Fivetran — the company helping enterprises move data from various sources into cloud databases — announced Thursday it has acquired Census, a reverse extract-transform-load (ETL) platform that lets businesses push data from databases back into operational tools. Census was founded in 2018 and raised more than $80 million in venture funding from some of the biggest names in the business — Sequoia, Andreessen Horowitz, and Tiger Global among them — making it one of the better-funded reverse-ETL startups in the category.
Terms were not disclosed, but Census carried a $630 million valuation in 2022. The entire Census team will migrate to Fivetran, and the Census brand will eventually fold into the Fivetran platform.
Why Buy Instead of Build?
Fivetran co-founder and CEO George Fraser told TechCrunch the deal matters for several reasons. Customers had been asking Fivetran for reverse-ETL capabilities for years. The company considered building its own — and even prototyped one — but Fraser said Fivetran realized partnering with a company that had already solved the problem was the better use of resources.
From a technical perspective, if you look at the code behind these services, they are actually very different. To do this, you have to solve a completely different set of problems.
Once acquisition became the sensible route, Census was the natural target: the two companies shared many customers, and the platforms’ philosophies aligned. “People who like Fivetran are also going to like Census, versus Informatica or building their own connectors,” Fraser said. “The philosophical choices of the two products are broadly similar, so they tend to attract the same customers — which matters a lot when you think about synergies.”
A Decade in the Making
The founding teams go way back — and that history turned out to be an advantage. Fraser met the Census team, including CEO Boris Jabes and co-founder Anton Vaynshtok, during Y Combinator’s Winter 2013 batch. Fraser and his co-founder Taylor Brown were going through YC themselves, while Jabes and Vaynshtok were building Meldium, a password and account-management system acquired by LogMeIn in 2014. They stayed in touch, and even discussed the concept that became Census years before the company launched.
Before Census’s founders started the company, we talked to them about their idea. I joked with Taylor at the time that it might eventually be acquired, because there was so much synergy between the two things. In a way, it feels like it was destined.
The reverse-ETL category’s rise also tells a larger story about how the data industry matured. Five years ago, the dominant paradigm was one-way: extract data from everywhere, load it into Snowflake or BigQuery, and let analysts query it. The tools that won that era — Fivetran, Airbyte, Matillion — were all built on the same assumption: data flows toward the warehouse and stays there. But the rise of AI agents and real-time personalization broke the assumption. An agent that recommends products, a churn model that triggers a retention email, a support copilot that reads the customer’s latest usage — all of these need warehouse-computed insights delivered back into operational systems within minutes, not after the next batch job. Reverse ETL stopped being a nice-to-have the moment the warehouse stopped being the end of the pipeline and became the middle of it.
The deal’s timing also reflects a broader consolidation wave in data infrastructure. As AI workloads drive up the strategic value of clean, well-routed data, the companies that control the pipes are racing to become one-stop shops — dbt Labs adding transformation, Snowflake adding application runtime, and now Fivetran adding the last mile. For Census, the exit logic is equally clear: reverse ETL as a standalone category was always at risk of being absorbed by the forward-ETL platforms upstream, and a $630-million-valued exit to the category leader — with the founding team intact — beats a slow squeeze from both sides. For customers, the near-term promise is fewer vendors and native integration between warehouse and operational tools; the longer-term question is whether the combined platform can add the AI-serving layer that the next wave of enterprise buyers will expect as standard.
What Reverse ETL Means for the Data Stack
The deal fills the last gap in the modern data stack’s round trip. Forward ETL — Fivetran’s core product — moves data from operational sources (SaaS apps, databases, event streams) into a central warehouse where analysts can query it. Reverse ETL closes the loop by syncing warehouse data back into the tools that operations teams use daily: CRM systems, marketing-automation platforms, support desks. Without it, the warehouse is a read-only archive; with it, the warehouse becomes the operational brain of the business, pushing computed customer segments, churn scores, and product-usage signals back into the tools where actions actually happen.
For Fivetran, the timing reflects a market shift. The data-pipeline category consolidated rapidly as enterprises reduced vendor counts, and the buyers demanding reverse ETL were increasingly the same teams deploying AI workloads that need clean operational data fed back into models and agents. Owning both directions of the pipeline — and the largest existing customer overlap in the category — positions Fivetran as the default full-cycle data-movement platform at precisely the moment AI agents need reliable, low-latency access to operational data. The YC friendship that made the deal smooth is a nice story; the data-stack logic that made it inevitable is the real headline — and a reminder that in infrastructure, the best acquisitions are the ones both sides saw coming for a decade.
