Glossary · Data and enrichment

What is waterfall enrichment?

The short answer

Waterfall enrichment is a way to fill a missing field, such as a work email or a phone number, by asking several data sources one after another. The first source gets every record; each next source is asked only about the records still empty, and the waterfall stops for a record as soon as one source returns a usable value.

Updated
Read
4 min
Written by
Antoine Blitz
On this page8 sections
  1. Real data
  2. How does waterfall enrichment work?
  3. Why does each extra source in a waterfall enrichment add less?
  4. Waterfall enrichment vs Waterfall ICP: what is the difference?
  5. What makes a waterfall enrichment reliable?
  6. How do you add BlitzAPI to a waterfall enrichment?
  7. Related terms
  8. Questions people ask

Real data

What a real waterfall returns, step by step

Five data providers, each asked for the current employer and title of the same 3,445 LinkedIn profiles in our September 2026 benchmark. The waterfall is rebuilt from their answers: each step keeps only the profiles that the steps before it had not returned.

From the B2B data provider benchmark, September 2026, run on

Benchmark

Profiles left for each step

Source 1
3,445
Source 2
330
Source 3
77
Source 4
57
Source 5
42

Each step counts only the profiles that the steps before it had left empty.

Benchmark

Records returned at each step

Source 1
3,115
Source 2
253
Source 3
20
Source 4
15
Source 5
1

A returned record is an answer, not a checked one. The report checks every answer against the live profile, step by step, and names the providers.

How does waterfall enrichment work?

A waterfall sends every record to a first source, then passes only the records still empty to the next one, and so on, until each record has a usable answer or every source has been tried. It starts with one field and one rule for what counts as an answer.

  1. Source 1Gets every recordThe records it fills leave the waterfall
  2. Source 2Gets the missesOnly the records still empty
  3. StopFirst usable answerOr the last source has been tried
Only the misses flow down: each source is asked about fewer records than the one before it.

The rule for “usable” decides more than the list of sources does. A waterfall that accepts any address it is given fills more rows than one that accepts only verified addresses, and more of those rows fail when you send to them. Decide it before you pick a single provider.

Why does each extra source in a waterfall enrichment add less?

Each extra source in a waterfall enrichment adds less because the easy records are filled first. The waterfall above shows it on real data: the first source returned a record for 3,115 of the 3,445 profiles. The four sources after it received 506 queries between them and returned 289 more. The last one was asked about 42 profiles and returned one.

Two things follow. The order matters, since whoever goes first answers most of the questions. And the tail is where the money goes: when a provider bills every request, found or not, the cost per answer tends to rise down the chain.

Returned is not the same as right. The benchmark report checks each of those answers against the live profile and shows, step by step, how many were accurate.

Waterfall enrichment vs Waterfall ICP: what is the difference?

Waterfall enrichment cascades through data sources to fill one missing field; Waterfall ICP cascades through job titles to find the right person inside one company. The names are close; the jobs are not.

Waterfall enrichment Waterfall ICP
What cascades Data sources Job titles, in priority tiers
Starts from A record with a missing field One company
Stops when A source returns a usable value The slots you asked for are full, or no tier is left
Gives you The missing field, such as an email The people to contact at that account, best tier first

Waterfall ICP is a people search that tries a ranked list of titles inside one company, for example from the ideal buyer down to the founder, until it has the number of people you asked for. The decision makers guide shows it at work.

Which is which?

Five steps from real workflows. Which kind of waterfall is each one?

  1. Ask a second provider for the phone numbers the first one did not find

  2. Search "VP Marketing", then "Head of Marketing", then "CMO" at one account

  3. Stop at the first source that returns a verified work email

  4. Fill three contact slots at one account, best title tier first

  5. Order three email providers by how many answers they return on your lists

Show the answers
  1. Ask a second provider for the phone numbers the first one did not findWaterfall enrichmentThe same field, asked of another source, only for the records still empty.
  2. Search "VP Marketing", then "Head of Marketing", then "CMO" at one accountWaterfall ICPWhat cascades is a list of titles inside one company, not a list of data sources.
  3. Stop at the first source that returns a verified work emailWaterfall enrichmentThe stop rule belongs to the field: the first usable value ends the waterfall for that record.
  4. Fill three contact slots at one account, best title tier firstWaterfall ICPThe tiers are ranked buyers; the search fills the slots tier by tier, and with one slot it stops at the first tier that finds someone.
  5. Order three email providers by how many answers they return on your listsWaterfall enrichmentOrdering the sources is the main design choice of a data waterfall.

