On this page
Ask five sales teams which data provider is best and you’ll get five answers. They’re all right, for their market. One provider is strong on US tech companies, another on European mid-market, another on mobile numbers. None of them is strong everywhere.
Waterfall enrichment is the answer the GTM community settled on, popularised by tools like Clay. Instead of choosing one provider, you ask several in sequence and keep the first reliable answer. Done well, it lifts coverage and cuts cost per record at the same time. Done badly, it burns credits and pollutes your CRM.
How a waterfall works
For each field you want (work email, mobile number, job title, company headcount), you define an ordered list of sources:
- Ask provider A.
- If A returns nothing, or a result that fails validation, ask provider B.
- Continue until a source returns a valid result, or the list runs out.
- Write the result to your CRM, along with which source supplied it and when.
The important part is step 2: fails validation. A waterfall that accepts the first answer, whatever its quality, just spreads bad data faster.
Designing the order
Most teams order providers by price. That’s the wrong first criterion. Order them by hit rate for your target market, then by cost.
To find the right order, run a bake-off:
- Take a sample of a few hundred records that represent your ideal customer profile. Include the regions and company sizes you actually sell to.
- Run the same sample through every provider you’re considering.
- For each provider, measure the find rate (how often it returns something) and the valid rate (how often that something passes verification).
- Rank by valid rate per unit of cost.
It’s common to find that an expensive provider earns its place first in one region and doesn’t belong in the waterfall at all in another. That’s fine. Waterfalls can, and often should, branch by geography.
Validation is not optional
Every result should pass a check before it’s accepted:
- Emails go through a verification service that checks the domain, the mail server and, where possible, the mailbox. “Catch-all” domains need their own rule: decide whether you’ll send to them, and with what volume.
- Phone numbers are normalised to an international format and checked for type (mobile, landline, VoIP).
- Job titles are matched against your persona definitions, so “Head of RevOps” and “Revenue Operations Lead” both land in the same bucket.
- Company data is cross-checked with the company’s own domain. A headcount that disagrees wildly between sources gets flagged for review instead of being written.
Bounce rates matter far beyond a single campaign. Mailbox providers judge your sending domains on them, so one batch of unverified emails can damage deliverability for weeks. We cover this in our guide to cold email infrastructure.
Enrich at the right moment
When you enrich matters as much as how.
- On creation. Inbound leads should be enriched within seconds of a form submit, so routing has what it needs. We cover this in speed-to-lead routing.
- On signal. When an account shows buying intent, enrich the buying committee then, not months earlier when the data might go stale.
- On a schedule. Job titles and companies change constantly. A quarterly refresh of active accounts keeps the CRM honest.
Enriching your entire total addressable market up front is usually a waste. Most of that data will be out of date before anyone uses it.
Make it observable
Every enriched field should carry three pieces of metadata: the source, the date, and the confidence (verified, unverified, catch-all). This makes three things possible:
- You can see each provider’s real contribution and cancel the ones that don’t earn their place.
- Reps can see how fresh a phone number is before they dial it.
- You can re-run enrichment for records that are older than your freshness threshold, and nothing else.
A simple dashboard of coverage by field, by segment, over time, is often the single most useful report a GTM team has.
Common mistakes
- Overwriting good data. A waterfall should never replace a value a rep entered by hand, or a recently verified value, with an unverified one. Write rules for which sources can overwrite which.
- No spending limits. Put a ceiling on credits per record and per day. A loop bug in an enrichment workflow can spend a month’s budget in an hour.
- Enriching junk. Filter out personal email domains, competitors, students and existing customers before you spend a credit.
- One giant workflow. Split enrichment into small, named steps. When a provider changes its API, you want to fix one step, not untangle everything.
The result
A well-built waterfall gives you higher coverage than any single provider, a lower cost per usable record, and data your reps actually trust. That trust is the real prize. Reps who trust the CRM use it, and a CRM that gets used is the foundation for everything else in GTM engineering.
We design and run enrichment pipelines for B2B teams. See our lead enrichment and scoring service.


