Clay enrichment fills the empty columns in a Clay table, such as a work email, a mobile number, a job title or a company's headcount. It does this by asking data providers one at a time until one of them returns an answer. That ordered chain is called the waterfall.
You pay on two meters. Every enrichment that runs uses one action, and data credits are charged on top, depending on the type of data.
Clay's docs say a fully enriched record typically costs 6 to 20 data credits. Clay prices data credits from $0.05 each, so one complete contact costs roughly $0.30 to $1.00 in data credits at the floor price, before actions. When an enrichment finds nothing, whether you get the charge back depends on the outcome, as the refund table below shows.
Enrichment cannot choose the people in the table, and it cannot protect your sending domain. When a campaign goes quiet, open the list before you touch the email. The file usually explains the silence better than the copy does.
What is Clay enrichment and how does the waterfall actually run?
Every row in a Clay table is one person or one company. Every enrichment column is a lookup that tries to fill one field for that row.
A waterfall chains several providers for the same field. Clay's guide to building a data waterfall says waterfalls run multiple providers in a predetermined sequence, so the same task is not repeated and no extra credits are spent. Provider one gets the first try. If it comes back empty, provider two gets the row, and so on down the chain.
How the parts work in practice:
- Starting point. Pre-built waterfalls sit behind "Add enrichment" at the top right of a table. You search there for the data point you want, for example a phone number.
- Inputs. The waterfall needs a starting value to search from, for instance an email address or a LinkedIn profile link. For a work email, plan on at least a full name and a company domain in every row. Without a domain, an email finder has nothing to build from.
- Order. You can reorder providers, add new ones or delete them. A toggle next to each provider skips it without removing it.
- Winner column. An optional setting outputs the name of the provider that succeeded, and the individual provider columns can be hidden.
- Run rules. You can switch on auto-update, or define rules that decide whether the waterfall runs.
- Templates. A custom waterfall can be saved with "Save as template", with a name, a description and a category, so the next table starts from the same setup.
Turn on the winner column from day one. After a few hundred rows it shows which provider actually covers your market. That is the only evidence worth using when you reorder the chain.
Run conditions are the cheapest control in the product, because a row that never runs never costs anything.
Do you need Clay at all?
Run a quick test before you pay for anything. Take 200 accounts from your real target list and pull them through the single database you already have. Count how many rows come back with a usable work email and the right title.
- If most rows come back complete, one provider is enough. A waterfall would add cost and setup work without adding many contacts.
- If one source leaves big holes in a big list, a waterfall starts to pay for itself, because each extra provider fills rows the first one missed.
- If your whole market is a few thousand companies, coverage stops being the main question. I cover whether a waterfall is worth it for your market size in a separate piece.
I would rather stitch a list together from proven providers such as RocketReach, ZoomInfo, UpLead, Apollo and Sales Navigator than use anything scraped. I also won't shrink a list to a few hundred perfect-fit rows just to make it look clean. A campaign needs a list that is big enough and clean enough at the same time. If budget is the constraint, compare the cost of B2B contact data per record and the alternatives to ZoomInfo before you commit to a platform.
| Option | What you get | How you pay | Fits when |
|---|---|---|---|
| Clay waterfall | Several providers tried in order for each field | A platform plan plus actions and data credits, priced in the next section | A big list where one source leaves gaps, and someone on the team can build and maintain tables |
| Single database (Apollo) | Emails and phone numbers from a database Apollo puts at 240M+ contacts (a vendor figure). Apollo says this enrichment works on any plan, including free (Apollo) | Your Apollo plan, with the same limits as your account | Most of your accounts already show up with the right people |
| Sales Navigator plus an email finder | People found on LinkedIn, with emails added by a finder tool | A Sales Navigator seat plus per-lookup finder fees, both priced by the vendors | A narrow, title-driven list where you want to see every profile yourself |
| A done-for-you list | A finished, verified file built by someone else | Per list or per engagement, quoted by the provider | You need the list now and nobody in-house wants to learn enrichment |
What does Clay enrichment cost per contact?
