Seeking Feedback on Data Enrichment Using Clay for ICP Filtering

Mark K. · 2026-01-14T21:45:24.926Z

Hey everyone! My name is Mark, I am an intern working in growth at a healthtech start-up. I have been using clay to web scrap and enrich data on our enterprise customer base to filter for ICPs. I would love to get some feedback on the enrichment I built as I am quite new to clay and I wonder if there is ways to improve it. If anyone knows the best way to contact someone that can help at clay or otherwise I'd love to know! Thanks

Seeking Feedback on Data Enrichment Using Clay for ICP Filtering | Clay

3 comments

· Sorted by Oldest

Honourable A.
·
·
what enrichment did you work on that you need feedback for - you can share screenshot or recordings btw you can ask questions in this channel if you need help figuring things out or reach out to the 🔒[private feed] if it has to do with clay itself And you can also join a clay cohort if you want to get the basic right
Mark K.
·
·
So I have a list of health systems in the US and I am basically trying to gather data around their scale, financials and their current tech usage. I have been using an AI enrichment to scrap for this but the credits can be quite costly and I find the more I include in my search prompt the lower the accuracy. Its not that I have a technical issue its more so I would just like to see what an expert thinks of my current process.
Honourable A.
·
·
yea it a common issue getting lower accuracy when you tend to over complicate your prompt also i dont know what exact informations you are looking for when it comes to the scale, financials and tech usage so hard to give you something specific for the tech usage you could use built with to get the tech stack or is that not what you are looking for? as for the others, AI is surely your best bet in getting the info - scale and financials. to reduce your credit cost it is best you use your own API keys - saves so much cost and you dont use your clay credit also to get good accuracy on your prompt i would advice starting with a small batch to see what results you get in those small batches. and you can use AI to help write your prompts so you are getting the right data or close to the right data. there are some open AI GPTs that could help you write your prompts and keep testing until you nail it so for example if you are testing for like 20 rows and have a prompt to get the info you need - if like 13 out of 20 is right then i feel you are on the right track happy to look at your actual prompt if you share it - can probably help you refine it further

Honourable A.
·
·
what enrichment did you work on that you need feedback for - you can share screenshot or recordings btw you can ask questions in this channel if you need help figuring things out or reach out to the 🔒[private feed] if it has to do with clay itself And you can also join a clay cohort if you want to get the basic right
Mark K.
·
·
So I have a list of health systems in the US and I am basically trying to gather data around their scale, financials and their current tech usage. I have been using an AI enrichment to scrap for this but the credits can be quite costly and I find the more I include in my search prompt the lower the accuracy. Its not that I have a technical issue its more so I would just like to see what an expert thinks of my current process.
Honourable A.
·
·
yea it a common issue getting lower accuracy when you tend to over complicate your prompt also i dont know what exact informations you are looking for when it comes to the scale, financials and tech usage so hard to give you something specific for the tech usage you could use built with to get the tech stack or is that not what you are looking for? as for the others, AI is surely your best bet in getting the info - scale and financials. to reduce your credit cost it is best you use your own API keys - saves so much cost and you dont use your clay credit also to get good accuracy on your prompt i would advice starting with a small batch to see what results you get in those small batches. and you can use AI to help write your prompts so you are getting the right data or close to the right data. there are some open AI GPTs that could help you write your prompts and keep testing until you nail it so for example if you are testing for like 20 rows and have a prompt to get the info you need - if like 13 out of 20 is right then i feel you are on the right track happy to look at your actual prompt if you share it - can probably help you refine it further