How to run a marketing postmortem

Guide · Frameworks · 4 min read · last verified 2026-07-27

A marketing postmortem is a structured review held after a campaign has ended — succeeded, failed, or fizzled — whose purpose is to separate what happened from why it happened, and to convert the difference into one concrete process change. It is a learning ritual, not a verdict. Teams tend to repeat marketing mistakes not because the mistakes are subtle but because nothing in their process forces the lesson to be extracted, written down, and reattached to the way work happens. The postmortem is that forcing function, and it works to the degree that it stays blameless, factual, and small in its conclusions.

A postmortem is not the kill decision

Keep two exercises apart. Deciding whether to stop investing in a channel is a resource-allocation decision, made against thresholds you set in advance — that discipline has its own playbook in When to kill a marketing channel. The postmortem happens after the outcome is already settled, and its product is not a decision but an explanation. Merging the two tends to ruin both: if the meeting might still kill the channel, everyone in the room has a stake in shading the story, and the facts arrive pre-spun. Hold the postmortem when there is nothing left to win except understanding. That is when people are most able to be accurate.

Facts before explanations: build the timeline

The discipline borrowed from incident investigation — read the record before interviewing the people — applies directly to a marketing postmortem, and the order matters: physical evidence first, human narrative second, so that the narrative can be checked against something it cannot argue with. Before the meeting, one person assembles a dated timeline of facts only — what launched when, what changed mid-flight, which creative went live, which audiences were added or removed, what was decided in which conversation, and what was happening outside the company at the time. External context belongs on the timeline too: what buyers were asking during the campaign window is part of the flight record, and a tracked evidence source such as Magrios can supply that outside view alongside your internal logs.

The timeline is circulated before the meeting, and the meeting opens by walking it. No interpretation is permitted during the walk — corrections of fact only. This single rule does most of the work, because memory tends to rewrite itself in the direction of the outcome: after a failure, people tend to remember having had doubts; after a success, they tend to remember the plan working as designed. The timeline is the check that keeps those rewrites honest.

The record is what makes the ritual possible

A postmortem without records is a séance — the team gathers to summon what it wishes it remembered. If the campaign was run as a documented experiment, with its hypothesis, launch conditions, and decisions logged as they happened, the timeline nearly assembles itself; the discipline that makes this cheap is described in How to document growth experiments. If no such record exists, the first postmortem will be painful and partially fictional, and its most valuable output will be the process change that says: we write things down from now on. That is a perfectly good first lesson, and teams that learn it once tend not to need it twice.

Find the assumption that broke

With the timeline established, the analytical work is to name the specific assumption that failed. Every campaign encodes assumptions — this audience feels this pain, this message will reach them in this place, this offer clears their bar for action. "The campaign didn't work" is not a finding; it is a refusal to look. A finding names the belief: we assumed the people seeing the message were the people feeling the problem, and the questions buyers actually ask suggest otherwise. Push each candidate explanation until it terminates in an assumption someone can own without shame — held, reasonable at the time, now shown wrong. Beliefs can be updated; vague failure just gets re-run with new creative on top.

Blameless does not mean consequence-free

Blameless means the explanation is not allowed to terminate in a person. "The launch email went out late" may be a fact on the timeline; "because nothing tells anyone an email is unowned until it is late" is the level a postmortem works at. The reason is practical, not therapeutic: people tend to stop volunteering facts the moment facts are used against them, and a postmortem runs on volunteered facts. But blameless does not mean nothing changes. The consequence lands on the process — a check added, a handoff clarified, a threshold agreed in advance — which is precisely the consequence that prevents recurrence, something blame has a poor record of doing.

One change, one artifact, one follow-up

Resist the closing temptation to fix everything. A postmortem that produces a list of improvements tends to produce zero improvements; a postmortem that produces one process change, with an owner, tends to produce one. Choose the change that would have caught the broken assumption earliest, and write the whole thing down in a short standing format: timeline, assumption that broke, change adopted, owner. The artifact outlives the meeting — it is what lets a new marketer inherit the team's scar tissue instead of re-earning it, and its lessons feed naturally into your reporting rhythm, from the weekly growth review where the process change gets checked, to the monthly investor update where the candid version of the story belongs. Then schedule one follow-up look: did the change stick, and did it catch anything. A postmortem is only finished when its one change has survived contact with the next campaign.