Define the measure before comparing it
A metric is interpretable only when its population, time period, collection method, and counting rule are clear. Use the same definition wherever a measure appears, and say when a channel cannot provide comparable evidence.
For a proposed measure, record the question it answers, its evidence source, who owns that source, and what it cannot establish.
Reach, engagement, and outcome are not interchangeable
Use each evidence type to answer the question it can support.
| Details | Reach | Engagement |
|---|---|---|
| Question answered | Was the intended audience reached or offered access? | Did people respond or interact? |
| Possible evidence | Defined distribution record, attendance record, or access evidence. | A reply, question, attendance interaction, or click where captured reliably. |
| Interpretation check | State the intended audience, denominator, channel, and reporting period. | Define what counts as an interaction and how duplicate activity is treated. |
| Common limitation | Delivery or access does not prove a person read, understood, or acted on content. | An interaction does not by itself demonstrate understanding, behavior change, or an outcome. |
| Outcome evidence | Assess separately against the stated objective using suitable evidence; avoid attributing a change to communication alone without support. | Assess separately against the stated objective using suitable evidence; avoid attributing a change to communication alone without support. |
Questions that improve comparability
Resolve these points before interpreting a scorecard.
What denominator should I use?
Name the audience population relevant to the objective and the period. Explain whether the evidence represents that population, a subset, or only people with recorded activity.
Can audience counts be added across channels?
Not safely unless overlap is known and the counting rules are compatible. The same person may appear in more than one channel's evidence.
Can qualitative feedback be reported as evidence?
Yes. Describe what was observed, how it was collected, and its scope. Treat it as a signal or explanation to investigate, not as a population-wide result unless the method supports that claim.
When is a channel-to-channel comparison fair?
Only when the audience, period, metric definition, and collection method are sufficiently consistent. Otherwise present the measures separately and explain the difference.