Where to start: the ICP as data
Sources and enrichment
Wire sources that return the fields your ICP needs: Sales Navigator for the fit envelope, registries for firmographics, job boards for hiring signals. Merge on domain. Then run enrichment as a cost-ordered waterfall: free sources first, credits only on the hard finds, a validation gate at the end. This is where most Clay bills are won or wasted.
Scoring and sync
Keep scoring simple. Two or more live intent signals on a fit account is an A, one is a B, fit without motion is a C. Complexity in scoring formulas is where signal programs die. Sync scores to your CRM with dedupe on domain, so sales works one clean list and Tier A accounts trigger outreach within 48 hours.
Workflows as code, from day one
Since Clay released its API and CLI, prototype in the UI but export workflows through the CLI and commit them. Snapshots make rollback a command, diffs make changes reviewable, and documentation makes the system survive whoever built it. The full method is in the GTM as code playbook. Think in reusable functions, because most tables are just combinations of them.
Questions people also ask
How do you set up Clay properly?
Start from the ICP as data fields, wire sources that return those fields, run a cost-ordered enrichment waterfall with validation, score by signals and sync to the CRM.
How do you avoid burning Clay credits?
Order enrichment cheapest first so credits only hit hard-to-find records, and enrich at the moment of need.
Should Clay workflows be version controlled?
Yes. Since Clay shipped its API and CLI, workflows can live in a repo, versioned and reviewed.