Clay pricing explained: plans, credits and what a month really costs

This page is for founders, RevOps leads and agency owners who are pricing Clay before they buy, or who already use it and want to know why the bill looks the way it does. You'll get more from it if you know roughly how many records you enrich a month, how many of them need a phone number, and whether you'll connect your own data provider keys.

How we checked

Bebop, made by Blackpearl Group, wrote this page. It sells a different kind of product and appears once in the comparison table below. These are the rules we followed:

• Facts. Every Clay price, allowance and rule comes from Clay's own pricing page, help docs and blog, checked on 26 October 2026 and listed in the sources. Prices are list prices before tax.
• Other vendors. Prices for other tools come from their own pricing pages on the same day. We don't use third-party price estimates.
• No hands-on scores. We haven't benchmarked Clay's data against other tools, so there are no ratings here. The two-week test below is how to measure it on your own accounts.
• Where Clay's pages disagree, we say so and give both figures.

Clay's plans on 26 October 2026

All paid plans include unlimited seats, so a five-person team pays the same as one person. Annual billing saves 10% and grants the whole year's credits up front.

One display quirk: on the day we checked, the Launch plan card labelled its monthly Data Credits "3K", while Clay's FAQ, its plans help page and the tier menu all give 2,500 for the $185 entry price. Budget on 2,500.

How the headline price is built

The $185 and $495 figures aren't fixed plan fees. Each is the cheapest Actions tier plus the cheapest Data Credits tier for that plan, and you can raise either one on its own:

The monthly Data Credits tiers on Launch run 2,500 ($125), 6,000 ($290), 10,000 ($460), 20,000 ($880) and 50,000 ($2,125). That works out to $0.050, $0.048, $0.046, $0.044 and $0.0425 a credit, so the unit price falls only slowly as you buy more. Monthly Actions tiers on Launch run from 15,000 ($60) to 200,000 ($540).

Two things follow. First, most teams that outgrow Launch outgrow it on Data Credits, not Actions: Clay says its Launch and Growth plans are sized so 90% of customers never hit their Actions limit. Second, moving from Launch to Growth is mainly a features decision. Growth's entry Actions tier costs $205 against $150 for the same 40,000 Actions on Launch; the difference buys CRM sync, the HTTP API and web intent signals.

What a Data Credit and an Action actually buy

Clay's help docs separate the two meters like this:

The cost of a record depends on what you enrich. In Clay's own worked example, a LinkedIn profile costs 0.5 Data Credits, an email found costs 0.5, and validating it with select providers costs nothing, so 100 contacts with profiles and emails used 95 Data Credits. Clay's docs put a fully enriched record at 6 to 20 Data Credits and say plainly that "emails are cheap, phone numbers are expensive". Clay doesn't publish one price per phone number, because it depends on the providers in your waterfall; the table shows the estimate before you run a column.

AI columns have their own rates. About 80% of models charge a flat number of Data Credits per row: for web research, Clay's own Helium model costs 1 and Argon costs 3, and GPT-5 Mini costs 1. Larger reasoning models are priced on the tokens each run actually uses, with no markup. Clay holds back an estimate before the run, based on the 75th percentile of past runs, and returns the difference afterwards. Every AI run also costs 1 Action.

Bringing your own API keys skips Data Credits for that provider but still costs 1 Action per run, and you pay the provider directly. That only saves money if you already pay for the provider, or if its own price is lower than Clay's credits for the same data.

Rollover, top-ups, trials and legacy plans

Clay introduced this model on 11 March 2026. Its announcement put Growth at $495 against $800 for the legacy Pro plan, cut the top-up premium from 50% to 30%, and warned that customers on Pro or using their own API keys may pay more under the new model, because Actions are charged even when the data comes from your own keys.

Worked example: a small agency's monthly Clay bill

Illustrative example. The agency, volumes and phone cost are invented. Clay's prices and per-item costs are from its pricing page and help docs, checked 26 October 2026.

The team: Forgeline Growth (forgeline.example) is a founder-led agency of three people that runs outbound programs for mid-sized metal fabricators and plastics molders. For its own pipeline it targets 300 manufacturers a month, two contacts at each: the plant manager and the VP of operations. That's 600 people.

The table, step by step:

The bill: 1,470 of 2,500 Data Credits and 2,520 of 15,000 Actions fit Launch at $185 billed monthly ($167 annually). The unused 1,030 Data Credits roll over, up to the 5,000 cap. Cost per emailable contact: $185 ÷ 540 = $0.34.

