Waterfall Enrichment: How It Works and When It Pays Off

Picture a list of 1,000 contacts that need email addresses. The workflow sends each one to the first provider. Contacts that come back with a verified address leave the process. The rest go to the second provider, and so on until every contact is either filled or has exhausted the chain.

The key parts are the order of providers, the rule for what counts as a valid result, and a stop rule so you don't pay twice for the same field. A verification step between providers checks each returned address, because a provider returning an unverified guess would end the waterfall too early.

Say provider A fills 600 of the 1,000, provider B fills 150 of the remaining 400, and provider C fills 60 of the remaining 250. In this example three providers reach 810 contacts, and the third one added the fewest new results. That pattern is normal: each extra provider adds fewer new results at a higher cost per result.

Order by a mix of cost, accuracy and fit for your audience:

• Put the cheapest provider with a decent fill rate first, so the most contacts leave the chain early.
• Move a provider up if it's strong for a specific region or seniority you target.
• Put a costlier, higher-accuracy provider later only if it's reserved for gaps in high-value accounts.
• Keep a provider out of the chain if you can't explain where its data comes from, since you inherit any consent problem.

Test on a sample of your own contacts, not the vendor's demo list, and re-test every few months, because provider coverage shifts. Apollo and Cognism are examples of databases teams place in waterfalls, often for different regions. The three-way comparison covers how the options stack up.

Follow these steps:

1. Define the field you want and what "valid" means, for example an email that passes verification, or a phone number that was confirmed within a recent period.
2. Choose two or three providers and order them using your sample test.
3. Add a verification step after every provider, not only at the end.
4. Set a stop rule: as soon as a valid value exists, skip the remaining providers.
5. Write the result and the provider that supplied it back to your CRM, so you can see fill rate by source.
6. Cap spend per record or per run, so a bad list doesn't drain your credits.
7. Schedule a refresh for old records instead of re-running everything.

You can build this in a workflow tool, in Salesforce flows or with a custom LLM-based step for hard-to-find fields. Treat anything an AI step finds as unverified until it passes the same checks.

Skip it when your list is small enough to check by hand, when one provider already fills most of your contacts, or when your team can't yet act on the extra data. A waterfall adds complexity, since you now have several vendors, several contracts and several data sources to vet.

There's also a compliance side. Every provider in the chain is another source of personal data, and you're responsible for how you use it. If you sell into the UK or EU, get privacy advice before adding a provider you haven't reviewed.

Cost is the other check. Divide what you pay by the number of verified contacts you gain, and compare that to the value of a meeting from those contacts. If the third provider costs more per added contact than a meeting is worth, drop it.

Track four numbers by provider and by segment:

• Fill rate: the share of contacts each provider returned a value for.
• Verified rate: the share of returned values that passed verification.
• Cost per verified contact, including the credits spent on failed lookups.
• Downstream results: bounce rate on emails sent and connect rate on calls.

Downstream results matter most. A provider can have a high fill rate and still deliver stale data that bounces, which hurts your sending reputation. For the wider process of building and cleaning lists, see the CRO's data enrichment and prospecting playbook and the ZoomInfo, Cognism and Apollo comparison. Phone data adds its own wrinkles, covered in local presence dialing.