Clay Waterfall Enrichment (2026): Hit Rates and Real Cost per Email
•
Clay waterfall enrichment chains data providers in a fixed order and stops at the first confident match, which lifts coverage into the low 90s from anywhere between 40% and 90% on a single provider. But coverage is not your hit rate. Multiply coverage by accuracy and the real number drops: the best provider in Clay's own benchmark delivers 86.9% of a list, and a five-provider chain lands about 90%.
The part nobody publishes is what happens when you send. Every single provider in that benchmark returns a list that bounces above 2% if you mail it raw. This post does the arithmetic Clay's own data supports but nobody has run.
What waterfall enrichment actually does
Waterfall enrichment queries several data providers in a fixed sequence and keeps the first result that clears your confidence bar. A record only reaches the second provider if the first returned nothing usable, and only reaches the third if the first two both missed. The pool shrinks at every step, so your most expensive source runs on the fewest records.
Since the March 2026 pricing change, Clay does not charge for a lookup that comes back empty. You pay for matches, not attempts. That single billing detail is what makes a long chain affordable, and it shows up in the cost math below.
The logic works on any field with more than one supplier behind it. Work email is the common case because there are dozens of vendors and a verification step sits right next to it. Mobile numbers and firmographics run identically. Only the providers change. If you are still getting oriented in the platform itself, start with the complete Clay guide and come back here for the provider math.
The provider numbers Clay publishes
Clay runs a public benchmark of work email providers against ground-truth data and publishes three figures for each: data quality, data coverage, and credit cost per lookup. This is the most transparent dataset in the category and it is the foundation for everything below. Clay tests by region; every figure in this post is the North America cut, last tested 22 October 2025. If you sell into Europe or APAC, read the section on limits before you copy any of this.
Quality is the share of returned addresses that are correct. Coverage is the share of records the provider returns anything at all for.
Provider | Quality | Coverage | Credits |
|---|---|---|---|
Hunter | 97.15% | 52.87% | 0.4 |
Findymail | 96.26% | 90.26% | 0.5 |
Kitt AI | 95.22% | 72.49% | 1 |
BetterContact | 95.05% | 81.66% | 1.5 |
Wiza | 94.72% | 85.02% | 1 |
Enrow | 94.10% | 70.94% | 0.2 |
Icypeas | 93.83% | 51.90% | 0.2 |
Dropcontact | 91.81% | 75.45% | 1 |
Datagma | 91.33% | 62.01% | 0.4 |
FullEnrich | 87.98% | 83.17% | 1.5 |
Zeliq | 86.10% | 85.56% | 2 |
LeadMagic | 81.95% | 68.02% | 0.3 |
Read the shape before the names. Hunter is the most accurate source in the set and reaches barely half a list. Findymail reaches nine in ten and gives up under a point of accuracy to do it. The two cheapest options sit at opposite ends of coverage. Nothing wins on both axes, which is the entire reason waterfalls exist.
Coverage times accuracy is your real hit rate
Here is the step everyone skips. Coverage tells you how often a provider answers. Quality tells you how often that answer is right. Neither number on its own is your hit rate. The product of the two is.
A provider that answers for 90% of your list and is correct 96% of the time gives you 86.9 usable addresses per hundred, not 90.
Provider | Coverage | Quality | Actually deliverable |
|---|---|---|---|
Findymail | 90.26% | 96.26% | 86.9% |
Wiza | 85.02% | 94.72% | 80.5% |
BetterContact | 81.66% | 95.05% | 77.6% |
Dropcontact | 75.45% | 91.81% | 69.3% |
Enrow | 70.94% | 94.10% | 66.8% |
Datagma | 62.01% | 91.33% | 56.6% |
LeadMagic | 68.02% | 81.95% | 55.7% |
Hunter | 52.87% | 97.15% | 51.4% |
Icypeas | 51.90% | 93.83% | 48.7% |
When a vendor page or an agency post quotes you an 85 to 92% find rate, it is quoting coverage. The number that decides whether your campaign works is in the right-hand column.
Every provider in this benchmark bounces above the threshold
Now flip it. The addresses a provider returns that are wrong do not sit quietly in a spreadsheet. You send to them and they bounce.
If a provider is 96.26% accurate, then 3.74% of what it hands you is bad. That is your floor bounce rate before anything else goes wrong, and it is above the line almost every sending platform draws. Instantly's published guidance is to keep hard bounces at or below 2% and to auto-pause campaigns that cross it. Smartlead ships the same circuit breaker. Above 5%, you are into blacklist territory.
