Hlib Storchak · 2026-07-03 · ~9 min readQuick answer
A waterfall is a sequence of email finding and verification providers, ordered so each one only runs on the contacts the previous one missed. You stop paying full price for every provider on every contact, and your find rate climbs because different providers hold different data. Order matters more than any single tool. Get the sequence right and you lift coverage without lifting spend.
Why one provider is never enough
Every email finding tool sits on top of a different data source. Some crawl public web pages, some buy from data brokers, some infer patterns from known company domains, some hold user-submitted contact lists. No single one of them sees the whole market. I have run the same list of a thousand contacts through two providers separately and watched the overlap come back under sixty percent. That gap is not a bug in either tool, it is just where their data does not reach.
If you only use one provider, you are not missing a few edge cases. You are missing whatever that provider structurally cannot see, on every single list, forever. A waterfall exists to close that gap without doubling your bill.
What a waterfall actually is
The idea is simple and the execution is where people get sloppy. You take your list of target contacts and run it through provider one. Whatever provider one finds and verifies, you keep and stop there. Whatever it misses, and only that subset, you send to provider two. Repeat with provider three, four, sometimes five, each one only ever touching what is left over.
The result is a stacked find rate that no single provider could hit alone, and a cost that scales with the size of the miss, not the size of the full list. You are not buying five lookups per contact. You are buying one lookup for most contacts and five only for the handful that are genuinely hard to find.
The core rule. Never run a contact through provider two if provider one already found and verified it. That single rule is the entire economic argument for waterfalls. Skip it and you are just paying for redundant data on the easy contacts while still missing the hard ones.
Step 1: audit what each provider is actually good at
Before you build any sequence, run a test batch, a few hundred real contacts from your actual ICP, through every provider you are considering. Do not trust marketing claims about find rates, they are measured on the vendor's own friendly sample. Measure on your list, your industries, your seniority levels.
Track three numbers per provider: raw find rate, verified deliverable rate, and price per verified find. A provider that finds a lot of emails but bounces half of them is worse than it looks. A provider that is expensive per lookup but has a high verified rate can still be cheap per verified contact.
| Provider type | Typically strong at | Typically weak at |
|---|---|---|
| Broad data broker (Apollo, ZoomInfo style) | Large enterprise, US-heavy coverage | Small teams, non-US SMB |
| Pattern-inference finder (RocketReach, Findymail style) | Known domains with a guessable email format | Companies with irregular or obfuscated formats |
| Waterfall-native enrichment (Clay, LeadMagic style) | Stacking multiple sources automatically | Cost if you do not gate the sequence properly |
| Verification-only tools | Confirming deliverability before you send | Finding new contacts at all |
Step 2: order the sequence by cost and coverage, not habit
Put your cheapest, broadest provider first. It should clear the easy majority of your list at the lowest price per find. Put your most expensive or narrowest provider last, because by the time a contact reaches it, it is already a hard case and you are paying a premium to close the gap, on a much smaller remaining list.
Do not default to whichever tool you personally like best or whichever one your rep pushed hardest. Order by the numbers from your audit. I have seen teams put an expensive specialist tool first out of habit and burn budget on contacts a five-dollar tool would have caught anyway.
Step 3: gate every step on verification, not just a find
A found email is not a usable email. Every step in the waterfall needs its own verification check before you decide whether to pass the contact forward or stop. If provider one returns an email but it fails verification, that contact should still fall through to provider two, it is effectively a miss.
This is where sloppy waterfalls quietly fail. Teams check for the presence of an email address and stop there, then wonder why bounce rates climb. Verify at every gate, not just once at the very end.
Verification should also catch role and format changes
Beyond simple deliverability, check that the email actually maps to the person and title you targeted, not a generic inbox or an old employer. Job changes are constant, and a technically deliverable email that reaches the wrong person is worse than no email, because it burns a touch on someone who cannot buy.
Step 4: build the sequence in your enrichment tool, not by hand
Doing this manually across spreadsheets does not scale past a couple hundred contacts. Tools built for waterfall enrichment let you define the provider order, the verification gate at each step, and the fallback logic once, then run it on any list size. Clay is the one I reach for most, because the waterfall logic is native to how it works rather than bolted on.
