Most measurement problems are trust problems. When leadership does not believe the dashboard, the numbers get replaced by opinion, and the marketing team loses the ability to argue for investment with evidence.
Trust is not built with better dashboards. It is built by being precise about what each number can and cannot tell you, and by never changing a definition mid-quarter.
The three layers
Platform data: what happened today
Platform reporting is operational. It tells you whether a campaign is delivering and at what cost. It is fast, granular, and self-interested—each platform claims credit for conversions it may only have assisted.
Use it daily, for execution only. Never use it to settle a budget argument between channels.
Attribution and incrementality: what actually contributed
This layer exists to correct platform bias. There are three credible approaches, and mature teams use at least two:
- Marketing mix modelling (MMM). Statistical, channel-agnostic, good for budget allocation across quarters. Requires clean spend and revenue history.
- Incrementality testing. Geo holdouts and conversion lift studies. The most credible evidence, the least frequent.
- Multi-touch attribution. Useful for relative comparison within a channel mix, unreliable as an absolute source of truth.
Triangulate, then decide
When all three disagree—and they will—trust the incrementality test for direction, MMM for magnitude, and platform data for execution. Document the decision and move on.
Business outcomes: what it was worth
The layer most marketing dashboards omit: margin, payback, retention by cohort, and pipeline contribution. Without it, you cannot tell the difference between growth and spending.
| Layer | Owner | Cadence | Changes strategy? |
|---|---|---|---|
| Platform | Channel manager | Daily | No |
| Attribution / MMM | Analytics lead | Monthly | Occasionally |
| Business outcomes | Marketing + finance | Quarterly | Yes |
2 of 3
Minimum number of attribution methods to triangulate before a budget decision.
1 quarter
Freeze metric definitions for a full quarter. Consistency is what makes trends readable.
90 days
Cohort retention window that separates durable acquisition from expensive churn.
Five rules that make reporting credible
- Define every metric once, in writing. Include the formula, the source, and the exclusions.
- Separate measurement from reporting. The person who builds the model should not be the person presenting the target.
- Publish misses as loudly as wins. Credibility is built on the week you show the number that failed.
- Annotate the timeline. Every spike has a cause. Record it or the chart becomes folklore.
- Report ranges, not single points. Forecasts with a band communicate uncertainty honestly.
Start with one question
Do not begin by instrumenting everything. Begin with the single decision the measurement needs to inform this quarter—usually "where should the next $50k of budget go"—and build the smallest stack that answers it.
A measurement stack that answers one question well earns the right to answer five.
