We’re cancelling Clay by November 1.

The reason is not the tool. It is what we changed about the way we handle data.

Outbound Single Source of Truth, the diagram from the original post

Data is the core of outbound and the main asset of GTM.

It is common to think copy is the key factor. That the more «correct» and «beautiful» the words, the higher the reply rates.

You can write a dozen variants and all of them will work poorly if there is no foundation of data underneath. And that foundation is signals: a specific fact about a company or a person, with a date.

Then the question comes up: where and how to store all of it. In a CRM? In outbound tools? In Clay? In a spreadsheet?

We tried every option, and each one is inconvenient in its own way.

You do not want to overload a CRM: you hit limits and inflate the bill.

Outbound tools keep only what you pushed into them, meaning the already validated list. Everything that got filtered out lives somewhere else. And a CRM on top of that.

Clay was convenient at small volumes, but with limits on API calls it stopped solving data orchestration efficiently.

So we deployed a separate PostgreSQL database, ours runs on Supabase. That became our Outbound Single Source of Truth.

What it gave us:

— talking to the base directly, in queries

— speed went up noticeably, limits on data volume disappeared

— a playbook deploys with a single function, nothing to build by hand

— any interface plugs in on top: the same CRM or BI for visualisation, we use open-source Metabase

— the base can be driven by an AI client over MCP: analysis, orchestration, building

— the data sits with us and moves between tools without losses

Right now we are running several GTM projects on this architecture, and we plan to cancel Clay by November 1.

More to share as we go.