Provider | Bad addresses if you send raw |
|---|---|
Hunter | 2.85% |
Findymail | 3.74% |
BetterContact | 4.95% |
Wiza | 5.28% |
Enrow | 5.90% |
Icypeas | 6.17% |
Datagma | 8.67% |
FullEnrich | 12.02% |
LeadMagic | 18.05% |
Not one of them clears 2%. The most accurate provider in the entire benchmark, on Clay's own testing, still hands you a list that trips an auto-pause.
This is the finding that reframes the whole exercise. The waterfall is a coverage instrument. It fills more rows. It does nothing about the rows it fills incorrectly. Build the most accuracy-obsessed chain the benchmark allows, Hunter and Findymail in the top two slots, and the blended accuracy across everything it returns still comes out at 96.6%. That is a 3.4% bad rate. The best chain available, and it is still over the line.
Verification is not an optimization you add later. It is the step that decides whether you can send at all.
What a deliverable email actually costs
Clay's guide walks through the cost model in credits. Converting credits to dollars is the part that gets left out, and it is the number you actually budget against.
Clay meters two things separately. Actions cover platform orchestration and run around four tenths of a cent each. Data credits buy the actual lookups, and those are what a waterfall spends. The published credit ladder prices them at $0.05 each at 2,500 a month, $0.046 at 10,000, and $0.0425 at 50,000, with the annual tiers running a little lower again. The docs put a fully enriched record at 6 to 20 credits once you add company data, phone and AI columns.
For work email alone, the arithmetic is small. Lead with Findymail, back it with Hunter, Wiza and Enrow, and 100 records costs about 51 credits, or $2.55 at the entry tier. Findymail alone fills 90 of the 100. Hunter picks up half of the 10 that fall through. The last two providers exist for the handful nobody else could match.
Email enrichment in Clay is cheaper than most of the internet believes. The cost problem is not the credits. It is what you spend on top of them, and what you get back.
Run three chains across 2,000 records with a realistic overlap penalty, verification at half a cent an address, and a verifier that catches 80% of the bad ones:
Quality first | Max accuracy | Cheapest first | |
|---|---|---|---|
Chain leads with | Findymail | Hunter | Icypeas |
Addresses returned | 1,967 | 1,929 | 1,904 |
Blended accuracy | 96.2% | 96.6% | 94.3% |
Bounce rate, unverified | 3.8% | 3.4% | 5.7% |
Bounce rate, verified | 0.8% | 0.7% | 1.2% |
People you can reach | 1,893 | 1,864 | 1,796 |
Total spend | $60 | $61 | $34 |
Three things fall out of that table and none of them is what the category tells you.
Lead with quality at volume, not with the highest quality score. Hunter is the most accurate provider in the benchmark and the wrong pick for slot one, because it fills barely half the list and pushes the other 47% down into more expensive lookups. Findymail is a point less accurate and covers nine in ten, so putting it first reaches 29 more people, costs a dollar less, and lands within half a point on blended accuracy. The top slot is a coverage decision made among providers you already trust.
Ordering changes your bill, not whether you can send. The gap between the most accurate chain here and the cheapest one is 45% of the budget and 2.3 points of raw bounce. Both are over 2%. The ordering debate that every guide runs is a procurement argument dressed up as a deliverability one.
Verification is about a quarter of your spend and all of your risk. On the quality-first run, $50 goes to enrichment and $10 to verification. That last $10 is the only part of the budget deciding whether a campaign sends or an inbox gets auto-paused, and it is the line most teams cut first.
Change the chain, the list size, or switch verification off and watch what happens:
Catch-all is where the model breaks
Everything above assumes verification does its job. On a meaningful slice of any B2B list, it cannot.
A catch-all domain accepts mail for any address, real or invented, so an SMTP probe returns the same answer either way. The verifier has nothing to work with and returns "unknown" or "risky." BounceZero's analysis of 10.2 million verifications found 33.1% of checks land on catch-all domains, and that 12.3% of all addresses checked were flat-out invalid.
The follow-up study is the more useful one, and it corrects a number the rest of the industry keeps repeating. Nearly two-thirds of that 33.1% comes from two consumer providers that refuse to answer SMTP probes at all, not from businesses configuring accept-all. On mail you can actually probe cleanly, true catch-all is about 1.4%; on Microsoft 365 and Google Workspace domains it is 8.2%. The two situations look identical to a verifier and need opposite handling.
