A retrospective is not for assigning blame — it is how the team and the system get better after every campaign. The best insight comes from failure, but only if it is safe to speak honestly. A retro with no action is wasted time: every lesson must become one concrete change in the playbook, the Brand Hub, or the process.
Information gathering
Ask up to 4 questions:
- Which campaign, and over what dates? Name, start and end dates, original objective.
- Final data: revenue, leads, CPA, ROAS, spend — final numbers versus target. (If a report already exists from
07-marketing-report-global, use it as the input.) - Who attends the retro? Content, performance, design, sales, leader?
- How long ago did the campaign end? Run the retro within 5 days of the end, while memory is fresh.
Principles
- Mandatory after every major campaign. No skipping because the team is busy. No retro means paying tuition without learning the lesson.
- Record only claims backed by data. "Creative A felt strong" is not an insight; "creative A hit $18 CPA against a $28 account average" is.
- No individual blame. Talk about systems and process. Everyone has the floor.
- Every lesson becomes one concrete change — to the playbook, a template, the Brand Hub, or a process — with an owner and a deadline.
- Document so it still makes sense in 6 months — and link the retro into the next campaign's brief.
Workflow
1. Prepare (before the session)
- Pull final data: revenue, leads, CPA, ROAS, spend, timeline adherence, budget variance.
- Send the data to the team before the meeting — the meeting is for discussion, not for reading numbers.
- Book 60 minutes with a named facilitator (usually the leader). For distributed teams, pick a time that works across the relevant timezones and require written pre-reads from anyone who cannot attend live.
2. Campaign scorecard
| KPI | Target | Actual | % achieved | vs previous campaign |
|---|---|---|---|---|
| Revenue | ||||
| Leads | ||||
| CPA | ||||
| ROAS | ||||
| Spend | ||||
| Timeline (on schedule?) | ||||
| Budget (within plan?) |
Overall verdict: Success (>90% of KPIs) / Mixed (60-90%) / Miss (<60%).
When comparing against category benchmarks, use the right region and industry row in references/benchmarks-global.md — and adjust for seasonality before declaring a CPA increase a failure, since Q4 media costs run materially above Q1.
3. What worked (repeat next time)
| # | What worked | Evidence (numbers) | Why it worked | How to apply next time |
|---|---|---|---|---|
| 1 | ||||
| 2 | ||||
| 3 |
Name the winning patterns explicitly:
- Content angle: which angle converted best
- Creative format: video / image / carousel — which performed
- Audience: which segment responded
- Timing: which day or hour performed best
- Channel: best CPA and ROAS
4. What did not work (avoid next time)
| # | What failed | Evidence | Root cause | Decision: stop / fix / retest |
|---|---|---|---|---|
| 1 | ||||
| 2 |
Root cause must go past the symptom. "The ads were bad" is not a root cause; "audience overlapped with the previous campaign > frequency rose above 4 > CTR fell below 1%" is.
5. Surprises
- Positive surprises: what beat expectations, by how much, why — what do you take from it?
- Negative surprises: what came in worse, by how much, why — how do you prevent it?
6. Process and execution review
| Item | Assessment | Problems hit | Improvement |
|---|---|---|---|
| Brief quality | |||
| Timeline adherence | |||
| Team communication | |||
| Review and approval process | |||
| Launch-day execution | |||
| Data and tracking setup |
Biggest process bottleneck: the single thing most worth fixing next time.
7. Team callouts
Record specific contributions: [Name] — [the specific contribution and its positive impact on the result]. Recognize publicly, correct privately.
8. Playbook and Brand Hub updates plus action items
| Document to update | What changes | Owner | Deadline |
|---|---|---|---|
Launch playbook (60-launch-playbook-global) |
Add or amend step: [X] | Leader | |
| Campaign brief template | Adjust section: [Y] | Leader | |
| Brand Hub — customer insight | Add new insight from this campaign: [Z] | ||
| Brand Hub — KPI framework / internal benchmarks | Update the benchmarks you actually achieved (CPA, CVR) | ||
| Winning creative library | Store the winners plus the reason they won |
Every action item needs an owner and a deadline — never "the whole team will handle it."
9. Input for the next campaign
- 3 things to definitely repeat: 1... 2... 3...
- 3 things to definitely do differently: 1... 2... 3...
- Hypothesis to test next campaign: [hypothesis + how it will be measured]
Facilitation guide
- Timing: within 5 days of campaign end.
- Format: 60 minutes, full team, facilitator (leader): 10 min scorecard > 15 min what worked > 15 min what did not > 10 min process > 10 min action items and owners.
- Ground rules: no individual blame, systems and process only; everyone speaks; constructive, not a venting session.
- After: the leader publishes the document within 24 hours, shares it with the team and leadership, and links it into the next campaign brief.
Output structure
File name: campaign-retro-[campaign-name]-[YYYYMMDD].md — the 9 sections above: scorecard > what worked > what did not > surprises > process review > team callouts > playbook and Brand Hub updates > input for the next campaign > action items with owners.
Related skills
07-marketing-report-global: the final report supplies the scorecard data.60-launch-playbook-global: execution lessons are written directly back into the playbook.02-campaign-brief-global: the "lessons from the previous campaign" section of a new brief comes from this retro.34-ai-marketing-os-global: new insights and real benchmarks are written into the Brand Hub.65-team-performance-review-global: team callouts and observations feed individual reviews at period end — keep the two documents separate.
Quality checklist
- Retro run within 5 days of campaign end
- Scorecard covers all KPIs against target with a Success/Mixed/Miss verdict
- Every "what worked" and "what did not" entry has numeric evidence — no impressions
- Failures trace to a real root cause, not a symptom
- Benchmark comparisons adjusted for region, industry, and season
- At least one playbook update and one Brand Hub update
- Every action item has an owner and a deadline
- 3 things to repeat, 3 things to change, and a hypothesis for next time
- No individual blame anywhere in the document
- Document shared within 24 hours and linked into the next campaign brief