Today I’m thrilled to share some big news: Census has signed an agreement to be acquired by Fivetran. Putting the two companies together completes a feedback loop we’ve both been working on for years—Fivetran moves data into the warehouse, Census moves trusted, modeled data out to all business functions. Under one roof we’ll be able to give data teams a single, simple path from raw events all the way to action. This marks an exciting new chapter for our company, our team, and most importantly, for our customers.

How We Got Here

Shared Foundings 

When we started Census in 2018, we had a simple yet seemingly impossible goal: “to build a source of truth that could coordinate all of the business operations of a company.” We had experienced first-hand how tools like Salesforce couldn’t scale when organizations had large amounts of data even before the AI era. I was personally obsessed with improving the prevailing messy data integration approach by creating a central hub that would federate to any part of the business. I laid out a high-level recipe for what ingredients one would need to deliver trusted data everywhere.

1. 1. 1. 1. 1. Gather all the raw data from the 4 corners of your business
            2. Unify it into a set of reusable, well-governed data models
            3. Deploy these models wherever the business needs them

This picture only lived in our seed stage pitch deck from 2018, which I’ve included here for posterity 🙂

This is a lot to build when starting out. While talking to our early potential users at Figma, Atrium and Notion, we stumbled on an initial “slice” of the problem: CRMs lacked actionable customer usage data to power sales workflows. At the same time, cloud data warehouses were emerging as a great potential store for this data marrying ease of use with infinite scale. This missing wire was a good first step on the journey to building the brain behind every business: we’d connect data warehouses into SaaS apps. Our prototype was so useful that our first customer – Figma, which was a lot smaller then – bought it immediately even though frankly, we didn’t know how to describe it. This led our first couple of users to call it “reverse Fivetran” which is the origin of the term “Reverse ETL” in the industry today!

Fivetran hasn’t just played a part in the name. Our two companies have been a part of each other’s stories for a lot longer than most realize. The founders met over 12 years ago. We were some of the very first users of Fivetran, at a dinner table in Mountain View, when it was something altogether different than what it is today. More importantly, Fivetran was an essential partner on day 1 and a happy customer to boot. Along with Figma, the data team at Fivetran has been using Census in production since the beginning.

Evolving With Customers

I love working on hard design and engineering problems at Census, but what brings me the most pride and joy is giving users a tool that creates leverage and helps them grow more powerful. I’ve had a front row seat to watch our customers build fantastic businesses using the operational superpowers they developed with us. It’s a privilege to serve them, and they have shaped the product every step of the way.

Our first few users/customers at Figma and Fivetran along with others like Notion, Rippling, and Loom were pioneers of modern GTM operations (what people nowadays call “GTM Engineering”). They understood that spreadsheets & Salesforce wouldn’t cut it, and they wanted the flexibility that came with SQL. This led to our first design enhancement beyond a pipeline that maps tables & columns: allowing users to write arbitrary queries that drove custom ops workflows (e.g. “VIP customer list for express support”).

The Rise of Cloud Data Warehouses

Redshift was our first warehouse integration and for a time, our most popular. It was a lot better than its predecessors. However, connecting tools like Census meant constantly fighting database overload. We started to question whether we’d built the right architecture. Luckily, Snowflake, BigQuery, and Databricks took the world by storm. With their separation of compute & storage, you could run Census and dozens of other workloads on centrally managed data without performance problems. These slow data “sinks” that Fivetran filled with data became able to run operational workloads. We joined a small chorus of players to bless this pattern so that data teams could embrace it.

The Advent of the Modern Data Stack & Reinventing the CDP

This combination of tools centered on the premise that data models should be treated as software assets with version control and software development practices. We evolved early towards this approach with integrations with dbt, GitHub, and a suite of monitoring capabilities. As everyone flocked to it, the pattern became known as the “modern data stack”, which is now so common that it’s just “the data stack.” 

An early picture of what became the core pieces of the “modern data stack” from our 2020 launch.

With Census connecting the cloud data warehouse to every app, it became the de facto source of truth for the most valuable customer metrics & insights. Marketing teams started to shift from traditional CDP solutions that were operating on outdated, incomplete data towards taking action from the warehouse. This pattern of data activation is now the modern standard for marketing organizations at Activision, OrangeTheory Fitness, Canva, Sonos, and thousands more.

The AI Wave

Every business leader I talk to is excited about AI and so am I. It feels like endless possibilities are ahead of us. Customers want to build an agent, they want to “talk” to their data, they want to automate & improve decisions, but it’s all on a wobbly foundation! Data quality is a bigger problem than ever, which is why we recently launched new capabilities for dealing with duplicates with entity resolution, and more tools for enriching data from multiple sources.

