What Is Waterfall Enrichment?

Each row runs down the list of providers until one answers. The first provider is asked for a match; if it returns nothing, the row falls through to the second, then the third, and so on until either a field comes back or the list is exhausted. Rows that matched at step one never reach step two.

That stop-at-first-match rule is what separates a waterfall from simply querying everyone. Querying every provider for every row costs the full price on all of them and then needs a rule to reconcile conflicting answers, which the waterfall order has already decided.

Because coverage is uneven in predictable ways rather than randomly. Providers assemble data from different sources, so one is stronger on a region, another on direct dials, another on small private companies. The rows a provider misses are not a random sample of your list, they are a segment.

A waterfall converts that unevenness into coverage. Each additional provider adds only the rows the ones above it could not match, so the gain per provider shrinks quickly, which is also why a waterfall of two or three well-chosen sources usually captures most of the available lift.

Cost and accuracy, in that order of effect. Because the first match wins, put the provider that is cheapest per successful match and most accurate for your segment at the top, and the expensive specialist last so it is only asked about rows nobody else could resolve.

The order is also your conflict resolution rule, which is easy to miss. If two providers hold different job titles for the same person, the one higher in the waterfall is the one you keep, so ordering by which source you trust for a field matters as much as ordering by price.