What is waterfall enrichment?

Waterfall enrichment fills an empty field by trying data sources in order and stopping at the first trustworthy answer. The definition, a worked example, why it beats a single source, and the two rules - provenance and pay-on-answer - that keep it honest.

GuideBY THE ASTROFABRIC TEAM · SEP 1, 2026 · 4 MIN READ

Waterfall enrichment is a method for filling an empty data field by querying sources in a ranked order and stopping at the first one that returns a value above a confidence bar. If the first source has no answer, the request falls to the second, then the third, until the field fills or the sources run out - which is where the name comes from. Each field runs its own waterfall, so a row is enriched field by field, and the source that answered is recorded on the value.

The definition

Three parts make a waterfall. A source order for the field, ranked by measured accuracy first, freshness second and cost third. A confidence bar that decides whether a returned value counts as an answer - an observed value passes, an estimate may need a second source to agree. And a stop rule: the run ends the moment the bar is met, so a field that fills on the first source never spends on the second. The complete operating manual, including how to order sources per segment and how an agent plans the waterfall per row, is AI agents for waterfall data enrichment.

A worked example

ONE ROW, THREE FIELDS, THREE WATERFALLS
FieldSource 1Source 2Source 3Result
Employee countEstimate: 180 (below bar alone)Estimate: 195 (agrees within tolerance)Skipped190, corroborated, two sources, dated
Verified emailNo answerPattern guess (below bar)Found and verified deliverableFilled on source 3, verified, dated
Commerce platformDetected on the live siteSkippedSkippedFilled on source 1, observed, dated

Notice what the row cost: one corroborated estimate, one verified email and one site detection - five calls that answered, and the misses on the way to the email were part of finding it. A pay-per-attempt model would charge for every call; a pay-on-answer model charges for the answers.

Why it beats a single source

Every data provider's coverage has a shape - strong in one geography, thin in another, rich for one role family and sparse for the next - so on a list that spans segments, the best single source still leaves a large share of fields empty, and the empty share concentrates in exactly the companies that were hard to find in the first place. A well-ordered waterfall roughly doubles the fill rate of its best single source. It also surfaces disagreement between sources, which is information: a value two sources contest deserves lower confidence than one they agree on.

The two rules that keep it honest

Provenance on every value. The source, the date and the confidence travel with the field, so a reader can tell an observed value from an estimate and the next run can tell whether the value is fresh enough to keep. Without it, a waterfall is a guessing machine with extra steps. Pay when a source answers. A field that fills on the fifth source should cost one answer rather than five lookups, and a field that never fills should cost close to nothing; otherwise the waterfall punishes exactly the hard-to-find rows it was built to fill. This is how AstroFabric's enrichment agent runs: source by source until the field is filled, provenance recorded, charged only when a source answers, and the row handed to verification before it is delivered.

Frequently asked questions

What is waterfall enrichment?

A method for filling an empty field by querying data sources in a ranked order and stopping at the first value that clears a confidence bar. Each field runs its own waterfall, and the source that answered is recorded on the value.

What is the difference between waterfall enrichment and regular enrichment?

Regular enrichment asks one source and accepts whatever comes back, including nothing. Waterfall enrichment falls through a ranked sequence of sources per field until one answers above the bar, which roughly doubles fill rates on a mixed list and records where every value came from.

What is a waterfall in the context of B2B data?

The ranked sequence of providers a field falls through - first choice, second, third - ordered by accuracy, freshness and cost for that field in that segment, with a stop rule at the first trustworthy answer.

How is waterfall enrichment priced?

The honest model charges only when a source answers, so misses on the way to a filled field cost nothing and unfillable rows cost close to nothing. Ask any provider whether a lookup that returns nothing is charged, and whether the spend ceiling is enforced before the call.

Sources

⟨ RUN IT INSTEAD OF READING IT ⟩

Every playbook on this blog ships as a runnable mission.

Open a workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, on your plan's monthly credits.

⟨ KEEP READING ⟩
GuideEnrichment & data

AI agents for waterfall data enrichment: the complete guide

Why one data source never fills a list, how a waterfall runs field by field with provenance on every value, the ordering and conflict rules that keep it honest, and what changes when an agent plans the waterfall instead of a person.

Sep 1, 2026 · 12 min read
GuideEnrichment & data

Company and person data for AI agents: the complete guide

What an agent needs from company and person data that a human operator never asked for: typed fields, provenance, freshness and cost per call. The company families, the person families, identity resolution, the data contract, delivery and the compliance rails.

Sep 1, 2026 · 12 min read
GuideEnrichment & data

What is technographic data?

Technographic data is the record of which technologies a company uses - detected, dated and tracked over time. What it observes, how it is collected, why the changes matter more than the snapshot, and how it is used in prospecting and enrichment.

Sep 1, 2026 · 4 min read