Practical framework / 5 MIN READ
Give every rate a denominator you can defend
A shared name is not a shared calculation. Make the population, counting rule and time basis explicit.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources checked 2026-11-06
A conversion rate can improve because more people converted, because fewer people were counted, or because the counted population changed. A useful metric definition lets another person reproduce the number and explain which of those happened.
What you will leave with: Create a compact metric contract that makes a dashboard number reproducible and a comparison interpretable.
Use this when
- Two teams report different conversion rates for the same period.
- A rate rose while the number of successful customers fell.
This lesson governs reported measures, not the implementation of individual analytics events or campaign tags.
Write the question before choosing the division
Name the decision the measure serves: whether a trial introduction helps eligible accounts start using the product, for example. Then write the numerator in ordinary language. Does success mean any start, a completed setup, or a completed setup that remains usable? Define how retries, cancellations and repeated actions count.
Write the denominator with equal care. All visitors, eligible visitors, sessions and accounts describe different populations. If a company has several users, an account-level rate answers a different question from a user-level rate. A team should be able to explain why the chosen denominator represents the people who had the opportunity to act.
Attach scope to the metric name
Google Search Console defines CTR as clicks divided by impressions. That is a concrete platform definition, not permission to treat every number called engagement rate as comparable. For a custom metric, append scope: completed setups per eligible new account within seven days.
Specify the reporting clock, timezone, eligibility date and outcome window. Decide whether the numerator is linked to the same denominator cohort or merely occurs in the same calendar period. Record exclusions and the owner who can approve a definition change. A short name may appear on the chart, but the full definition should remain accessible beside it.
Check the components before celebrating the rate
Display numerator and denominator next to the percentage. Ask whether their changes are operationally plausible. A broken eligibility filter can remove difficult customers and manufacture improvement without changing anyone’s experience. A new source of visitors can lower a rate while adding valuable customers.
Recalculate a small, de-identified example by hand and compare it with the report. Include a repeat action, an excluded record and an outcome near the time boundary. This exercise checks the interpretation of the calculation, not just whether a query runs. Keep the test examples so future revisions can be checked against the same expectations.
Version the definition without rewriting history
When the business question changes, create a new version with an effective date. Explain why the old and new series differ and, where feasible, show both during a transition. Do not silently stitch them into a single trend line. Historical recomputation is useful only when the underlying data supports the revised definition.
In review meetings, distinguish a result change from a definition change. The owner should record what changed, what previous decisions may need revisiting and where the replacement metric appears. Consistent definitions reduce arguments about arithmetic and leave room for the more important argument about what the organization should do.
Make the reasoning concrete.
A software team counts 120 completed setups among 600 eligible new accounts. A second report divides the same 120 by 1,000 visiting users. These illustrative totals describe a 20% account rate and a 12% user-based ratio, not contradictory arithmetic.
Account decision
Opening: Use the 600 eligible accounts.
Hypothesis: This denominator matches the onboarding decision.
Risk: Account identity and eligibility still need validation.
Traffic decision
Opening: Report visiting users separately.
Hypothesis: A traffic measure can diagnose acquisition composition.
Risk: Combining user counts with account outcomes can confuse interpretation.
Choose the account metric for this decision, preserve the user traffic measure as context, and label both clearly rather than averaging the percentages.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Decision
What choice will this number change?
Numerator
What counts once, and what never counts?
Denominator
Who had a genuine opportunity to succeed?
Clock
Which start date, timezone and outcome window apply?
Version
Who owns the definition and how will changes be marked?
Watch for the wrong conclusion.
- Dividing account outcomes by users without explaining the units.
- Hiding counts behind a percentage.
- Changing eligibility after seeing the result.
- Comparing calendar outcomes with an acquisition cohort as if they were identical.
Your next action
Choose one disputed metric and write its numerator, denominator, counting unit and clock in a sentence everyone can reproduce.
Sources & scope
Frameworks, worksheets and hypothetical examples are original editorial synthesis. Linked primary sources support only the stated concepts. Source checking is not an independent audit of effectiveness. No named human author or reviewer is claimed.
- Google Search Console Help: Performance report overview →Primary platform documentation · Supports: Definitions of clicks, impressions, CTR and average position. Supports these Google Search measures only; the editorial measurement workflow is not a Google recommendation.