Whatever tool you use, the requirement is the same. It has to only spend on the remaining unfound contacts at each step, and it has to log which provider actually found each contact, because that log is what feeds your next audit.
Step 5: re-audit every quarter
Provider coverage is not static. Data providers buy each other, deprecate sources, and expand into new regions. A sequence that was optimal in January can be paying a premium for a provider that quietly improved, or missing gains from one that quietly declined. Rerun the same audit process every few months on a fresh sample and reorder if the numbers moved.
Waterfall vs single provider, side by side
| Element | Single provider | Waterfall |
|---|---|---|
| Find rate on a mixed ICP list | Capped by that provider's coverage | Stacks across providers, consistently higher |
| Cost structure | Flat per contact | Cheap first pass, premium only on hard remainder |
| Setup effort | Low, one integration | Higher, needs sequencing and gating logic |
| Failure mode | Silent gap you never see | Visible remainder you can decide to chase or drop |
| Best for | Small, narrow, single-region lists | Larger or mixed-region ICP lists |
When a waterfall is overkill
If your ICP is a few hundred accounts in one country with a well-known corporate email format, a single good provider probably clears ninety percent of your list and the marginal gain from stacking a second one is not worth the setup time. Waterfalls earn their complexity on volume and on messy coverage, not on every list by default.
I would not build a five-step waterfall for a hundred accounts. I would for ten thousand across a dozen countries, where the coverage gaps compound fast.
Common mistakes that quietly break a waterfall
- No verification gate. Passing raw finds through as if they were confirmed, and paying for it in bounce rate later.
- Wrong order. Running the expensive specialist first, so you pay premium prices for contacts an earlier cheap step would have caught.
- Never re-auditing. Treating the sequence as fixed forever while provider coverage shifts underneath it.
- Ignoring the remainder. Not tracking which contacts fall through every step, so you lose visibility into your true coverage gap.
- Over-building for a small list. Adding steps that cost more in setup time than they return in incremental finds.
What good waterfall data feeds downstream
The point of finding more contacts is not the email address itself, it is what it unlocks. A higher find rate means a bigger sendable list from the same ICP, which means more of your targeting work actually reaches someone instead of dying at the enrichment step. It also means fewer compromises on your ICP just because a contact was hard to find. Waterfall enrichment is infrastructure work, it should feed straight into your ICP and your sending, not sit as a separate side project.
How I run this
I audit provider performance against the client's actual ICP first, never against vendor benchmarks. I sequence providers cheapest and broadest to narrowest and most expensive, gate every step on real verification, and log which provider found each contact so the sequence stays honest. Then I re-audit on a schedule, because I have booked over 2000 meetings for B2B clients and the lists that kept converting were the ones where enrichment was maintained, not set once and forgotten.
FAQ
How many providers should a waterfall have?
Most working waterfalls I build run three to four steps. Beyond that the marginal gain per additional provider drops fast while setup complexity keeps climbing. Start with two, measure the lift, add a third only if the remainder is still large enough to justify it.
Does a waterfall slow down list building?
Per-contact processing takes a little longer since some contacts pass through multiple steps, but tools built for this run the sequence automatically, so at the list level it is not meaningfully slower than a single provider, and the coverage gain is worth the small added latency.
Should verification be a separate step or built into each provider?
Either works as long as it happens at every gate, not just at the end. Some providers return a verification status inline, others need a dedicated verification tool between steps. What matters is that an unverified find never counts as a stop condition in the sequence.
Is Clay the only tool that does this?
No, several enrichment platforms support provider stacking, but Clay is the one where waterfall logic is a native building block rather than something you engineer around. That is why I reach for it most, though the concept works with any tool that lets you sequence providers and gate on verification.
What do I do with contacts that fall through every step?
Keep them visible rather than silently dropping them. Some are worth a manual look or a LinkedIn-based alternate channel, especially if the account carries a strong buying signal. Others are simply not reachable this way and should not block the rest of the campaign from going out.