Audit the Data Relationships Behind Attribution

The decision to make

Attribution depends on a chain of relationships: event to participant, participant to account, account to opportunity, and opportunity to the measured outcome. A clean chart cannot repair a broken link. Missing associations may systematically exclude certain seller workflows or regions, so an overall completeness percentage can conceal bias. Quality work should start from the business population that ought to be represented, then explain every included, excluded, and unattributed record. The audit must distinguish absent data from present but incorrect data.

Build the practical approach

Create a dated eligible opportunity base from the CRM and reconcile it to the reporting output. Validate stable IDs, contact-company links, participant-opportunity links, campaign membership states, event time, ingestion time, outcome dates, amount basis, and reporting currency. Define required fields for each report type rather than applying one completeness rule to all analyses. Distinguish missing, stale, ambiguous, duplicate, and structurally invalid records. Sample included and excluded opportunities across regions, deal types, and seller workflows; inspect relationship evidence as well as field presence. Retain association effective dates so current contact relationships do not silently rewrite past interactions. Test campaign links at creation and after redirects, preserving agreed case-sensitive names and campaign IDs. Track late-arriving events and define whether they revise previous reporting snapshots or enter a correction ledger. Maintain an exception queue with owner, due date, observed impact, and correction reason. Publish material coverage limitations beside the report before interpreting differences, then prioritize fixes by affected decisions and record value rather than merely by the number of missing fields.

HubSpot requires deal Amount, Create date, Close date, and an associated contact for deal revenue attribution. HubSpot: Attribution report definitions.

The practical workflow

1. Fix the eligible CRM base
2. Test identity relationships
3. Reconcile included amounts
4. Inspect excluded records
5. Own the exception queue

Compare the approaches

Work through an illustrative scenario

Illustrative scenario: two regions show different attributed revenue despite similar eligible pipeline. One region logs meetings against opportunities but frequently omits participant associations required by the report. Another attaches a generic administrative contact to every deal, which appears complete but is not reliable stakeholder evidence. The analyst must decide whether backfilling any contact will solve the gap. It would improve a superficial completeness measure while weakening validity. The team verifies meeting participants where evidence exists, leaves unresolved deals unattributed, and changes the recording workflow. It labels the affected coverage by region and postpones campaign-efficiency comparisons that depend on those relationships until the discrepancy is understood.

Measure whether the work is useful

Define eligible-base reconciliation as included opportunity value plus excluded value plus explicitly unattributed value, with mutually exclusive treatments, compared with the CRM base under the same amount and currency rules. Association completeness is opportunities with required links divided by eligible opportunities; association validity is audited links supported by evidence divided by audited links. Campaign match rate is qualifying events assigned to a recognized campaign ID divided by qualifying events expected to carry that ID. Duplicate rate uses records failing the agreed uniqueness key divided by evaluated records. Late-arrival rate is events ingested after the reporting cutoff divided by relevant events. Report audited sample sizes, exception age, and coverage by segment. Track whether corrected records change a material program decision.

Google Analytics UTM values are case-sensitive, and missing relevant UTM parameters can appear as '(not set)' in reporting. Google Analytics: Campaign URL parameters.

Avoid the common failure points

Backfilling can invent historical relationships when effective dates are ignored. A populated field may contain a placeholder or wrong contact, so completeness is not validity. Duplicate rows can inflate both events and opportunity amounts. UTM consistency cannot recover an unlogged conversation. Currency conversion changes can masquerade as pipeline movement if the rate basis is unstated. Do not remove unexplained records to force reconciliation. Preserve raw evidence and correction provenance, and label estimated or unresolved associations. A downstream model change should not be used to conceal an upstream measurement failure.

Your next-action checklist

• Completeness audit: Inspect linked records. Check the limitation: cannot prove correct associations.
• Relationship audit: Sample included and excluded deals. Check the limitation: requires record-level review.
• Tagging validation: Test redirects and naming. Check the limitation: misses offline activity.

Use the comparison to choose a bounded next step. Record the evidence, the responsible owner, and the review decision before extending the play to additional accounts.

Questions this guide answers