Choose the Right Denominator for Lead Quality Metrics

“Two-thirds of the leads are good” sounds clear until someone asks which leads were counted. The answer may mean two-thirds of reviewed records, two-thirds of the original list or two-thirds of records left after inconvenient cases were removed. Those are different statements.

A denominator worksheet makes the reporting population explicit. Begin with the original candidate set, explain each transformation and preserve the unresolved cases. The objective is not to choose the largest or smallest denominator, but the one that matches the decision the metric is supposed to support.

Name the question before calculating

A review-acceptance metric asks how many completed reviews met the stated criteria. A delivery-readiness metric asks how much of the defined unique candidate set is ready for the intended next use. Review coverage asks how much of that set has received a completed review. None is a substitute for the others.

Write the numerator and denominator in words alongside the formula. Include the entity type, period and status rules. If the report mixes people and companies, resolve that mismatch before arithmetic. If it excludes records, state why they are outside the defined population rather than silently removing them after seeing the result.

Technical counting behavior can also hide missing cases. SQLite: Built-in aggregate functions distinguishes counting all rows from counting non-null values in one field. A query that counts a completion-date column can omit unfinished records by construction. Verify what your reporting tool actually counts instead of trusting a column heading.

Reconcile a fictional delivery batch

Imagine a research team receives twenty candidate rows for one service territory. This example is invented and is not a performance benchmark. Identity review finds four extra rows that duplicate accounts already represented. The approved population is therefore sixteen unique accounts, with the duplicate lineage retained.

At the reporting cutoff, twelve unique accounts have completed reviews. Eight meet the stated delivery criteria, four are unsuitable and four remain pending. The worksheet reconciles every level:

• Candidate rows received: 20.

• Duplicate extra rows removed from the account count: 4.

• Unique accounts in scope: 16.

• Completed reviews: 12, comprising 8 accepted and 4 unsuitable.

• Pending reviews: 4, still included in the unique in-scope population.

Now report the different measures: acceptance among completed reviews is 8 divided by 12, approximately 66.7%. Delivery readiness across the defined population is 8 divided by 16, or 50%. Review coverage is 12 divided by 16, or 75%. Each percentage has a valid use, but only with its complete label.

Do not let pending work improve the headline

If the team reports only 66.7%, a reader may assume most of the whole delivery is ready. The pending four could eventually be accepted, rejected or remain unresolved. Their absence from the completed-review denominator is appropriate for that specific measure, but their existence must remain visible.

Do not label pending cases unsuitable merely to eliminate uncertainty either. That would change a missing review into a negative business judgment. Preserve the distinction between “reviewed and does not fit” and “not yet established.”

Likewise, duplicates should not be treated as four separate failed accounts if the metric concerns unique-account fit. They can be reported as an input-quality issue in their own right. The same defect can affect a process metric without becoming four additional business entities.

Test denominator changes with a bridge

Suppose two pending accounts are later found to belong outside the originally agreed territory. Ask whether the initial population definition was wrong or the scope genuinely changed. Record the decision and its effect on the report. Do not retrospectively shrink the population simply because the new percentage looks more attractive.

A bridge should show the previous population, additions, justified removals and current population. Keep historical versions where needed for comparison. When scope changes, report the new measure as a revised basis rather than implying direct comparability with the old figure.

Before release, ask a colleague to reconstruct the counts from the row statuses and identity references. The worksheet passes when all original rows have an explained destination, each percentage answers one named question and pending work remains discoverable. That is a stronger quality statement than a polished percentage with an invisible denominator.

Continue with Measure Sales Navigator Prospecting with a Small, Honest Funnel or Compare Two Target Segments with a Fair Sales Experiment.