Clay meters two things. Clay's pricing page says actions measure platform usage, such as running an enrichment, and data credits buy data and AI from the vendors in Clay's marketplace. Clay's credits documentation adds the per-run detail. Each enrichment always uses 1 action, while data credits vary from 0.5 to 10+ depending on the data type. All prices and plan sizes below are as of the time of writing.
| Plan | Price shown | Actions per month | Data credits per month | Note |
|---|---|---|---|---|
| Free | $0 | 500 | 100 | Waterfalls are included, phone number enrichment is not |
| Launch | $167 on the plan card, from $185 in the FAQ | 15,000 | 2,500 to 10,000 | - |
| Growth | $446 on the plan card, from $495 in the FAQ | 40,000 | 6,000 to 100,000 | Automatic CRM sync and enrichment start here |
| Enterprise | Custom | 100,000+ | 100,000+ | CRM sync every 15 minutes instead of once a day |
The pricing page shows two prices for each paid plan: $167 and $446 a month on the plan cards, and "starting at" $185 and $495 in its FAQ. Look at the billing period on your checkout screen before you compare plans.
The worked math at the floor price
Here is the same 6 to 20 credit range at two list sizes:
| Volume | Data credits needed | Data credit cost at $0.05 |
|---|---|---|
| 100 full records | 600 to 2,000 | $30 to $100 |
| 1,000 full records | 6,000 to 20,000 | $300 to $1,000 |
Treat these figures as estimates. They assume the lowest published credit price, and what you actually pay per credit depends on your plan and billing. Data is also only one line item in what a booked meeting costs, which I break down in cost per B2B meeting breakdown.
Actions are cheaper, and Clay prices them from less than $0.01 each. They still have a monthly ceiling. Say one record passes through five enrichment columns: email, title, phone, headcount and a verification step. That record uses five actions. At 1,000 records you have used 5,000 actions, which is a third of Launch's monthly allowance.
Now set the credits against the plan. Launch's smallest allowance of 2,500 data credits covers about 125 records at 20 credits each, or about 417 at 6. The largest Launch allowance of 10,000 credits covers about 500 to 1,667 records.
A team that needs thousands of fully enriched contacts a month will run out of data credits long before it runs out of actions. At outbound volume, the math gets heavy fast. In my campaigns, a monthly target of 30 to 35 booked calls means reaching at least 10,000 people, and enriching 10,000 records at even 6 credits each uses 60,000 data credits.
Your own API keys
You can connect your own provider accounts. The pricing page says each run then counts as one action but uses no data credit. The docs give an example: enriching 100 contacts costs 95 data credits through the marketplace, and 50 when the email lookups go through your own email provider key.
At the floor price, that is about $4.75 against $2.50 in Clay credits. Your email provider now bills you directly, though, so the real saving is the gap between the two invoices.
Top-ups and rollover
Running short in the middle of a month is expensive. One-time data credit top-ups carry a 30% premium, and actions cannot be topped up at all.
On monthly plans, unused data credits roll over up to 2x the monthly limit. Annual plans roll over far less, 15% per Clay's credits documentation. Actions do not roll over. Work out the month's volume first, then buy the plan that fits it.
When does Clay refund credits, and when are you still charged?
Clay's credit refunds documentation explains how billing works. Each record is billed when a data provider bills Clay for the API call behind the enrichment. Refunds are automatic, so there is nothing to request, and a refund returns both the data credits and the action.
| Run outcome | Refunded? | What to do |
|---|---|---|
| Success | No | Nothing: you paid for an answer and got one |
| Empty response | No | Count it as spend; a provider that returns these often belongs lower in the chain |
| No data | Only for the actions Clay lists | The list covers email and phone finding, verification, people and company enrichment and more; check it for your action |
| Error | In most cases | Rerun the failed rows instead of assuming they cost nothing |
| Timed out | Yes | Rerun later |
| Invalid input | Yes | Fix the input column, for example a malformed domain |
| Missing input | Yes | Fill the empty input cells first |
| Rate limit wait time exceeded | Yes | Run smaller batches |
| Retried but failed | Yes | Rerun later |
| Run condition not met | Yes | Nothing: your filter did its job |
One outcome catches people out. Clay treats a verification result of invalid as a real answer, so it charges for it. You get a refund only if the verifier gives no verdict at all.
