Clay Phone Enrichment: Which Providers to Waterfall, What Each Number Costs, and How to Send the List to a Dialer

Clay makes it easy to chain phone providers: add a column, pick a provider, add another one behind it, and let the table fill. What it does not decide for you is the order of those columns, how much each mobile really costs once the cascade has run, and what happens to the list after the last column. Those three choices decide whether your reps spend the afternoon talking to buyers or listening to voicemail greetings. This guide uses the mobile phone test Clay published on May 3, 2026, the prices we checked on November 6, 2026, and the public API of a dialer to walk through each step.

The phone waterfall in five lines

If you are new to the idea, our comparison of waterfall enrichment tools explains it in detail. In short, applied to phone numbers:

1. 01

   You start from people you already chose:

   a first name, a last name, a company domain and ideally a LinkedIn URL.
2. 02

   The first provider looks for a mobile:

   if it finds one, the row is done.
3. 03

   The rows still empty go to the next provider:

   and so on down the list of columns.
4. 04

   You only pay for results:

   Clay does not charge data credits for a lookup that returns nothing.
5. 05

   The last column is quality control:

   format, number type and source, before anyone dials.

The whole game is in steps 2 and 3: which provider comes first, and which ones follow. Clay’s interface lets you build the cascade in a few minutes. Getting the order right takes data.

What Clay’s 9,806-number test actually measured

Clay ran the test with TitanX, a company that scores phone numbers. Clay published the results on its blog on May 3, 2026, with the methodology on Clay University. The setup is worth understanding before you reuse the conclusions, because it explains both their strength and their limits.

The ten providers were Forager, Bytemine, Wiza, Datagma, Leadmagic, RocketReach, People Data Labs, ContactOut, Findymail and Prospeo. Clay kept only unique pairs of a number and a full name, then had the numbers scored. In North America and EMEA, TitanX used manual cold calls plus eleven other signals over four days and sorted each number into four grades. In APAC, an offshore team called each number at least five times until it reached a clear outcome.

Each provider then got a quality score per region, where a P2 counts for half a P1, a P3 for a quarter, and a bad number counts against the provider. That last rule matters: a provider that returns many wrong numbers is penalized, not just ignored.

The provider order by region

Clay published its recommended waterfall for each region as images, so the table below is our reading of the figures Clay published in text: coverage leaders, accuracy leader and budget picks. “Coverage” means the number of mobiles returned in the test; “quality” means the score described above.

ContactOut deserves a note of its own: it returned 2,878 mobiles across the three regions, close to half of the 6,000 contacts, the largest volume in the test. Clay also describes it as one of the pricier options. In a waterfall, that makes it a strong second column in North America and APAC rather than the first one.

Clay also writes that Prospeo had strong APAC coverage but could not be included in the APAC scoring. Treat Prospeo as unmeasured in APAC rather than weak.

The providers in the test, one by one

The prices below are the entry prices Clay listed in its article, next to what we checked on the vendors’ own pages on November 6, 2026 when we had it. They are what you pay with your own account; inside Clay, you pay in Clay data credits instead, unless you connect your own key.

Findymail and RocketReach are North American volume players in the test (1,620 and 1,600 mobiles), Wiza and People Data Labs are also strongest in North America. Clay lists Findymail from $49 a month for 1,000 credits, RocketReach from $108 a month for email and phone, Wiza from $83 a month and People Data Labs from $98 a month for 350 records. For the full price grid of the providers you are more likely to run outside Clay (Lusha, Apollo, Kaspr, Cognism, FullEnrich and others), see our phone enrichment pricing tool.

What a number really costs in a Clay waterfall

There are two ways to pay for each column, and they change the math.

On the Clay side, the numbers we checked on November 6, 2026: phone enrichment starts on the Launch plan, which Clay’s FAQ lists at $185 a month for 2,500 data credits while its pricing page also shows $167 with 3,000 data credits; that puts a data credit at roughly 6 to 7 cents. Clay’s documentation puts a fully enriched record at 6 to 20 data credits depending on the provider, so roughly $0.45 to $1.50 per record with a phone. Unused credits can accumulate up to twice the monthly amount. The free plan does not include phone enrichment.

