End the campaign with a decision, not a retrospective deck

A useful campaign review reconstructs what the team believed, what actually happened and which operating decision will change before the next budget is committed.

The campaign has ended. The review contains delivery totals, screenshots and a slide called “learnings”. Everyone agrees that collaboration could improve, targeting should be refined and the next campaign needs more time.

Then the next brief repeats the same assumptions. The review documented activity but failed to change the system that produced it.

A worthwhile post-mortem should leave a small number of decisions with owners. Its purpose is neither to defend the campaign nor to identify a person to blame. It is to make the next commitment better informed.

Recover the original prediction

Begin with the approved brief and forecast, not the final presentation. What behaviour was expected to change? Which audience was meant to respond? What conditions had to hold for the plan to work?

If those statements were never recorded, acknowledge the gap. Reconstructing intent from memory after results are known invites a convenient story in which the campaign always aimed to achieve whichever metric improved.

Keep original targets visible alongside any legitimate mid-campaign changes. A revised objective can be sensible, but the review should explain why it changed and who approved the change.

This creates a distinction between an ambitious forecast, a flawed mechanism and a delivery failure. Each calls for a different remedy.

Build a timeline before a verdict

Record the meaningful events: approval, launch, tracking changes, creative revisions, stock constraints, sales handoffs and external disruptions. Include dates so the group can examine which explanations are even chronologically possible.

Then place outcomes against that timeline. If qualified enquiries fell before the creative change, blaming the new creative is questionable. If sales response times increased during the peak, poor progression may not be solely an audience problem.

Google’s post-mortem guidance describes learning from incidents without treating individual blame as the organising principle. The application here is to marketing operations: reconstruct conditions and improve the process that shaped people’s choices.

A blameless review still has accountability. Owners must complete agreed changes; the difference is that “be more careful” is not accepted as a sufficient system improvement.

Separate observation from explanation

Write “demo requests increased” as an observation. Write “the new message created stronger intent” as a proposed explanation. The second statement needs evidence beyond the first.

The Magenta Book’s evaluation framework encourages testing the proposed mechanism and considering alternatives. A marketing review can use that discipline without pretending every campaign has an experimental design.

For each important explanation, ask what else could produce the pattern. A price promotion, sales incentive, seasonal shift or reporting change might account for part of it. List unresolved alternatives rather than voting for the most comfortable story.

Also distinguish evidence of no meaningful effect from a measurement system unable to detect one. Those findings should lead to different decisions about the next campaign and the next measurement investment.

An illustrative review that changes something

Imagine a B2B campaign that produces many downloads but few useful sales conversations. The first reaction is to demand a more aggressive follow-up sequence.

The review samples the downloads and finds that the material appealed mainly to students and general researchers. Sales also received incomplete account information. The actual problem is a mismatch between the content’s audience and the campaign’s commercial objective.

One decision is to redesign the offer around an operational problem faced by target buyers. Another is to test the qualification fields and routing before launching paid distribution. A third might be to preserve the broad educational asset under a different purpose and budget.

These are hypothetical findings, but the distinction matters. More follow-up effort would not necessarily repair audience fit and could make the experience worse for people who never wanted a sales conversation.

Limit the action list

A review that produces twenty improvements often produces none. Select the changes most likely to prevent recurrence or improve the next consequential decision.

Each action needs an owner, a due date, a completion condition and the next project where it will be used. “Improve lead quality” is an aspiration. “Approve target-account eligibility before the next media booking” is an operating change.

Some findings justify stopping an activity rather than improving it. Make that option legitimate. Otherwise reviews become an endless programme of small optimisations around an approach that no longer deserves investment.

Keep unresolved questions in a separate learning list with a proposed way to answer them. Do not disguise uncertainty as an action item nobody can complete.

Close the loop visibly

At the next campaign approval, ask which previous review actions apply and whether they were completed. This makes the post-mortem part of planning rather than an archive the team visits only when challenged.

Review successful campaigns too. A good outcome can conceal weak process, favourable timing or an unusually strong sales contribution. Repeating the creative without understanding those conditions may produce disappointment.

Over time, the organisation should accumulate a record of predictions, evidence and changed decisions. The most useful review is the one a future team can consult before making the same expensive assumption again.