Most Clay failures are planning failures that surface as technical symptoms. Answer these six questions before building anything.
1. What decision does each column feed?
Every enrichment column costs Actions and possibly credits on every row, every run. Start from the decision the data supports, work backwards to required fields, and refuse speculative columns.
2. What is the source of truth?
For each attribute (industry, headcount, email...), decide which system wins when sources disagree: Clay, the CRM, or a named provider. Write it down; undocumented precedence rules are the root of most CRM data drift.
3. What is the dedupe key?
Agree the stable identifier (website domain for accounts; email for people) before any write-back. Enriching and writing without a dedupe key manufactures duplicates.
4. What are the CRM write rules?
Which objects, which fields, create vs update, who approves. Write only validated records. Include provenance fields (source, timestamp, workflow) on every write so future drift is diagnosable.
5. What does success look like?
Measurable: match rate ≥ X%, email validity ≥ Y%, cost per usable contact ≤ $Z. Without thresholds, "working" is an opinion.
6. Who owns it?
A named owner for the workspace and its CRM writes — accountability before automation.
Need help applying this?
Describe your project and the ClayForEnterprises team will review it — the resources stay free either way.