Dead addresses are common. ZeroBounce's 2026 list decay report, from an email verification vendor, found that 23% of the email addresses it analyzed were invalid. That report also puts yearly decay of an email list at a quarter or more.
So if your table verifies 1,000 emails and 200 of them are dead, you pay for all 1,000 checks. Deduplicate and filter before you verify, never after.
When you use your own API key, no data credits are charged, so none come back. The action is still charged, and it is refunded on the outcomes in the table above.
Refunded cells read "Charged: 0 (Refunded)". A fully refunded step looks exactly like a step that cost nothing. That means a provider failing on half your rows can look harmless in the grid. Check the run status and the table's total spend instead.
How should you set up enrichment so you pay only for contacts you will send to?
This order keeps rows you will never email away from the paid lookups:
- Begin with the customers you already have. My first question on any new table is who bought before and who is buying now, and the lookalike search comes after that. The full method for building the list before you enrich it is a separate piece.
- Write down the real buyer titles. Target the titles that actually sign contracts in that niche, and avoid a bare "manager". A CTO, a head of procurement and an owner are three different searches.
- Exclude generic titles before a single credit is spent. I strip Assistant, Manager and Associate out of a target list on purpose. In Clay, put that logic in a filter column and use it as the run condition, so excluded rows never trigger a paid lookup.
- Remove duplicates and current customers. Those rows should never be emailed, so they should never be enriched.
- Enrich what is left. Only at this point does the waterfall run.
- Verify every address, then send only to the verified ones. I check each address with NeverBounce or DeBounce and hold bounces under 0.5%. Here is what bounce rate is safe and why.
The saving is easy to calculate. Say a 1,000-row import holds 300 generic titles. Filtering first keeps those 300 rows away from every paid lookup. At the floor price, that saves $90 to $300 in data credits on that one table.
Where do enriched lists still break?
Enrichment fills cells. It never decides who belongs in the table. A perfectly enriched list of the wrong people is still the wrong list. It just has phone numbers now.
The usual order goes like this: a campaign goes quiet, the team rewrites the subject line, and nobody opens the file the emails went to. The email takes the blame because everyone can see it. No rewrite saves a bad list or a weak sender.
Check these places before blaming anything else:
- Targeting that is wrong but neatly filled in. A headcount range that is far too wide mixes several kinds of buyer into one market, and enrichment makes every one of those rows look complete.
- One huge region shared by the whole team. Split territory into specific countries per rep, so no buyer hears from two of your reps at once.
- A strong list sent from a weak sender. Same message, two companies. One got 2% replies, the other 7.5%. Nothing in the copy changed. Buyers check the sender's profile, website and proof before they answer. A plain message from a sender they can check out beats a perfect one from a ghost.
- Volume on a broken base. Thousands of personalized messages a week still go unanswered when the base underneath them is wrong. Test the list at small numbers first. A weak result there will repeat at scale. Enriching more rows will not fix it.
- Gaps in the sending setup. Enrichment does not buy domains, set authentication records or warm up mailboxes.
On that last point, Google's rules are specific. Since February 1, 2024, everyone sending to Gmail accounts must set up SPF or DKIM. Anyone sending over 5,000 messages daily must also add DMARC alongside SPF and DKIM, though the DMARC policy may stay at none (Google Workspace Admin Help).
Spam rates reported in Postmaster Tools should stay below 0.10% and should never reach 0.30%. Bulk marketing mail also needs one-click unsubscribe.