The practical consequence for a Clay waterfall: your verified list still carries a residual risk tier. Do not fold catch-all results into your main sequence and do not delete them either. Segment them, send at low volume from a separate mailbox, and watch the bounce rate on that segment on its own.
How to build the chain, and what to measure
Start with one field and 500 to 1,000 records that look like accounts you actually sell to, including the messy ones. Anything smaller and you are reading noise.
Add each provider as its own enrichment column and set a run condition so it only fires when the column above it is empty. That conditional is the waterfall; without it you are paying every provider for every row. Then add two columns most people skip: a source attribution column recording which provider landed each match, and a verification column at the end of the chain.
Source attribution is what turns the chain into something you can optimize. If your third provider is landing 3% of your list, it is not earning its slot. If it is landing 20%, move it up. Without that column you are guessing.
Then measure the only two numbers that matter. Your deliverable rate is confirmed-good addresses divided by records entering the waterfall, not addresses returned. Your cost per deliverable email is total credit spend plus verification spend divided by that same number. Both will be worse than the figure on any vendor page, and both are the ones your pipeline actually runs on.
Finally, decide where the clean records land. Enrichment that ends in a CSV decays from the moment it is exported, so the chain should write back into the system your reps work out of. The Attio and Clay integration covers that side of the pipe.
Where this approach has limits
Every number here is North American. That is the right cut for most seed to Series B teams selling into the US, and the wrong one for everybody else. Coverage moves hard across regions: a provider that fills nine in ten US contacts can fall well short of that on European or APAC lists, where both data availability and privacy regimes work against you. Clay publishes each regional cut separately. If your ICP sits outside North America, pull yours and rebuild the order from it, because the provider that belongs in your top slot may not be the one that belongs in this one.
The chain model here also assumes each provider's coverage applies independently to whatever fell through. It does not, quite. Records that survive three providers are hard for a reason, and later steps underperform their benchmark numbers. The calculator handles this with an explicit overlap penalty rather than hiding it, but treat any modeled coverage above the mid-90s as optimistic.
And accuracy is perishable in a way no benchmark captures. Contact data decays as people change jobs, so an address that was correct in October is not necessarily correct today. That is an argument for running enrichment as always-on infrastructure on a refresh schedule rather than as a one-off before each campaign, which is exactly what Clay is built to do well. The platform is not the constraint here. The default configuration is.
Frequently asked questions
What is clay waterfall enrichment?
Clay waterfall enrichment queries multiple data providers in a fixed order and keeps the first result that clears your confidence threshold. A record only falls through to the next provider when the one before it returns nothing, so you reach the coverage of the whole stack while paying for a fraction of the lookups.
How much does clay waterfall cost?
A quality-led work email chain runs roughly 51 data credits per 100 records, about $2.55 at Clay's published credit pricing, because only successful lookups are billed. Add verification and expect a real cost per deliverable email of about three cents. Clay plans start at $185 a month; verify current pricing before you buy.
What find rate should I expect from waterfall enrichment?
Expect 90 to 95% coverage on a well-targeted North American B2B list, but 86 to 90% deliverable once you account for accuracy. Vendor find rates quote coverage, not correctness. Multiply a provider's coverage by its quality score to get the number your campaign runs on. European and APAC lists run lower.
What are the best waterfall enrichment tools?
Clay is the most flexible option because conditional enrichment logic takes seconds to configure and it publishes provider benchmarks. Apollo and ZoomInfo bundle waterfall logic into their own platforms with less control over the chain. The right answer depends on whether you want to own the provider order.
What is waterfall enrichment on Apollo?
Apollo's waterfall pulls from its own database plus third-party sources to fill emails and phone numbers on net-new and saved contacts. It is simpler to switch on than a Clay chain and gives you less say over provider order and confidence thresholds, which matters most when deliverability is the constraint.
Do I still need email verification if I run a waterfall?
Yes. A waterfall raises coverage; it does nothing about the addresses it fills incorrectly. On Clay's benchmark every provider returns a list that bounces above 2% unverified, and even a chain built from the two most accurate providers blends to 96.6%, a 3.4% bad rate. Verification is what gets you under the threshold.
Sparsh Gupta, Founder of Automation Jinn and an Official Attio Expert Partner, builds enrichment and CRM systems for seed to Series B B2B SaaS teams. If you want your waterfall audited against your own list instead of someone else's benchmark, book a discovery call.