On the direct side, here is the cost of one mobile found on the entry plans we checked, as recorded in our pricing tool:

Provider (own account)	Entry plan checked November 6, 2026	Credits per mobile	Cost per mobile found
Leadmagic	$99 a month, 5,000 credits	5	about $0.10
Prospeo	$39 a month, 1,000 credits	10	about $0.39
Datagma	$49 a month, 3,000 credits	30	about $0.49
FullEnrich (itself a waterfall)	$29 for 500 credits	10	about $0.58

The last line of that calculation is the one that matters. A column that is cheap per lookup but returns many wrong numbers lowers the price on the invoice and raises the price per conversation. That is exactly why Clay’s test scores providers on quality, and why the first column should be the most accurate one. To turn your own figures into a cost per conversation and per meeting, use the phone enrichment pricing tool: it takes a match rate, a share of wrong numbers and a pickup rate by number type.

Before you export: the columns to normalize

A Clay table after a waterfall usually holds one phone column per provider, in as many formats as there are providers. A dialer needs one clean number per contact, and your reps need to know what kind of number it is. Add these columns before anything leaves Clay.

1. 01

   One final number, in E.164:

   a single column that takes the first non-empty provider column and writes it as +[country code][number], with no spaces, dots or leading zero. Dialers and APIs expect this format, and it removes the “01 23” versus “+33 1 23” duplicates.
2. 02

   The number type:

   mobile, direct line or switchboard. Some providers label it, most do not. A number shared by several contacts at the same company is a switchboard, whatever the label says.
3. 03

   The source:

   the name of the provider that found the number. You need it to compare providers on your own results, and in Europe to tell the person where you got their number.
4. 04

   The enrichment date:

   numbers decay. A date column tells you when a list needs a refresh.
5. 05

   A do-not-call flag:

   a column for contacts who must not be called, filled from your CRM, your suppression list and the national registers that apply.

If you prefer to clean outside Clay, export the table and run it through our free tools: the phone number validator for E.164 format and invalid numbers, the direct dial checker to sort mobiles, direct lines and switchboards, and the CSV cleaner for broken files from Excel. If your numbers come from several exports rather than one Clay table, the waterfall enrichment merger joins them, keeps the best number per contact and shows what each provider really added.

Sending the list from Clay to a dialer

Two routes, depending on how often you run the table.

Clay’s HTTP API column shows the response status in each cell, so a failed row is visible and can be re-run. Integration guides written for Clay add two practical warnings: requests can be retried, so the receiving system should cope with duplicates, and a shared secret in the Authorization header is the usual way to authenticate, since Clay does not sign requests by itself.

Example with Skipcall’s public API

Skipcall’s public API is documented at docs.skipcall.app. Two endpoints are enough for a Clay table, as documented on November 6, 2026:

1. 01

   Create the contact:

   POST https://api-public.skipcall.app/v1/contacts with a Bearer API key. Required fields: firstName, lastName and phoneNumbers, a list of numbers in E.164 format. Optional fields include email, company, country, role, linkedinUrl and custom fields. A multi-workspace key adds an x-workspace-id header to choose the workspace.
2. 02

   Add it to an existing sequence:

   POST /v1/sequences/{id}/contacts with up to 100 contact IDs per call. Contacts are assigned round-robin among the sequence’s assignees unless you name one, and contacts already in the sequence are ignored.
3. 03

   Call from Skipcall:

   reps work the sequence from the power dialer or the parallel dialer. The API does not place calls; it puts the right contacts in front of the right reps.

A minimal body for the first call looks like this, with Clay column names inserted where the values go:

{
  "firstName": "Jane",
  "lastName": "Doe",
  "company": "Acme Inc.",
  "country": "United States",
  "role": "VP Sales",
  "phoneNumbers": ["+14155550123"]
}

API access is included on the Starter, Intermédiaire, Expert and VoIP plans, with a rate limit of 100 requests a minute (1,000 on Expert). A Clay table that sends a few hundred rows at once should therefore pace its requests or send them in batches. For other automation patterns around a phone system, see our guide to telephony APIs and automation and the Clay integration page.

One table, three regions: conditional columns

Most teams do not call a single region. The tempting shortcut is one waterfall for everyone, ordered by a global ranking. Clay’s figures show why that wastes money: the provider that leads North American coverage is not the one that leads in EMEA, and the cheapest good number comes from Bytemine in one region and Leadmagic in the other two. A single order means some of your contacts always start with a provider that is weak where they live.

The fix fits in one table. Clay lets you set a run condition on each enrichment column, so a column only runs when a formula is true. Normalize the country first, in a column that holds a region code (NAMER, EMEA or APAC), then give each provider column a condition on that code and on the final phone column still being empty.

