Should Snowflake acquire Hightouch or Census?
Acquire Census (~$400M), not Hightouch. Census has tighter product-market fit in operational activation (Salesforce, HubSpot), lower integration friction than Hightouch's broader reverse-ETL scope, and gives Snowflake immediate defensibility against Salesforce Data Cloud's activation sprawl.
Moves: (1) Price Census at $350–450M (10–15x ARR multiple for proven reverse-ETL), (2) Keep Hightouch independent; license their stack to Snowflake for enterprise bundling, (3) Launch Snowflake Activation Suite within 18mo combining Census core + warehouse-native sync, (4) Invest in Klue competitive intel to monitor Salesforce acquisitions in the same space.
The Case For Acquiring
- Completes the warehouse loop. Snowflake ingests data, transforms it (dbt), and now activates it back into CRMs/marketing tools without leaving the ecosystem. Eliminates partner tax.
- Defensive against Salesforce Data Cloud. Salesforce is aggressively bundling reverse-ETL. Owning Census or Hightouch neutralizes their workflow entanglement.
- High-margin SaaS add-on. Both Census (~$30M ARR) and Hightouch (~$50M ARR) are pre-profitability but high-gross-margin. Acquisition consolidates data+activation into a single P&L.
- Immediate customer cross-sell. Snowflake's 10,000+ customer base is reverse-ETL's core TAM. Bundling Census reduces their CAC to near-zero.
- $300–700M is asymmetric. For a strategic moat, that's 3–5 weeks of Snowflake operating cash. Low risk relative to upside.
The Case Against
- Partner ecosystem backlash. Hightouch and Census have strong relationships with Fivetran, dbt, Airbyte customers. Acquiring one signals Snowflake will commoditize all of them.
- Integration complexity. Both platforms rely on customer data platforms (Segment, RudderStack, Polytomic) for ingestion. Owning Census doesn't solve the bigger TAM (CDP orchestration).
- Standalone is already working. Census and Hightouch are independent, profitable paths. Acquisition adds integration overhead with limited revenue upside (both are <$60M ARR).
- Talent & culture drain. Acquisition talent retention in fast-growth SaaS is 50–60%. Census and Hightouch would lose their technical founding teams within 2 years.
- Build-vs.-buy optics. Building native reverse-ETL into Snowflake over 18–24mo is faster, cheaper, and avoids regulatory scrutiny (Snowflake already has ~$10B market cap; acquisition adds regulatory risk).
What Snowflake Should Actually Do
- Do not acquire Census or Hightouch. Instead, license their IP (Hightouch's sync engine, Census's operational connectors) for $10–20M/year. Own the integration, not the company.
- Build Snowflake Activation as an in-warehouse reverse-ETL service. Use Force Management's CRO advisory to define the UX (operational simplicity, not platform complexity). Target launch Q3 2026.
- Partner with Polytomic (not Census/Hightouch). Polytomic is newer, cheaper, and more modular. License their connector SDK for Snowflake's Marketplace; revenue-split incentivizes their roadmap to match yours.
- Invest in Pavilion + Bridge Group intel on Salesforce's Data Cloud roadmap. If Salesforce acquires Hightouch, accelerate Snowflake Activation to Q2. If they stay hands-off, maintain library partnerships and ship cautiously.
- Establish a reverse-ETL vendor board. RudderStack, Polytomic, Hightouch, Census, and Workato all address different TAM slices. Monitor Klue signals on who's winning operationally, and align Snowflake's roadmap with the winner, not the loser.
- Run a $50M acquisition bake-off with all five vendors (RudderStack, Polytomic, Hightouch, Census, Workato). Require 2-year integration commitments; the winner gets preferential Marketplace placement + bundling rights. Keeps competitive pressure on all.
- Ship Snowflake Activation to a closed beta (500 customers) by Q3 2026. Measure NPS, CAC payback, and activation velocity. If > Hightouch/Census cohort averages, full launch; otherwise, acquire the best performer.
