I asked some Clay experts I found thru Linkedin for their best tips on not wasting Clay credits, and compiled the most popular answers.
Test in Sandbox Mode First - Test workflows on max 50 rows before running on full dataset; prevents accidental bulk enrichments
2. Reorder Enrichment Waterfall - Put cheapest providers first (e.g., Clay enrichments → PeopleDataLabs → others); can save 3x credits 3. Start with Companies, Not People - Filter companies by ICP first, then enrich contacts; avoids paying for irrelevant contacts 4. Write better structured prompts. Use structure: Input → Action → Output → Prefix → Example; use JSON schema output for clean data 5. Optimize OpenAI Costs - Use GPT-4o-mini for research ($0.001/call) then GPT-4o for final output; can cut $15k spend to $300 6. Set “Only Run If…” Conditions - Add conditional logic to enrichments; only run when specific fields are blank/meet criteria 7. Know Your Strategy First - Write down desired outcome before building; automate intentionally, not everything More here: https://bloomberry.com/blog/how-to-stop-wasting-clay-credits-with-these-7-tips/
Really cool! :)
This post will remain pinned to the channel for one full week!
I’m curious on the groups thoughts on number two. I am finding that when I sort based on lower credit cost providers, while I might get a cell phone result earlier in the waterfall, the number is for the wrong person. I am using the filters to only return results if the cell phone number is indeed a cell phone and confirmed valid. To respond to this, I have begun tracking the successful data provider and then let my team flag the contact mobile number as invalid in the CRM, which pushes to a segment that I can then go in and check the next data provider. I’d love to hear if there’s a better workflow because it’s very manual.
