’t fully integrated them yet, since we're a financial product and need to ensure exact accuracy. However, we do find them helpful when doing integrations with Metronome. For instance, we feed our full OpenAPI spec into an LLM, and it can generate SDK-like scripts based on our APIs, which has been very useful.

What’s been the hardest technical challenge you’ve had to overcome to get Metronome to where it is today? 

In the past, one of the most significant experiences we had was scaling alongside some of the fastest-growing companies in the world. As a two-year-old company, we went through our own version of a "ChatGPT moment" when we had to rapidly scale alongside OpenAI. We heavily relied on Kafka, which is built for that level of scale, and it worked fine for the most part. However, we were also doing stream processing on top of Kafka, such as generating continuous views of invoices through stream processing.

What we found was that, when you’re operating at that kind of scale with rapid growth, many of the primitives in the scaling ecosystem revealed bugs. We were forced to battle-test Kafka streams, making sure everything worked as the traffic scaled up non-linearly. There was a three-month period where we were working 24/7 to ensure that our Kafka deployment kept up with this massive growth.

Currently, many of our customers want both the power and speed of a data streaming architecture, but they also want the flexibility of more SQL-like query patterns. To address this, we've been heavily leveraging modern OLAP databases alongside our streaming architecture. This allows us to process events in real time, while also offering a more expressive, SQL-based paradigm for exploring certain types of data workloads. The challenge is balancing these two—maintaining an efficient, low-latency pipeline while also providing a more powerful, albeit slower and more expensive, way to query data. All the while, we must ensure 100% financial accuracy since we’re a financial system.

How do you plan on Metronome progressing over the next 6-12 months? Anything specific on your roadmap that new or existing customers should be excited for? 

We've spent the past year and a half working with some of the largest usage-based billing companies in the world to build a data model that is perfectly suited for scaling, from a small team of two people all the way to massive, global-scale businesses. We've invested a lot of effort into refining this new data model and product line, which we are launching today!

In the short term, over the next one to two quarters, our main focus is on making this powerful system much more approachable and self-serve. We're putting a lot of effort into improving our documentation, quick start guides, and overall developer ecosystem, so that small businesses can get up and running on Metronome without needing to speak with anyone. Historically, we've focused on the most complex, large-scale enterprises to prove the system's capabilities at the highest level. Now, we're ensuring that smaller businesses can confidently use Metronome as the one and only billing system they’ll ever need.

Lastly, tell us a little bit about the team and culture at Metronome. How big is the company now, and what do you look for in prospective team members that are joining?

We're hiring across the board right now, especially in engineering, design, sales, and marketing. These are our key areas of focus as we continue to grow. Culturally, Metronome is a hardworking team, deeply committed to helping scale some of the fastest-growing companies in the world. Because we operate as a financial system, accuracy is crucial, and we take that responsibility very seriously.

That said, we balance the intensity of our day-to-day work with a fun and whimsical company culture. We're big on karaoke, and every year we host an offsite camp in Northern California, complete with karaoke competitions, games, and team-building activities. Despite the seriousness of our work, we embrace a lighthearted atmosphere. Our mascot is a gnome, and you’ll find gnome-themed items all over the office—it’s a playful, quirky vibe that counters the rigor of our mission.

Conclusion

To stay up to date on the latest with Metronome, learn more about them here.

Read our past few Deep Dives below:

• Our chat with Datastax's Chief Product Officer, Ed Anuff ❇️

• TensorStax is building Autonomous AI for MLOps 🌐

• Dashworks, your AI-powered internal search tool 🔍

• Delphi - the verified digital cloning platform 👥

• CodeGPT is intertwining human language with code 🌏

If you would like us to ‘Deep Dive’ a founder, team or product launch, please reply to this email ([email protected]) or DM us on Twitter or LinkedIn.

Join Slack | All Events | Jobs