glossary Systems mechanics Data, Reporting & Intelligence Systems · Revenue OS

Denominator Integrity

Denominator integrity is the question of which population a conversion ratio is computed over. A metric produced inside a single system can only count what that system was told about, so inquiries that never became a record are silently excluded. When that happens, the ratio measures the performance of the captured subset rather than the performance of the business, and it can improve while capture degrades. Restoring integrity means rebuilding the denominator from the upstream log, not the downstream record.

Definition

Every conversion ratio has two populations behind it: the events that converted, and the events that could have. The second population is assembled from whichever system produced the report, which means the report inherits that system's blind spots. An inquiry that arrived by phone and was never keyed into a CRM, a form submission that failed validation, a chat that ended before a transcript was stored — none of these are in the denominator, and none are visible as absences. Denominator integrity is the discipline of sourcing the denominator from the system furthest upstream in the path, then reconciling it forward, so that failures of capture appear as failures rather than as improvements.

Why it matters

A ratio that moves the wrong way under stress is load-bearing in the worst possible way: it reassures precisely when it should alarm. Staffing decisions, media budgets and channel comparisons are all made against conversion ratios, so a denominator that quietly excludes the failures propagates into every one of those decisions. Checking denominator integrity is cheap — an export, a join and a recount — and it is worth doing before any conclusion is drawn from a conversion metric, in any business that takes inquiries through more than one system.

Explore related entities

[CLAIM BOUNDARY] Rebuilding a denominator changes what a number describes; it does not establish what the excluded population was worth or why it was excluded. Attribution of cause requires separate evidence, and no figure on this site quantifies the gap for any business.