What it costs to run a list through Clay

Ten guides will show you how to build a table in Clay. Almost none of them tells you what the table costs to run, and the ones that mention price answer it in a single heading at the bottom of a tutorial.

Here is the number, computed from Clay's own published rates on 3 September 2026. A 500-row list, enriched with a work email and one AI research column, costs about 725 data credits and 1,000 actions. That is 29% of a month on the $185 Launch plan, so roughly $0.11 a row if you use the whole allowance and $0.37 if you run one list and stop. It cannot be done on the free plan at all.

Tamozo sells lead context, so we have a commercial interest in how you answer this question. Every figure below comes from Clay's own pages, dated, so you can check them.

The interesting part is not the total. It is where the total comes from.

Two meters, and only one of them will ever run out

Clay charges you on two axes and they are easy to confuse.

Actions are Clay's own orchestration work. Clay's docs describe them as the orchestration you do in Clay, covering enriching data, running AI research and sending data to other tools, and put the cost at "a few tenths of a penny" each. One action is charged per enrichment, per AI use, per CRM sync, per API call. Unused actions expire at the end of the billing cycle.

Data credits buy third-party data and AI from Clay's marketplace. Clay puts these at "a few pennies" each. They roll over, capped at twice your monthly limit.

For anyone below enterprise volume, actions do not matter. Clay says so itself, in a pricing memo it published on 11 March 2026: "We've intentionally set Action limits generously, and expect 90% of customers will never hit them." The arithmetic below agrees. At 500 rows with two columns you use 6.7% of Launch's actions and 29% of its credits.

So when a guide presents Clay's pricing as a two-meter puzzle, it is technically correct and practically misleading. There is one meter that decides your bill.

A 500-row list, priced from Clay's own numbers

Every input here comes from Clay's documentation.

Clay's docs carry a worked example of enriching 100 contacts. Finding people in Clay's own database costs nothing. A LinkedIn profile enrichment costs 0.5 data credits. A work email costs 0.5 data credits, charged only when the lookup succeeds. Validation is free on selected providers. Clay's illustration assumes a 90% email hit rate, which is an assumption inside an example rather than a published match rate, so treat it as Clay's own working figure and not as a benchmark.

An AI research column is priced separately, and this is where it gets expensive. Clay charges one action plus data credits per run, and the credits depend on the model. Clay publishes two prices per model, one for content generation and one for web research, and a research column is the second of those. On the web research column, only six of the sixteen models Clay lists carry a fixed rate at all:

Take Clay Helium, Clay's own mid-priced model, and run 500 rows with an email lookup and one research column:

Against the tiers. Clay publishes two prices for each plan, and both are real: the higher one is month-to-month, the lower is what the same plan costs per month if you pay for a year up front. Clay's pricing page opens with the annual figures showing, which is why the two numbers look like a contradiction if you read the cards and the page text in the same sitting.

The saving is a flat 10%. Every figure in the rest of this article uses the month-to-month price, because that is what you pay if you are trying Clay for a quarter rather than committing to a year.

The research column is 69% of your bill

Look again at the split. The email lookup on 500 rows costs 225 credits. The single research column costs 500. The AI column is more than twice the cost of the thing most people think they are buying. Swap Helium for Argon and the research leg becomes 87% of the run.

One number from Clay's own docs belongs next to that arithmetic, because it is the honest counterweight to it. Clay says a fully enriched record typically costs 6 to 20 data credits. Our 1.45 a row is email plus one research column, which is a deliberately thin build. Add phone numbers, firmographics and a second research pass and you are in Clay's range, where 500 rows costs 3,000 to 10,000 credits and the Launch allowance is gone in a single list.

This matters because of what people picture when they picture Clay for lead generation. They picture finding email addresses. If that is genuinely all you want on 500 rows, you spend 225 credits, which is 9% of the Launch allowance, and you are carrying a $185 monthly subscription to do it.

Clay earns its price when the research column is doing work you would otherwise do by hand: reading a company's site, classifying a persona, checking whether a signal is real. Priced per row, that judgment is the product. The email is close to a rounding error.

The free plan tops out at 68 rows a month

Clay's free plan gives you 100 data credits a month and caps tables at 200 rows.

Run the arithmetic on that. Email plus one Helium research column costs 1.45 credits per row. At 100 credits a month, the free plan's ceiling is 68 rows. And the 200-row table cap stops a 500-row list before credits enter into it.

Sixty-eight rows a month is a demo, not an allowance. That is worth knowing before you plan an evaluation around the free tier, because the usual sequence people imagine, try it free on a few hundred rows and upgrade if it works, is not available. Clay's 14-day trial with 1,000 data credits is the real evaluation surface, and phone enrichments are excluded from it.

