China Campaign Retrospective Template
Use a China campaign retrospective template to normalize evidence, compare outcomes with the original decision, diagnose funnel and execution gaps, and assign next actions.
- VERSION
- 2.0.0
- UPDATED
- 2026-08-01
- LICENSE
- MIT
- FILES
- 04
npx skills add fffffrost/market-skills --skill campaign-retrospectiveTurn campaign evidence into the next owned decision.
A retrospective should not celebrate activity or explain results after the fact. Reconstruct the original decision contract, normalize the evidence, separate outcome from diagnosis, and end with a small set of owned changes that can be tested next time.
Recover the original plan
Restate the objective, baseline, target, audience, proposition, budget, timing, and planned decision thresholds.
What did the team agree success would mean before launch?Normalize the record
Align dates, sources, attribution windows, definitions, missing data, platform changes, and confidence levels.
Which numbers can be compared, and which cannot?State what happened
Compare business and behavior outcomes with the baseline and target before interpreting causes.
What changed, by how much, and with what confidence?Locate the break
Analyze audience, message, channel, creative, conversion, sales handoff, and retention without collapsing them into one verdict.
At which decision stage did evidence diverge from the plan?Review operating quality
Check delivery, approvals, timing, ownership, experiment integrity, dependencies, and exceptions against the plan.
Was the result caused by the choice, the execution, or both?Commit the learning
Convert findings into stop, continue, change, and test decisions with owners, dates, evidence needs, and follow-up reviews.
Who will do what differently, and when will we know?CHINA CAMPAIGN RETROSPECTIVE TEMPLATE
ORIGINAL CONTRACT
Business objective:
Audience behavior objective:
Baseline / target:
Audience / proposition:
Budget / period:
Planned decision thresholds:
EVIDENCE QUALITY
Source | Date range | Definition | Attribution window | Missing data | Confidence
OUTCOME SUMMARY
Business result vs target:
Behavior result vs target:
Leading indicators:
Material unknowns:
FUNNEL DIAGNOSIS
Audience:
Message / proposition:
Channel and creative:
Conversion:
Sales / partner handoff:
Retention / follow-up:
EXECUTION REVIEW
What shipped as planned:
Delays / exceptions:
Approval or ownership gaps:
Experiment integrity:
DECISIONS
STOP:
CONTINUE:
CHANGE:
TEST NEXT:
ACTION REGISTER
Action | Owner | Due date | Evidence needed | Review dateFour distinctions that keep a retrospective honest.
The example is synthetic and does not imply a real performance benchmark. The linked case shows how evidence limits remain visible.
Record what changed against the original target before offering a causal story.
Expose mismatched definitions and missing windows rather than combining every available metric.
Separate a weak proposition from a late launch, broken handoff, or incomplete experiment.
Every change includes a due date and the signal that will confirm or reject it.
- Original brief
- Platform and CRM exports
- Execution timeline
- Team and agency notes
- Result scorecard
- Evidence-backed findings
- Attribution limits
- Owned action plan
Know where the workflow must stop.
- Cannot finalize a retrospective before the decision period is complete
- Missing or changed tracking definitions must remain explicit uncertainty
- Does not infer causal attribution from platform correlations
These conditions distort the result.
- Platform metrics are compared without normalizing definitions and time windows
- Agency reports and first-party evidence are accepted without reconciliation
- Actions lack owners deadlines measurement points or data repairs
How the skill works
- 01Reconstruct the original China contract
- 02Normalize platform and CRM evidence
- 03Analyze outcome funnel execution and explanation
- 04State attribution limits
- 05Convert learning into owned decisions
- 06Set the next measurement point
See it inside a complete job.
The case preserves task context, source material, execution trace, output excerpts, and human decisions.
Inspect the source before installing.
Instructions, templates, compatibility, and dependencies remain in one open directory.
- ├─SKILL.md
- ├─agents/openai.yaml
- ├─listing.yaml
- ├─references/retrospective-template.md