Measurement Design
Part of Media investment reviews
Explaining a campaign result across several channels
Explain a multi-channel campaign result using one business outcome, channel delivery, reporting rules and known limits.
Explain a multi-channel result by starting with one business outcome, then showing what each channel delivered and what its reporting can explain. Keep the timeline, material changes and competing explanations visible. Supplier conversion totals cannot simply be added to create a campaign result.
State the result once
Define the outcome in the business record: completed orders, accepted enquiries or another event. State the dates, eligible market and how duplicates and cancellations were handled. Show the observed total and a relevant comparison period if one exists.
Note differences between periods, such as a changed offer or stock position. A before-and-after difference describes what happened; by itself, it does not isolate media's effect.
Give each channel its intended job from the approved plan. One may have introduced an unfamiliar service, another answered active enquiries and a third reminded prior visitors. Review each against that job before ranking them by a single response rate.
| Layer | Evidence to show | Claim it supports |
|---|---|---|
| Campaign conditions | Offer, availability, dates and major changes | What customers could act on |
| Delivery | Spend, placement, audience and timing | Whether planned opportunities ran |
| Channel response | Interactions and supplier-credited actions with definitions | What each system recorded or credited |
| Business outcome | The business record and its duplicate-handling rule | What the organisation observed |
| Effect estimate, if available | Defined comparison and uncertainty | What the tested change may have added |
Reconcile the reporting rules
Ask each supplier which action it credits, which interactions qualify, its counting rule and its attribution window. Record the settings used for the exported figures; another supplier may use different rules.
If two platforms each credit activity connected to one sale, that does not make two business orders. Preserve both platform figures as diagnostics and count the order once where the business record permits. Mark outcomes with unknown channel allocation as unknown rather than distributing them by guesswork.
Steps to Reconcile Multi-Channel Reporting
- Identify credit rules per platformCheck which action is credited (e.g., click, view, conversion)
- Confirm counting methodDetermine if it's first-click, last-click, or linear attribution
- Record export settingsEnsure consistency across platforms
- Avoid double-counting ordersCount each business outcome once in the final record
Build a defensible account
Put events in date order: launch, delivery changes, offer changes, response and outcome review. Check whether the campaign reached the intended market and whether the destination worked. Record promotions, availability problems or other material conditions that coincided with the result. State which explanations the available records can test.
Consider a repair service whose video and search activity ran together while accepted bookings rose. Search may report more credited enquiries, and video may have reached people earlier.
Those reports describe delivery and credited response within their coverage. Without a suitable comparison, the review cannot assign the booking increase between channels or establish how many bookings media added.
A controlled lift study asks a different question from attribution reporting: it compares outcomes under a tested media change with a control under the study design. If no suitable effect estimate exists, say so. End with the strongest supported conclusion, its main limit and the decision it informs. Valid delivery with uncertain channel contribution might support maintaining a bounded mix while testing one change.