Add phones: the founder wants mobile numbers for 150 plant managers, who often answer their phones more readily than their email. Assume the phone providers in the waterfall average 10 Data Credits per number found (an illustrative figure inside Clay's 6 to 20 range for a fully enriched record; check the estimate on your own column). That's 150 × 10 = 1,500 more credits, for 2,970 in total, 470 over the allowance. Three ways to cover it:

• Use rollover. If last month left 1,030 unused, this month's 470 overage is already covered. Rollover makes a lumpy month cheap.
• Top up 470 credits. At about $0.065 a credit (the $0.05 entry rate plus the 30% premium), that's roughly $31, for a month of about $216.
• Raise the Data Credits tier to 6,000. $60 + $290 = $350 a month. The extra $165 buys 3,500 credits, about $0.047 each. The same $165 spent on top-ups buys about 2,540 credits, so moving up a tier only pays if you're regularly short by more than about 2,500 credits a month.

At $216 for 540 emails and 150 mobiles, the cost per contact is $216 ÷ 540 = $0.40.

The same contacts elsewhere, at list prices on 26 October 2026: Apollo charges 1 credit per email and 8 per phone number, so 540 + (150 × 8) = 1,740 credits, inside one Basic seat's 2,500 at $65 a month ($49 annually). Hunter's Starter plan covers the 540 emails (2,000 credits, $49 a month or $34 yearly) but not phones. Both are cheaper for the raw data. What Forgeline gets from Clay is the workflow: several providers tried in order, an AI research column for every account, and custom logic in one table. If the agency only needed emails, Clay would be the expensive choice.

Clay against other options at a glance

Prices from each vendor's own pricing page, checked 26 October 2026. Annual figures are the monthly equivalent when billed annually.

For a full list of replacements, by the job you want done, see Clay alternatives. If you're weighing Clay against an all-in-one database, the credit maths in Apollo alternatives applies here too.

Which Clay plan fits

Watch the cost of building, too. A Clay table is only as good as the person maintaining it. If nobody on your team will own the waterfall, the AI prompts and the exports, the subscription is the smaller part of the cost.

How to run a two-week test before you pay

The 14-day trial gives you 1,000 Data Credits and no phone enrichment. That's enough for a fair test of emails and research on about 50 accounts, if you plan it.

1. Pick 50 accounts you already know. Customers, recent lost deals and researched targets, with the buyer you'd want at each and an email you know works.
2. Build the table you'd really run. Same columns, same providers and the same AI prompt you'd use at volume. Note the Data Credit estimate Clay shows on each column before you run it.
3. Run 10 rows first. Clay's own AI pricing guide suggests testing complex prompts on 10 to 50 rows so you can see the real charge per row before running a whole table.
4. Read the meters. Open Settings, then Usage, and record Data Credits and Actions used per column and per record.
5. Score every record. Give +1 for the right person in their current role with a working email, 0 for a plausible record you can't confirm, and -1 for a wrong person or a bounce. Treat an unsupported AI claim as -1 too. This mirrors GTM-Bench, Blackpearl Group's June 2026 benchmark of AI prospecting systems, where several general-purpose AI agents finished with negative net scores because weak records counted against them.
6. Project your month. Multiply credits per usable record by your monthly volume, add a margin for phones, and pick the tier that covers it without regular top-ups.
7. Test phones on a paid month. If phones matter, buy one month of Launch on monthly billing, run 50 phone lookups and dial them before you commit to a year.

Accounts in test: ___      Contacts requested: ___
Columns run (provider or model, credit estimate shown):
1.
2.
3.

USAGE (from Settings, then Usage)
Data Credits used: ___   Actions used: ___
Data Credits per usable record: ___ ÷ ___ = ___

SCORES
+1 (right person, current role, working email): ___
 0 (plausible, can't confirm): ___
-1 (wrong person, left, bounced, unsupported claim): ___
Net score: ___

PROJECTION
Records a month: ___ × ___ credits = ___ Data Credits
Actions a month: ___
Plan and tiers that cover it: ___   Monthly cost: $___
Cost per usable record: $___ ÷ ___ = $___

Decision and why:

Where Clay's pricing catches people out

• Actions on your own keys. Connecting your own provider keys removes Data Credits, not Actions. Heavy users of their own keys can pay more under the March 2026 model than before.
• Exports cost Actions. Every record pushed to a CRM, sequencer or ad platform is an Action. A table that re-syncs often can use Actions faster than enrichment does.
• Annual rollover is small. Monthly plans can bank up to twice the monthly amount, but annual plans carry over only 15% of a year's credits, and only if you renew at the same tier or higher.
• Top-ups cost 30% more. Fine for an odd month. If you top up every month, raise your Data Credits tier instead.
• The trial can't test phones. Budget a paid month if mobile numbers drive your decision.

Next step: if the numbers above make Clay look like more tool than you need, compare the options by job in Clay alternatives.