What you’ll end up with
- The result against the target that was set before launch.
- A short account of what caused it, separated from what you observed.
- A named decision: repeat, adapt, or stop.
- Anything worth keeping written into the standard it belongs to, not just noted here.
Before you start
The process
1
1. State the result against the original target
Primary KPI, actual against target, and the counter-metric alongside it. Spend against budget, from budget reporting.No interpretation yet. Just the numbers and what was promised. This step is assembly from the flat file and the filed brief, which is why a builder pre-fills it and why pulling the target from memory never happens: the builder cannot remember a different number than the brief holds.
2
2. Read the change log before explaining anything
Budget shifts, creative swaps, audience changes, tracking fixes. Most of what looks like a performance story has a recorded cause, and attributing a mid-flight budget move to creative quality is the classic retrospective error.
3
3. Separate what you observed from what you think caused it
Write them as different sentences, so someone can accept your numbers and dispute your explanation.E.g. “Cost per trial finished at $52 against a $45 target. Two of the three creative concepts stopped delivering after week two, and spend concentrated into the third, which had the highest cost per trial of the set.”
4
4. Name what you would do differently, specifically
“Better creative” is not a finding. “Three concepts was too few for a six-week flight at this budget” is, because the next brief can act on it.Be equally specific about what worked. Retrospectives skew toward problems, and a team that only records failures slowly forgets what it is good at.
5
5. Decide: repeat, adapt, or stop
One of the three, named explicitly, with the condition attached.
6
6. Write findings into the standards they belong to
This is the step that makes the difference, and the one most often skipped.A copy insight goes into the copy brief or copy best practices. A structural lesson goes into campaign structure best practices. A budget assumption that proved wrong goes into budget planning. A finding that stays in a retrospective is a finding nobody applies.A builder can draft the edit to the standard; the owner approves it before it lands. The standards are the instructions every future campaign, and every agent running one, builds from, so a change to one is never self-approved.
7
7. Link it back and file it
Attach the retrospective to the campaign brief in the campaign library, so the brief carries its own outcome. That’s what makes the library worth searching before the next campaign.
Keep it short
A retrospective long enough to feel thorough is usually too long to be read. One page is enough for the result, the cause, the decision, and what changed as a result. The detail is already in the reports.Exceptions
A campaign stopped early still gets a retrospective, and it is usually the more useful one. Record why it was stopped and at what point that became clear, since recognising a failing campaign sooner is itself the lesson. Always-on activity gets a retrospective on a set cadence rather than at an end date, most naturally alongside the quarterly report.Related resources
- Marketing campaign brief Where the original target was set.
- Campaign library Where the retrospective is filed.
- Make a campaign The full path this closes out.
- Report on results Where the numbers come from.
- Experiment learning library Where any test results are recorded.
MCP wiring
MCP wiring
Type: process
Domains: campaign, reporting
Provides:
Uses:
Hands off to:
Domains: campaign, reporting
Provides:
process:campaign-retrospective, process:campaign-retrospectiveUses:
budget-plan, campaign-brief, campaign-structure-best-practices, copy-best-practices, copy-brief, experiment-log, flat-file, library:campaign, library:change-logHands off to:
process:budget-reportingTriggers: “Write the retro for campaign X” · “How did the campaign do?” · “Close out the campaign” · “Post-mortem for last quarter’s always-on” · “What did we learn from X?”Asks, in order:- Which campaign are we closing out? Options:
library:campaign. Skip when: only one recently ended campaign. - Here’s the result against the filed target and the change log. What do you think caused it? Options: your answer. Skip when: never skip; cause is the owner’s.
- Repeat, adapt, or stop? Options:
retro-decision. Skip when: never skip. - Any findings to write into a standard? I’ll draft the edit for you to approve. Options: your answer. Skip when: no findings.
Get the complete Marketing OS
Subscribe to Snow Doughnut, our free weekly newsletter, and get the Notion workspace where you do the work: tasks, a marketing calendar, name generators, briefs, and more.
Have us run it for you
We implement this system, use it to help you grow and scale, then hand it all off to you when you’re ready.