Waterfall enrichment: build for usable records

Waterfall enrichment queries data sources in a defined sequence until a record satisfies an acceptance rule. It can improve coverage and control cost, but only when the workflow defines required fields, valid values, provenance and failure handling. More sources do not automatically create better data.

In outbound, enrichment turns a thin account or person record into something the system can qualify, route and contact. The important output is a usable record for a specific play, not the maximum number of populated columns.

Start with a field contract

For every enriched field, document the business purpose, accepted format, freshness requirement, confidence rule, fallback order and writeback destination. Different fields deserve different treatment.

The contract keeps the workflow tool-agnostic. Sources can change while the business requirement remains stable.

The workflow

1. Normalize the input. Resolve obvious formatting differences and check for an existing canonical account or person before spending on enrichment.
2. Check internal truth. Reuse current, accepted CRM data where appropriate. Preserve commercial ownership and suppression states.
3. Query the first source. Use the source that best balances the field requirement, coverage and operating constraints.
4. Validate the response. Test format, identity, freshness and confidence. A non-empty value can still fail the contract.
5. Continue on failure. Query the next source only when the value is missing or unacceptable.
6. Resolve conflicts. Apply explicit precedence rules or route material disagreements to review.
7. Write back with provenance. Store the accepted value, source, checked time, confidence and workflow version.
8. Track the gap. Mark unresolved fields and decide whether the record can proceed, wait or be rejected.

Choose source order by field

A single global provider order is convenient but often weak. The best first source for corporate identity may differ from the best source for a person's current role or contact path. Design the waterfall at field level, then batch requests where it improves the operation.

The Waterfall Enrichment Yield Calculator shows the marginal coverage and cost contributed by each successive source. When evaluating the first data layer, compare the database-led workflow in Prospeo vs Apollo and the email-focused trade-offs in Prospeo vs Hunter.

Enrich after qualification where possible

Use inexpensive account-level evidence to qualify the market before buying or researching every contact. The TAM mapping workflow separates account qualification from buying-committee coverage for this reason.

Some fields are needed to qualify, so the order is not absolute. The practical rule is to avoid expensive person-level enrichment until the account passes the gates that can be checked earlier.

Preserve missing and conflicting data

Do not turn “unknown” into “false.” A missing technology field may mean no source found evidence, not that the company lacks the technology. Keep field status separate from value and record which sources were checked.

When sources disagree, compare observation dates and entity scope. A parent company and local subsidiary can both have correct but different attributes. Send unresolved identity and qualification conflicts to a review queue.

Write back safely

Define which system owns each field and whether enrichment may fill blanks, refresh stale values or overwrite accepted data. Keep raw provider responses outside core CRM fields when they are useful for audit but not ready as commercial truth.

The accepted record should connect to the shared GTM brain through definitions and versions. When a field helps activate a play, preserve its value at activation time so attribution can explain the decision later.

Monitor the workflow

• Coverage and acceptance by field, segment and source
• Conflict and manual-review volume
• Cost per accepted record rather than per returned value
• Age of fields with freshness requirements
• Writeback errors and records blocked from activation
• Downstream rejection caused by incorrect or missing data

These measures diagnose the data system. Qualified meetings and pipeline still decide whether the wider play is commercially useful.

Test the waterfall before production

Build a review set from the segments and geographies the play will target. Include incomplete records, subsidiaries, common company names, recent job changes and known conflicts. Run each field through the proposed source order and inspect why the workflow stopped.

For accepted values, verify identity and evidence. For rejected values, confirm that the next source was queried correctly. For unresolved records, check that the workflow produced the intended review or hold state. A test set should exercise disagreement and absence, since clean matches reveal little about fallback behavior.

After launch, keep a small sample for manual review and add real failure cases to it. When a provider or rule changes, rerun the same set. This turns enrichment quality into a maintained operating standard instead of a one-time setup decision.