Census is one of the two companies that invented the reverse ETL category. Alongside Hightouch, Census defined the playbook in 2019: take a SQL query against the warehouse, sync the result to a SaaS destination on a schedule, handle all the boring operational details (field mapping, upserts, retries, observability) so the data team doesn't have to. If you have ever pushed a customer health score from Snowflake to Salesforce in 2020-2026, there's a meaningful chance Census or Hightouch was the wire that did it.

What It Actually Is

Census is a managed pipeline product. You connect it to your warehouse (Snowflake, BigQuery, Databricks, Redshift, Postgres) and to one or more SaaS destinations. You define a "sync" by writing a SQL query or pointing at a dbt model; the rows that come back are the rows Census will push to the destination. You map the columns to destination fields once. Census then handles:

• Idempotent upserts: distinguishing new records from updates so you don't create duplicates in Salesforce.
• Change detection: only syncing rows that actually changed since the last run, instead of wastefully resyncing everything.
• Rate limiting and retries: respecting Salesforce API limits, backing off gracefully, retrying transient failures.
• Observability: a UI showing every sync run, every failed row, and every field-level error.

The set of supported destinations is large — Salesforce, HubSpot, Marketo, Pardot, Iterable, Braze, Zendesk, Intercom, Google Ads, Facebook Ads, Slack, Snowflake, and many more. The set of supported sources is essentially "any major data warehouse."

The Origin Story

Census was founded in 2018 by Boris Jabes (formerly CEO of Meldium, acquired by LogMeIn) and Bastin Gerald. The founding observation was that the modern data team, by 2018, had spent years investing in cloud warehouses, dbt, and analytics engineering — and the result of all that work was usually trapped in the warehouse. The customer health score that the data team built was sitting in a Snowflake table that the customer success rep, who lived in Salesforce, had no way to see. The audience the analytics team built was sitting in a dbt model that the marketer, who lived in Iterable, couldn't access.

Census started as a Salesforce-to-warehouse sync tool, then quickly pivoted to the inverse direction once the founders realized that pushing out of the warehouse was the actually-painful problem nobody had solved. The launch in 2019 was approximately simultaneous with Hightouch's, and the two companies have been the dominant pair in the category ever since.

Census raised significant venture funding from Sequoia, Andreessen Horowitz, and Insight Partners. In 2024 it acquired Sutro, a customer data infrastructure startup, expanding its activation surface area.

Where Census Differs from Hightouch

Census and Hightouch do approximately the same thing, and any comparison between them requires acknowledging that they are 80% identical products. The 20% that differs:

Census leans more analytics-engineering-flavored. Census has historically marketed itself as "the operational analytics platform for data teams," with strong dbt integration, SQL-first workflows, and a sensibility that feels native to the people who already live in dbt and Snowflake. The Census brand is built for the analytics engineer.

Hightouch leans more marketing-flavored. Hightouch has aggressively repositioned itself as "the warehouse-native CDP" and has invested heavily in audience builders, marketing campaign tooling, and identity resolution — the things that matter to a CMO buying a customer data platform. The Hightouch brand is built for the growth marketer.

Both products will sync your dbt model to Salesforce. The marketing and the surrounding feature set differ.

The Opinionated Take

In a duopoly category, the smaller player has to pick a defensible position, and Census's choice has been stay close to the data team. That's the bet: as the warehouse becomes the center of the operational stack, the data team becomes the buyer for activation tooling, and the product that feels native to a dbt-fluent analytics engineer wins. It's a coherent strategy and it's working. Census's customer list — Canva, Figma, Notion, Loom — is heavily weighted toward technically sophisticated companies where the data team has strong organizational power.

The risk in this strategy is that the data team is the smaller buyer compared to the marketing org, and the bigger commercial prize lives in the marketing department. Hightouch chased that prize harder. Whether that bet pays off long-term is one of the more interesting open questions in the data stack.

The other thing worth saying: the existence of both Hightouch and Census proves that "warehouse-native activation" is a real and durable category, not a feature waiting to be absorbed. The warehouses themselves have shipped competing features and neither of these companies has been killed by it. The operational details matter, and the warehouses, so far, haven't done them well enough.

Where Census Fits in the Stack

Census sits between the warehouse and the operational SaaS apps. It reads from the warehouse via SQL, and it writes to SaaS apps via their APIs. It is not an ingestion tool (that's Fivetran/Airbyte), not a transformation tool (that's dbt), and not a BI tool. It is the activation layer.