Fivetran Activations (formerly Census)

Overview

Census no longer exists as an independent product, and anyone comparing it today should know that first. Fivetran announced its acquisition in May 2025, getcensus.com now redirects to fivetran.com, and the product is Fivetran Activations.

The migration is well past announcement stage. Month-to-month Census plans ended in March 2026, the Census store was retired in August 2026, and Fivetran has said it will eventually deprecate the Census API.

As Activations, it is described as cloud-based managed reverse ETL configured without writing code, building audiences in your warehouse so there is no siloed, ungoverned copy of customer data.

Its sync engine works differently from most. Rather than reacting to events, it keeps source and destination in sync continuously, an approach Fivetran compares to cloud file storage.

What makes it distinct in this category is the bundle. It is the only reverse ETL product sold alongside a market-leading ingestion business and, since the June 2026 dbt Labs merger, first-party transformation, on one meter and one bill.

That meter is monthly active rows, the same unit as Fivetran's ingestion, which makes the round trip easy to model but means activation rows cost roughly ten times ingestion rows in Fivetran's own examples.

Modelling is SQL over warehouse tables, with a dbt integration for triggering syncs and an audience hub for building segments without SQL.

AI columns are a genuine addition, generating derived fields with large language models as part of the sync rather than as a separate enrichment step.

Its compliance list is the strongest of the three, covering SOC 1 and 2, ISO 27001, HIPAA, PCI DSS Level 1 and HITRUST, with explicit control over EU data processing location.

Two gaps to note against Hightouch: identity resolution is not published as an Activations feature, and no event collection SDKs exist, so it activates data rather than gathering it.

Key features and capabilities

The same headings are used for every reverse etl & activation entry, so two tools can be read side by side.

Where data comes from

• Twenty named sources including Snowflake, BigQuery, Databricks, Redshift, Azure Synapse and Microsoft Fabric
• Also Athena, ClickHouse, MotherDuck, Trino, Elasticsearch, Postgres, MySQL, SQL Server and S3
• Simple sources are supported too, including Google Sheets and CSV files
• Google Pub/Sub can act directly as a source for real-time flows
• It does not collect first-party event data through SDKs, unlike the other two

Business tools it syncs to

• More than 200 pre-built reverse ETL destinations on the marketing page
• The documentation publishes no destination count, listing them individually instead
• Only two published categories, available destinations and custom or partner destinations
• No support tiers such as generally available or beta are published for Activations destinations
• Connector availability changes between plans, so check the mapping when migrating from Census

Defining audiences and models

• SQL models over one or more warehouse tables, possibly across different schemas
• A dbt integration triggers syncs after models run, and dbt is now a sibling product
• An audience hub enables self-service audience creation without SQL
• AI columns generate derived fields with large language models as part of the sync
• Identity resolution is not published as an Activations feature, a real gap against Hightouch

Sync behaviour and scheduling

• Schedules from weekly down to continuous, which checks the source roughly every minute
• Standard cron expressions to minute granularity, or triggering from an API or orchestrator
• All syncs are incremental, sending only records new or changed since the previous run
• Six sync behaviours, upsert, update only, create only, mirror, append only and delete
• Live syncs are marketed at very low latency, but the documentation page for them is missing, so treat with care

Security, consent and observability

• The strongest published list of the three, SOC 1 and 2, ISO 27001, HIPAA, PCI DSS Level 1 and HITRUST
• Data, keys and credentials encrypted in transit and at rest, with customer-managed keys
• Column blocking stops a column replicating, and hashing anonymises it before it lands
• Staff can only access your data with your approval, which you can revoke at any time
• Private connectivity is Business Critical only; consent management is not published

Beyond syncing

• Sold alongside more than 700 ingestion connectors, so the round trip runs on one bill and one meter
• dbt transformations became first-party with the June 2026 merger
• Stewardship of the Great Expectations open-source project since May 2026, for data quality
• No event collection, personalisation, experimentation or decisioning engine is published
• The narrowest set of activation extras of the three, in exchange for the widest data platform around it

Where it runs and what it costs

• Cloud-managed, with a hybrid deployment option for stricter security requirements
• No fully self-hosted or open-source option
• Audiences are built in your own warehouse, and syncs read from and write back to it
• EU data residency is explicitly committed, with control over the processing location
• Billed on monthly active rows, the same unit as Fivetran ingestion

Pricing

Usage-basedFree to 3,500 activation rows

The only one of the three publishing real numbers. Billed on monthly active rows, with a free tier of 3,500 activation rows alongside 500,000 ingestion rows and 5,000 model runs, and a 14-day trial per new connection. Fivetran's own worked example prices a Snowflake to Salesforce sync of 12,007 rows at $202.25 a month, so activation rows cost roughly ten times ingestion rows. Tiers run Free, Standard with 15-minute syncs, Enterprise with one-minute syncs and Business Critical with private connectivity. Annual contracts save up to 22%.

Vendor pricing page →

Demos and videos

• How to set up Activations Tutorial · fivetran.com
• Fivetran Activations guided demo Product demo · fivetran.com

About Fivetran

Census was founded in 2018 in San Francisco with Boris Jabes as co-founder and chief executive, and had over 50 staff and hundreds of customers when Fivetran agreed to acquire it in May 2025 for undisclosed terms; its own funding history is no longer published anywhere first-party, since its blog now redirects. It is now part of Fivetran, itself founded in 2012 by George Fraser and Taylor Brown in Oakland, which completed an all-stock merger with dbt Labs on 1 June 2026. The combined company claims more than 100,000 data teams. Named Activations customers include HubSpot, Sonos, Skims and Grubhub.

Founded 2018 · Oakland, California · fivetran.com