What makes a waterfall enrichment reliable?

A waterfall enrichment is reliable when four rules are written down before the first run: one definition of a hit, the source and date stored with every value, a rule for conflicts, and checks over time.

  • One definition of a hit. A verified work email, a mobile number, a phone that rings at the person’s desk: say which, and refuse the rest.
  • Provenance on every value. Store the source and the date next to each field, so that a bad answer can be traced and a source can be dropped.
  • A rule for conflicts. When two sources disagree on the same field, the waterfall has to know which one wins, or it keeps the first answer by accident.
  • Checks over time. People change jobs and addresses stop working, so a filled field is true on the day of the call. Run it again on a schedule: see data enrichment.

How do you add BlitzAPI to a waterfall enrichment?

BlitzAPI joins a waterfall enrichment as one step, from wherever the waterfall runs: Clay, n8n, or your own code. Email Enrichment takes a profile URL and returns a verified work email, or found: false when there is none. A miss counts no record, so the step costs nothing when it finds nothing, and the next source runs only on the true misses. The integrations page shows how to call it from each tool.

  • Full definitionData enrichment

    Filling in what your records are missing. You start from what you have, such as a company website or a person's profile link, and get back the facts around it: industry, size, location, role, work email.

  • APIWaterfall ICP

    A people search that tries your list of job titles in order, inside one company, filling the number of contacts you ask for from your first choice down: ask for one and it stops at the first person it finds.

  • Short definitionEmail verification

    Checking that an email address exists and can receive mail before you write to it.

  • Short definitionWork email

    The address a person gets from their employer, on the company's own domain, rather than a personal address.

  • APIDirect dial

    A phone number that reaches one person, a mobile or a personal direct line, instead of a company switchboard.

All terms in the glossary

Sources and method

Written by
Antoine Blitz, Founder, BlitzAPI
Published
Measured on

The five sources are data providers tested in the September 2026 benchmark, queried in the order the report uses (it starts from the top of its ranking, a sorting rule rather than a recommendation). The report names them and checks every answer; this page shows the counts only.

The benchmark asked each provider for a person's current employer and title, not for an email: the counts show how a waterfall behaves, whatever the field. Each provider answered for all 3,445 profiles, and the cascade is computed from those answers.

The call shown for builders is not the source of the counts above: it is how BlitzAPI fits into a waterfall of your own. PROFILE_ID stands for any public profile URL.

Questions people ask

What is waterfall enrichment in Clay?

Clay offers waterfall columns: for one field, such as a work email, they try several providers in the order you set and stop at the first one that returns a value. You choose the providers, the order and what counts as a result. The logic is the same as on this page; Clay runs it for you, row by row.

What is waterfall enrichment in sales?

It is how sales teams reach more of the people on their lists. Instead of accepting the gaps of a single provider, they chain several: the first fills most emails or phone numbers, the next ones fill some of what is left. More contacts become reachable, but each extra source adds fewer answers and one more set of rules to maintain.

What are the downsides of waterfall enrichment?

The cost per found record rises toward the end of the chain, because later sources get the hard cases and answer fewer of them. Sources can also disagree, so you need a rule for conflicts and a note of which source supplied each value. And every extra step adds a call, which a live workflow feels.

How do you build a data waterfall?

Pick the field and decide what counts as a usable answer, for example a verified email rather than any email. Order the sources, usually by how many answers they return on lists like yours and by price. Send only the misses to each next step, store which source answered, and check the values again on a schedule.

With BlitzAPI

Put it to work

ProductEmail Enrichment APIPOST /v2/enrichment/emailBenchmarkB2B data provider benchmark, September 2026The full report, with its method, on 3,445 LinkedIn profiles.

For your team

For builders

Use BlitzAPI as one step of a waterfall. The call returns a verified work email, or found: false so that the next step knows it has to run. Send it with your API key from Clay, n8n or your own code.

Find a work email from a profile URL (one step of the waterfall)

curl -X POST https://api.blitz-api.ai/v2/enrichment/email \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"person_linkedin_url":"https://www.linkedin.com/in/PROFILE_ID"}'

API reference

Data shouldn't be
the problem

Grab. Connect. Build.