Gmail's help page also warns senders against purchasing addresses from third parties (Gmail Help). That warning is about off-the-shelf files bought from list sellers. A list built for your campaign from real sources and sized to your market is a different thing, whether your team builds it or someone builds it for you.
On the sending side, my team never sends cold email from the domain a company's invoices come from, and we warm every new mailbox for at least 10 to 14 days before the first message goes out.
Is it legal to email contacts that Clay enriched?
It depends on who the contact is and where they are. For campaigns aimed at the USA, the UK and Europe, two rule sets come up most often.
United Kingdom. The ICO's business-to-business marketing guidance draws its main line around named people:
- A work email that identifies a named person, such as a director's own address, is personal data under UK GDPR. A generic inbox with no known person behind it is not.
- Under PECR, the UK's rules for electronic marketing, the consent rules do not cover corporate subscribers such as limited companies. Even so, always say who you are and provide a valid opt-out address.
- Individual subscribers include sole traders and certain partnerships. Before emailing them, you need their consent. The alternative is the soft opt-in. It applies to people who bought from you or asked about your services and were offered an opt-out.
- A person's data being public does not mean they agreed to direct marketing.
- When personal data comes from third parties or public sources, the people concerned must get privacy information within one month at the latest.
- Any list you buy or sell needs a lawful basis behind it, and anyone can object to being marketed to, with no exceptions.
The ICO is reviewing this guidance following the Data (Use and Access) Act, so recheck it before each campaign.
For an enriched table, that means three things. Named people need a plan for the privacy notice. Opt-outs go on a suppression list that every future table is checked against. Sole traders get filtered out unless you have their consent or the soft opt-in applies.
California. Per the California Attorney General, the CCPA carve-out covering B2B personal information ended on December 31, 2022. The law reaches a for-profit business that meets any one of these tests:
- gross annual revenue above $25 million
- handling, through buying, selling or sharing, data on California residents or households, once that count hits 100,000
- earning half or more of its annual revenue from selling that information
Information a business reasonably believes the consumer lawfully made public is excluded. Opt-out requests must be handled within 15 business days at most.
Germany, Austria and the Czech Republic. My team does not cold email contacts in these markets without a compliant basis, and we use LinkedIn there instead.
This is not legal advice. The rest of the rules are in the full legal picture for cold outreach.
The order that keeps the bill small
Filter, then enrich, then verify, then send. Every step Clay runs out of that order shows up on the invoice, and none of them fixes a list of the wrong people. Before you buy a bigger plan, open last month's table and count the rows you should never have enriched.
Wondering whether Clay is worth its credits at your list size? Book a call and send me your row count, and I will tell you what I would pay for.
Questions people ask about Clay enrichment
Is Clay better than Apollo for a small team?
For most small teams, start with Apollo or whichever single database you already pay for. Move to Clay when the gaps start costing you deals. Clay rewards a person who builds and maintains tables. A team without that person pays for a platform it never really uses.
Can you enrich more than 50,000 records at once in Clay?
Only on Enterprise. Clay's pricing page reserves bulk enrichment of more than 50,000 records at a time for that plan. On smaller plans, split a large file into batches and run your exclusion filters on each batch first. That way a big import does not burn the month's data credits on rows you would have removed anyway.
Do Clay refunds free up workbook and budget limits?
Yes. Per Clay's refund documentation, refunded credits are also removed from the spend that counts toward workbook and budget limits, which frees that allowance. This matters if you cap spend per table. A run that fails and gets refunded does not leave the table stuck at its limit with nothing to show for it.
Does Clay enrichment break GDPR for EU contacts?
Enrichment is a data step, so the risk lies in what you do with named people's details afterwards. Keep every opt-out on a suppression list that new tables check before they run. Record which provider supplied each contact, which the winner column already gives you. Then take advice for each country before you email EU contacts at any volume.
