What is waterfall enrichment?
Waterfall enrichment queries multiple data providers in a set order, moving to the next only when the previous one returns nothing for a record. Each record is filled by the first source that has an answer, rather than by one provider for the whole file.
Why it matters
No single provider covers every region, company size and role equally. A single-source file therefore has gaps shaped by that provider's coverage rather than by anything about your market. Running sources in sequence fills those gaps without paying every provider for every row.
How a waterfall runs
- Order the providers, usually by measured accuracy and cost for the kind of record being enriched.
- Send the record to the first provider in the chain.
- If a usable value comes back, stop. The record is done, and no further provider is called or charged.
- If nothing comes back, pass the record to the next provider and repeat.
- Record which source answered, so the file carries provenance rather than an anonymous value.
Waterfall against the alternatives
| Approach | How sources are used | What you are billed for |
|---|---|---|
| Single provider | One source for the whole file | Every row attempted, matched or not |
| Waterfall | Sources in sequence, stopping at the first hit | Only the sources actually reached |
| Parallel | Every source for every row | Every source, on every row |
What to look for
- Whether provenance is kept. A value with no named source cannot be audited when it turns out to be wrong.
- Whether the order is configurable. The right first provider differs by region and by seniority.
- How conflicts are handled when two sources both match a record and disagree.
- Whether the chain stops at the first answer or at the first verified answer. Those are not the same rule.
Related terms
Data enrichment — How records are matched and filled, and what is returned when nothing matches.