Two more conditions save credits. The first is a seniority or account filter on the expensive columns: keep Forager for the accounts and roles you would call by hand, and let the volume providers handle the long tail. The second is a stop rule: once a row holds a mobile, no later column runs, even if a cheaper provider might return a second number. Two mobiles for the same person rarely double your chances; they mostly double the bill and confuse the rep who has to pick one.

Last point on regions: Clay’s test grouped Europe, the Middle East and Africa into one score, but coverage inside that block varies a lot from one country to the next. If you call mostly Germany, France or the Nordics, look at country-level evidence before trusting an EMEA average; our B2B phone data coverage map sorts what is published by country and by provider, and says where nothing is published.

Mistakes that make a Clay phone waterfall expensive

1. 01

   Putting the cheapest provider first:

   it fills rows with wrong numbers, and the waterfall never asks the better providers for those contacts.
2. 02

   Using the same order for every country:

   Clay’s own figures show different leaders in North America, EMEA and APAC. Split the table by region, or add a condition on the country before each column.
3. 03

   Judging providers on fill rate:

   a filled cell is not a working number. Call a sample and count the numbers that reach the right person.
4. 04

   Re-enriching contacts you already have:

   if an Apollo or CRM export already holds a mobile, run the waterfall only on the empty rows, or you pay twice for the same person.
5. 05

   Keeping one phone column per provider in the export:

   reps then call the first number they see, often a switchboard. Export one final number and its type.
6. 06

   Forgetting the source column:

   without it, you cannot compare providers on your own results, and in Europe you cannot tell a prospect where their number came from.
7. 07

   Never refreshing:

   people change jobs and numbers. Our B2B data decay calculator estimates how much of a list has gone bad and when to refresh it.

Improving Apollo phone data with Clay

It is one of the questions that comes back most often in Clay’s community: “I have Apollo, can Clay make the phone numbers better?” Yes, if you treat Apollo as the first column you already paid for, not as a source to replace.

1. 01

   Import the Apollo export into a Clay table:

   keep Apollo’s mobile, direct and corporate phone columns as they are.
2. 02

   Flag the rows to complete:

   no mobile, or a number shared by several contacts at the same company.
3. 03

   Run the waterfall on those rows only:

   with a condition on the flag, so the providers never see the contacts Apollo already covered.
4. 04

   Merge into one final number:

   the Apollo mobile when there is one, otherwise the first mobile from the waterfall.

The same logic works for numbers coming from your CRM, from Lusha or from any other export. If you only need to compare and merge several exports you already own, our waterfall enrichment merger does it in the browser without a Clay table. And if you use Apollo’s own calling, our review of Apollo’s dialer and its limits covers what it includes and when a dedicated dialer helps.

Test your own waterfall in one afternoon

Clay’s test is a starting point. Your market, your roles and your countries are what count. A short test tells you more than any benchmark, including Clay’s.

1. 01

   Take 100 real prospects:

   the roles and countries you actually call, not only well-known executives.
2. 02

   Run them through three providers in parallel, not in a waterfall:

   so each provider sees all 100 names and you can compare like for like.
3. 03

   Call every number returned, within the same week:

   same time slots, same caller ID.
4. 04

   Count per provider:

   numbers returned, mobiles among them, numbers that reached the right person, and the cost.
5. 05

   Build the order from those counts:

   the most accurate first, the one that adds the most new good numbers second, the cheapest last.

The rules do not change because the number came from Clay

The provider that found a number does not change what you may do with it. In the United States, manual business calls are generally allowed, but the TCPA restricts autodialed and prerecorded calls to cell phones without prior consent, and you must honor do-not-call requests. In the European Union, calling a business contact usually rests on legitimate interest under article 6 of the GDPR, and article 14 requires you to tell the person where you got their data. In the United Kingdom, screen numbers against the TPS and the CTPS before you dial. Our cold calling compliance checker lists the rules country by country, and our guide to cold calling cell phones covers the US mobile rules in detail. This is general information, not legal advice.

How Skipcall fits this workflow

Skipcall is a business phone and outbound calling platform, not a contact database: it does not sell lists and does not build them. Your Clay table stays the place where you find numbers; Skipcall is where you call them, with a power dialer or a parallel dialer on up to four lines, call recording and transcription when recording is on, and sync with HubSpot, Pipedrive, Salesforce and Attio. Contacts arrive by CSV or through the public API described above, and Skipcall’s built-in enrichment can complete the contacts you already have, on credits.

Clay waterfall checklist