Growth costs more per credit than Launch

Divide the price by the included credits and you get the implied cost per data credit, which is more useful than "a few pennies":

Moving up costs you more per credit, not less. You buy the higher tier for CRM integrations, the HTTP API, intent signals and action headroom. You do not buy it for a data discount on the included allowance. Buying credits on top, through the expansion options Clay publishes on both paid tiers, is cheaper per credit than the allowance that comes with the plan.

What a Claygent run costs, and why nobody can give you one number

If you have read that a Claygent run costs some specific number of credits, that number was invented.

Clay's own documentation says a run is one action plus data credits "which vary by the model you've selected." On the web research column, the published fixed rates run from 0.5 to 3 credits, and ten of the sixteen models Clay lists have no fixed research rate at all. Those are billed variably, at what the model actually cost to run, with Clay stating a 0% markup. Clay withholds an estimate up front, based on the 75th percentile of past runs for that model, then reconciles against the real cost and refunds the difference.

So the honest answer carries a model name attached, and for more than half the models there is no answer in advance at all, only an estimate. Anything else is one model's price presented as the product's price.

One more figure worth knowing if you are thinking about calling Clay from an AI assistant rather than a table: Clay's own FAQ puts credit usage per MCP invocation at "20-100 credits," because the interface is built for one-to-one operations. That is forty to two hundred times what the same email lookup costs in a table. On the free plan's hundred credits, it is between one and five lookups a month.

What Clay does not publish about its own accuracy

Clay publishes a lot of coverage numbers. Work email providers between 52% and 90% coverage. Three providers stacked in a waterfall reaching 90% to 95%. A provider table with incremental coverage and per-lookup prices.

Every one of those figures is about somebody else's data.

This needs stating precisely, because Clay does have a database of its own. Sourcing from it is the one thing in Clay that costs nothing at all: Clay's own worked example prices finding 100 people in the Clay database at zero data credits, and its docs list sourcing lists of accounts or contacts among the operations that consume no action either. You can build a list in Clay for free.

What Clay does not own is the enrichment. The email address, the phone number, the LinkedIn profile, the firmographics: those are bought per record from what Clay describes as 150-plus data and AI providers in its marketplace, and that is what the credits are for. So a Clay-wide accuracy or match rate is not a quantity that exists, because the answer would be a different vendor's number on every row and would change depending on which providers your waterfall happens to hit. That is the architecture, not an evasion. The one internal-test number Clay does publish, that the naive [email protected] pattern returned a valid email about 31% of the time in a software-industry dataset, is a figure about a method Clay is arguing against, with no sample size attached.

The coverage percentages themselves are sourced to "Clay's own data tests, run across thousands of real B2B contacts," with no sample size, no date and no methodology. Clay sells access to every provider in that table and takes margin on the credits you spend there. That does not make the numbers wrong. It does mean nobody has checked them.

We have the same problem, which is why the disclosure is at the top of this article rather than the bottom.

Who this is the wrong tool for

Clay never says who should not buy it. The published numbers do.

If you send by hand, work a list of a few hundred people a month, and mainly need contact details, Clay's subscription is a fixed monthly cost against a variable need. The free tier will not carry the evaluation, the entry plan is $185 a month, and the part of the bill that would justify it, the research column, is the part you are not planning to use. Our comparison of Apollo and Clay works through that trade-off from the other side, and how Apollo's credits actually work covers the same arithmetic on the seat-licensed model.

If the judgment work is the point, the price reads differently. A research column at one credit a row is cheap against the alternative of reading five hundred company websites yourself.

One caveat on the actions figure, since this article is about being precise. Clay does not publish whether a waterfall bills one action or one per provider attempted, and its work email waterfall cascades across more than a hundred providers. If it bills per provider, the action count on a hard row is much higher than the 1,000 above. It does not change the conclusion, because credits bind long before actions do, but it is an open number and Clay has not answered it.

So the decision is not really about Clay's price. It is about whether you have judgment work to automate. Count the rows you would enrich in a month, multiply by 1.45 credits for email plus one research column, and see where that lands against 2,500. If the answer is a few hundred rows and you only want addresses, the subscription is the wrong shape. If it is a thousand rows you would otherwise read by hand, the research column pays for the plan on its own.

One last lever, since it is the biggest one available to you: the model. Helium at 1 credit a row and Argon at 3 do the same job at three times the price, and the research column is already the majority of the bill. Changing that one dropdown moves your cost more than changing plan does.

Prices and mechanics in this article were read on clay.com and university.clay.com on 3 September 2026. Axios reported on 31 August 2026 that Clay was raising at a $7 billion pre-money valuation led by Wellington; Clay has not announced the round, and its own last published valuation is the $5 billion tender offer it announced on 28 January 2026.