Clay data enrichment is a spreadsheet-style workflow that takes a list of people or companies and runs each row through a chain of data providers in sequence, a waterfall, stopping at the first provider that returns a result. It finds an email for 80 to 92% of rows, the highest find rate of any finder approach, at an accuracy of 70 to 80%, because the first answer wins even when a later provider would have been right.
Clay is a table. Each row is a person or company, each column is an enrichment: find the LinkedIn profile, find the work email, find the phone, pull the tech stack, summarise the website with an AI prompt. The email column is where most of the value and most of the risk sit, because Clay does not have its own email database. It calls other providers, in an order you choose, and takes the first result.
That design is why Clay finds more addresses than any single finder, and why a Clay export bounces more than its confidence scores suggest. The provider that answered first is not necessarily the one with the freshest record. A full review of the product is in the Clay review; this page is about the enrichment mechanics and what to add.
Clay needs at least a full name and a domain. Rows missing either go through a company-lookup step first, which costs extra credits.
A database such as Apollo, Prospeo or People Data Labs is queried. If it has a record for that name at that domain, the waterfall stops here and returns it, with whatever age the record carries.
If the database misses, Hunter, FindyMail, Datagma or similar infer the address from the domain pattern. Each provider called costs credits; the waterfall stops at the first hit.
Clay marks the returned address as validated when the provider reported it as such or when a syntax and MX check passed. This is not a live mailbox check at export time.
The row records which provider answered. Few teams look at that column, but it is the first thing to check when a segment bounces.
Four situations where the waterfall returns something with confidence and that something bounces. The invalid and catch-all rates are from the BounceZero corpus, April to October 2026.
| Situation | What the waterfall does | Typical outcome | Fix |
|---|---|---|---|
| Database record is stale | Returns the old address because the database answered first | Bounces; B2B addresses decay fast, with only 19.2% of addresses still valid 90 days after verification in the BounceZero decay study | Verify at export; re-order waterfall so a pattern finder with live check runs first for small companies |
| Catch-all domain | Pattern finder guesses a format; the server accepts anything; Clay marks it found | Looks valid, cannot be confirmed; 2.3% of non-freemail (business and ISP) addresses, 12.4% of .org | A verifier that labels catch-all separately; route those rows out of cold volume |
| Mixed address formats at one company | First provider applies the common pattern to a person who uses a different one | Hard bounce | Verify; on a miss, run a second pass with the next format |
| Name variants and non-Latin names | Providers normalise differently; the first match may be a different person | Wrong person, or invalid | Keep the LinkedIn URL column and spot-check high-value rows |
Clay prices in credits and each provider in the waterfall consumes credits when called. These are planning figures from the published credit tables and are vendor-reported; your mix of providers changes them.
| Step | Credits per row (approx.) | On 5,000 rows | Note |
|---|---|---|---|
| Find work email, waterfall of 3 providers | 2 to 6 | 10,000 to 30,000 | Only providers actually called are charged |
| Find LinkedIn profile | 1 to 2 | 5,000 to 10,000 | Skipped if URL supplied |
| Company enrichment (size, industry, tech) | 1 to 3 | 5,000 to 15,000 | Often cached across rows for the same domain |
| AI column (summarise website, draft opener) | 1 to 4 | 5,000 to 20,000 | Depends on model and prompt length |
| External verification (BounceZero via HTTP column) | 1 BounceZero credit, $0.003 | $15 at Growth tier | Unknowns refunded |
The verification step is the cheapest column in the table and the only one that reports what will bounce.
Two ways to put a live mailbox check into the Clay table so every exported row carries valid, invalid, catch-all or unknown.
Add an HTTP API enrichment that calls the BounceZero single-validation endpoint with the email column as input. Map the response status to a new column. Rows run in parallel and the result lands next to the address. Documentation in API email validation.
Export the table to CSV, run it as a bulk job, and import the result column back by matching on email. Simpler for one-off lists above 50,000 rows and keeps Clay credits untouched. Details in verifying Clay exports.
Order the waterfall by freshness, not by hit rate. For companies under 200 people put a pattern finder with a live check first; for enterprises put the database first, since large companies have stable formats and database records age more slowly.
Verify at export, every time, and filter the sequencer import to valid only. Send catch-all rows from a secondary domain at low volume or move them to LinkedIn. Hold unknowns for a seven-day re-check. Keep the provider column and, when a segment bounces, look at which provider answered; that is usually the fix.
Re-run verification on anything older than ninety days before it goes back into a sequence. The waterfall is excellent at finding; it was never designed to confirm.
Pricing, strengths, where it fits
Database-first against waterfall-first
When a single finder is enough
Step-by-step export and re-import
How every finder mechanism works
The wider enrichment market
Clay marks an address as validated when the provider that found it reported it as valid or when a syntax and MX check passed. That is a check at indexing time, not a live mailbox check at export. For a list you are about to send to, run a verifier at export so each row carries valid, invalid, catch-all or unknown.
Three reasons. The database provider that answered first held an old record; the domain is catch-all so any guess looked right; or the person uses a different format from the company default. Verification at export catches the first two and flags the third as invalid for a second pass.
Clay finds an address for 80 to 92% of rows, the highest of any finder approach, at an accuracy of 70 to 80%. The gap between find rate and accuracy is the share that bounces, which is why the verification column matters more in Clay than in a single-provider finder.
For small and mid-sized companies, put a pattern finder with a live check first, then databases. For enterprises, put the database first because large companies keep stable address formats. Whichever order you choose, verify at export; the order changes the hit rate, not the need to confirm.
One BounceZero credit per address. A 5,000-row table costs 5,000 credits, which is the Growth tier at $15, and unknown results return their credits. That is usually the cheapest column in the table and the only one that predicts bounces.
Call the API from an HTTP column or verify the export in bulk. 100 free checks a month, unknowns refunded.
Ayoub built BounceZero's 5-stage validation pipeline, its dedicated BGP-announced IP infrastructure, and the Patroni HA PostgreSQL cluster behind every verification. Previously built high-volume email delivery infrastructure. Trained at 1337 Benguerir (École 42 network, 2019). Open-source: bgp_analyzer.
Where leads come from, how finders work, and what a verified lead is - with 2026 corpus data
Email finder for companies
Free email lookup in 2026
Apollo lead generation guide 2026
8 channels that build pipeline, by budget
Tools by stage with data accuracy notes
Sellers compared, with sample-check rules
Continue through related topics