What is enrichment waterfall?

Also called identity waterfall, resolution waterfall.

The enrichment waterfall is the fixed-order process that resolves a signal fragment to a named person and account, working from the strongest identifier to the weakest and across data providers cheapest first and expensive credits last, with a merge log and a quarantine so that every match is auditable.

The enrichment waterfall is how the system turns confetti into signal. Sensors deliver fragments: an anonymous IP that resolved to a company, a badge scan, a username in a thread, a form fill from last spring. Until you have decided whether they belong to the same account you do not have a signal. Resolution is the first real job of the build, collapsing fragments into people and people into companies, and it runs as a deterministic waterfall, an explicit order from strongest identifier to weakest. A LinkedIn profile URL wins when present, as the most stable handle on a person. Failing that, a verified email. Failing that, the company domain, which anchors the fragment to the right account even when you cannot pin the individual. A fixed order means the same inputs always resolve the same way.

The same ladder runs across providers, cheapest first, expensive credits last, with caching. We had been burning enrichment credits the expensive way, paying a premium per lookup through one provider's native integration, until David built a break-even calculator to decide, lookup by lookup, when to use our own API keys across a waterfall of providers and when to fall back to the pricier credits. Never pay twice for a record you already enriched, and never re-pay to enrich an account you just decided was not a fit.

Matching is probabilistic, so identity is a guess with a confidence attached. When two fragments merge, the merge is written to a log so it can be seen and reversed. When confidence is low the match is quarantined and waits for a second piece of evidence; it never silently becomes outreach. And because signals roll up to the account, one wrong individual does not break the read on a company that is clearly in motion.

intelligence layerThe intelligence layer is the shared, account-level store of the Demand Compass system that every sensor reports into and every motion reads from, holding the resolved signals, the scores and the merge log for each account; it is the one component of the stack a company never outsources.
Signal RadarThe Signal Radar is the instrument of the Demand Compass that reads every account's activity across both axes, the awareness signals and the readiness signals that never become a form fill, and feeds what it catches into the Market Map, so that "who is in the market this week" becomes a standing answer instead of a guess.
The Signal-Signature GapThe Signal-Signature Gap is the pattern in considered B2B deals where the person who emits the buying signal, the engineer reading your docs or testing the free tier, is rarely the person who signs, so a funnel that waits for the decision-maker to raise a hand waits for something that never happens.
TAM coverageTAM coverage is the share of a company's total addressable market that its demand engine actually touches, mapped as named accounts with signals attached, as opposed to the fraction a team reaches when lists are pulled by hand and the rest of the market is never mapped at all.
signalA signal is a dated piece of evidence that an account did something on one of the two Demand Compass axes: an awareness signal shows the account knows you, a readiness signal shows the account is in motion now, and every signal is evidence with a half-life, so it decays instead of counting forever.

From The Demand Compass, by David Moreira and Marcos Stubrin. All terms · The book