With this wave upon us, our users are pushing us to evolve again. Instead of relinquishing human agency to the machines, we launched AI columns to help the best data & operations folks take advantage of LLMs with infinite flexibility and control. It’s no wonder people are referring to modern growth ops as GTM Engineers because the leverage available to a single human using AI is amazing.

With all these changes, speed matters more than ever, in keeping data up-to-date, in reacting to change, in engaging with customers, in running experiments. In all our years working with customers, nobody ever asks for slower data delivery. We just never want to sacrifice trust & quality for speed. Every time we look at how to make any of these performance improvements, we come back to the same answer: a data feedback loop with multiple independent parts, which needs to be closed…

Closing the Loop & Merging with Fivetran

My crude drawing of the modern data feedback loop from our 2021 Series A announcement. With today’s news, this loop can be implemented with a single product!

Our customers (and Fivetran’s) keep asking for us to build more connectors, not just to power activation but for all the activity coming in and out of the warehouse. Together, we can deliver a better user experience for things like authentication, debugging, and a lot more. Last year I laid out a vision in which a single platform could ingest, unify, make sense of, and activate data – not too unlike that picture from 2018!

When you integrate the entire data lifecycle, you can build richer and more accurate semantic models. These in turn can unlock real AI-powered automation. After all, how can an autonomous marketing agent decide if a campaign change was successful if it can’t retrieve the results of its actions? The Fivetran founders & team share a lot of the same vision I have for the industry. Here’s some of the fundamental values that will make our combined products the definitive solution going forward.

• High quality integrations form the core of our offering. We both care about building tons of great connectors, with trust & reliability above all.
• Every app (AI or SaaS) should be connected to a company’s data. Whether it’s with a simple UI for operators to use or a deep developer platform for engineers to integrate, we both care about universal connectivity.
• A fundamental belief that data teams should be a growth driver, not a ticket closer!

My stick figure rendition of how data teams should operate in the 21st century (from our 2021 Series A announcement). Notice that the data movement arrows go in both directions!

Our Best Work is Ahead

The merger is expected to close soon, and we’re excited to share a lot more about what our teams are going to build. What you can be sure of today and tomorrow is that you’ll see the same faces in Zoom & Slack, the same rapid‑fire releases, and the same obsession with making data useful. There’s still a lot of work to transform data teams into the brains & nervous system of every enterprise. And there’s always more applications to connect, a trend that’s accelerating with AI and agents.

What began as an ambitious idea is now becoming real: a Universal Data Platform that unifies and activates your data—cleanly and instantly. It starts with robust ETL, uses tools like Entity Resolution to create a single source of truth, and syncs enriched data to every system you use—CRMs, ad platforms, support tools, and more. With built-in semantics and governance, AI agents can understand and act on your data. This isn’t a CDP or a dashboard. It’s the infrastructure for turning data into decisions, at scale.

What Changes for Customers?

• Everything you use today keeps working. Census will continue to run as a standalone product and API, with the same SLAs and the same team behind the screen.
• The roadmap gets bigger. Real‑time syncs, deeper governance, an even larger connector catalog—those don’t slow down; they accelerate with Fivetran’s scale. A lot of you have been clamoring for more ingestion capabilities in Census, they’re about to explode 🙂.
• Support that can span your entire data stack. From pipeline health to API limits, you’ll have a single place to get answers.

If you’re reading this far, I’m happy to tell you that for the month of May, you can contact your friendly Census or Fivetran customer representative for a celebratory extended trial of our product!

Thanks is not Enough

As we wrap up this chapter, I want to extend my deepest gratitude to everyone who believed in Census from the very beginning.

To our early customers— Figma, Clearbit, Atrium, Notion—thank you for seeing the promise in our product when it was little more than a scrappy prototype. You helped us validate the very concept of reverse ETL and cement a permanent piece of the data landscape. Today I marvel at the 1000+ companies like Activision, Sonos, Padlet, HubSpot, Weights & Biases, and Shopify that move billions of records through our platform.

To our investors—Andreessen Horowitz, Sequoia, Tiger Global, Insight Partners, and too many angels to name—thank you for believing in our team, our mission, and our market. From the earliest seed cheque to our Series B, you enabled us to dream bigger and execute faster​.

To our partners—Snowflake, Databricks, HubSpot, Braze, and many more—thank you for seeing the potential of what we could achieve together. Working with you demonstrated that when apps & data are well integrated, their value is greater than the sum of its parts.

Finally, to our phenomenal team—past and present—you are the engine behind everything we’ve achieved. Many of you joined in the midst of a pandemic or took a bet on us when we were still defining what the company would become. Your talent, resilience, and dedication have created a company culture I’m proud to be a part of.

Onwards and upwards

While my team will tell you I’m never satisfied, I am incredibly proud of what our team has built. We moved the industry forward not just with groundbreaking technology but by playing our part to elevate the scope of data & ops teams everywhere. I can’t wait to deliver much more inside the Fivetran family.