Start With the Outcome: Choose Campaign Metrics Before You Buy Media
Start With the Outcome: Choose Campaign Metrics Before You Buy Media
A campaign report can contain far more numbers than a team can act on. Impressions, clicks, completed views, form submissions, purchases, and cost measures may all be available, but availability is not the same as importance. A useful measurement plan begins before a media placement or campaign is selected: it starts with a clear description of the outcome the organization wants to observe.
This is a planning framework, not a prediction. A metric cannot guarantee a business result, and the appropriate measures depend on the organization, its offer, its audience, its consent practices, and the way it defines a valuable action. The point is to make those choices visible before a report arrives.
Describe the outcome in plain language
Start with one sentence that describes the change the campaign is meant to support. For example: “Help people discover a new resource,” “invite qualified readers to request more information,” or “make it easier for returning customers to complete a purchase.” Avoid beginning with a channel or a metric. “Get more clicks” describes an activity; it does not explain why the activity matters.
Google’s advertising guidance similarly frames measurement around the advertiser’s stated goal. Its examples distinguish traffic, conversions, brand awareness, and return on investment, then describe different measurements that may be relevant to each. That distinction is useful even when a campaign does not run on Google: first state the intended outcome, then decide which observable signals are relevant to it.
Separate the primary signal from supporting signals
A primary signal is the action or condition the team will use to judge whether the campaign is moving in the intended direction. A supporting signal adds context but should not quietly replace the primary goal. If the objective is a completed information request, a completed request might be primary while qualified visits, click-through rate, and form-start rate are supporting signals. If the objective is awareness, reach, frequency, engagement, or completed views may be more relevant supporting measures than form submissions.
Writing this hierarchy down prevents an easy reporting mistake: celebrating the most flattering number even when it is unrelated to the campaign’s stated purpose. It also makes discussions with stakeholders more concrete. A team can ask, “Did we measure the action we said mattered?” before asking whether a result looks large or small.
Define the action precisely
Terms such as lead, engagement, and conversion need a local definition. Google describes conversions as valuable actions chosen by the advertiser, with examples such as a purchase, newsletter signup, or request for information. That flexibility is useful, but it also means a report is only interpretable when the underlying action is named.
For each primary or supporting signal, record:
- the exact action being counted;
- where it happens, such as a page, form, store, or app;
- who owns the definition and can answer questions about it;
- the reporting period and time zone; and
- known limitations, including actions that cannot be observed or connected to a campaign.
This list is not a substitute for a tracking, consent, or privacy review. It is simply an editorial and operational record of what a dashboard label means.
Choose a small reporting set
Most campaign teams can begin with a compact scorecard: one primary signal, two or three supporting signals, spend or delivery context, and a short note about what changed during the period. The note matters. If the landing page, offer, targeting, creative, reporting window, or conversion definition changed, comparisons may need qualification.
For example, a traffic-oriented scorecard might include qualified visits as its primary signal, then clicks, click-through rate, and the share of visits that reached the intended page as supporting context. A request-oriented scorecard could use completed requests as the primary signal and show form starts, qualified visits, and cost per completed request alongside it. The examples are templates, not prescriptions.
Agree on the review question
Before launch, write the question the report should help answer. “Did this campaign create more impressions?” is a delivery question. “Did people who arrived from this campaign reach the information page we intended?” is a journey question. “Did the defined request action occur often enough to justify another test?” is a decision question. Different questions call for different evidence.
At the review, preserve uncertainty. A reported association does not by itself explain why an outcome occurred, and a short reporting window may not show the full customer journey. Rather than declaring a winner, record what the data can support, what it cannot support, and what should be tested or clarified next.
A one-page measurement brief
- Outcome: What change is the campaign intended to support?
- Audience and message: Who is the communication for, and what is the central promise?
- Primary signal: Which defined action or condition will anchor the review?
- Supporting signals: Which two or three measures supply context?
- Definitions and limitations: How is each signal counted, and what is missing?
- Review question: What decision should the reporting help the team make?
Completing this brief does not make a campaign predictable. It does make the measurement conversation more disciplined: the team begins with a stated outcome, uses metrics that connect to that outcome, and documents the limits of what those metrics can say.