- Guard against Salesforce's next move. Salesforce will likely bundle Einstein Activation + Data Cloud + native reverse-ETL within 18mo. Snowflake's only advantage is warehouse-native speed and dbt integration; lead with that, not M&A.
Comparison Table
| Path | Cost | Timeline | Risk | Outcome |
|---|---|---|---|---|
| Acquire Census | $350–450M | 6–9 mo integration | Partner backlash; talent drain | Immediate activation; 15–20% Snowflake revenue lift |
| License Hightouch IP | $10–20M/yr | 3–6 mo licensing | Dependency on vendor roadmap | Lower margin; Hightouch retains independence |
| Build in-warehouse (Snowflake Activation) | $20–30M (eng + ops) | 18–24 mo | R&D delay vs. Salesforce; integration debt | Warehouse-native; defensible moat; high margin |
| Partner board (5 vendors, bake-off) | $5–10M (incentives) | 12 mo evaluation | Diluted focus; scattered roadmap | Best-of-breed selection; optionality |
| Acquire Polytomic (wild card) | $200–300M | 12–18 mo | Lowest cultural fit; weakest TAM | Modular connectors; future-proofs reverse-ETL |
The Mermaid View
Bottom Line
Don't buy Census or Hightouch. License, partner, and build. Snowflake's strength is warehouse-native speed and dbt integration—not M&A.
Salesforce Data Cloud is the threat, but the best defense is a 18-month in-house Activation suite, backed by a vendor board (Polytomic leading) and Pavilion/Bridge Group competitive intel. If Salesforce makes a move, accelerate; otherwise, ship cautiously and keep the ecosystem partners profitable and independent.
Snowflake's margin advantage dissolves the moment it looks like a monolith.
Tags
["snowflake", "reverse-etl", "m&a-strategy", "salesforce-data-cloud", "data-activation", "census", "hightouch", "polytomic", "warehouse-moat", "cro-lens"]
FAQ
Does the article recommend acquiring Census or Hightouch? The Direct Answer opens by favoring Census (~$400M) over Hightouch for its tighter product-market fit in operational activation, but the deeper recommendation in "What Snowflake Should Actually Do" is to acquire neither and instead license, partner, and build.
The Bottom Line confirms: don't buy Census or Hightouch. The article suggests licensing their IP for $10-20M/year and owning the integration rather than the company.
What are the ARR figures cited for Census and Hightouch? Census is cited at roughly $30M ARR and Hightouch at roughly $50M ARR, both described as pre-profitability but high-gross-margin. The article argues that since both are under $60M ARR, acquisition adds integration overhead with limited revenue upside.
A Census acquisition is priced at $350-450M, a 10-15x ARR multiple for proven reverse-ETL.
Why does the article suggest partnering with Polytomic instead? Polytomic is described as newer, cheaper, and more modular than Census or Hightouch. The recommendation is to license its connector SDK for Snowflake's Marketplace with a revenue split that incentivizes Polytomic's roadmap to match Snowflake's.
The build-it-in-warehouse path ("Snowflake Activation") would launch Q3 2026 using Force Management's CRO advisory to define the UX.
What is the $50M acquisition bake-off the article proposes? It suggests running a bake-off across five vendors (RudderStack, Polytomic, Hightouch, Census, and Workato) with 2-year integration commitments, where the winner gets preferential Marketplace placement and bundling rights.
This keeps competitive pressure on all of them while preserving optionality. Klue signals would track who is winning operationally.
Why does the article warn about the Salesforce competitive threat? It predicts Salesforce will likely bundle Einstein Activation plus Data Cloud plus native reverse-ETL within 18 months. The article says Snowflake's only real advantage is warehouse-native speed and dbt integration, and it should lead with that rather than M&A.
If Salesforce acquires Hightouch, the recommendation is to accelerate Snowflake Activation to Q2.