Prepare the retrospective from frozen inputs
Schedule the review after the evidence cutoff, while leaving enough time to recover obvious gaps. Freeze the report version used in the discussion. Keep the original objective, intended audience, approved message, campaign dates, activity log, evidence register, qualitative review, product-send ledger, and any separate outcome source within reach.
Write the scope at the top of the worksheet. A campaign retrospective should not absorb unrelated monthly work merely because it happened nearby. It also should not rewrite the objective after seeing the outputs. If the initial objective was vague or unmeasurable, record that as a planning finding for the next campaign.
| Input | Version to bring | Question it answers |
|---|---|---|
| Campaign brief | Approved starting version | What did we intend to change and for whom? |
| Frozen report | Named cutoff and filters | What entered the final scope? |
| Evidence register | Included, excluded, and missing-proof rows | Which output claims can be checked? |
| Contact and send history | Campaign-linked actions only | What work was actually completed? |
| Qualitative review | Grid version and reviewer notes | How was the coverage read? |
| Outcome sources | Survey, analytics, or business record where available | What changed beyond outputs? |
Separate fact, interpretation, unknown, and decision
Run every discussion point through four columns. A fact is traceable to the frozen input, such as five evidence-ready items. An interpretation explains a possible pattern, such as the core message appearing more often in features than roundups. An unknown states what the available sources cannot answer. A decision names the next action without pretending the interpretation is proven causation.
This separation prevents a persuasive meeting comment from entering the next brief as settled evidence. It also lets a useful decision proceed under uncertainty. You can test a narrower message next time while still recording that the current campaign did not measure audience understanding.
| Column | Write | Reject |
|---|---|---|
| Fact | A dated row, count, passage, or recorded action | General impressions with no source |
| Interpretation | A pattern plus the evidence that prompted it | A causal claim the data cannot establish |
| Unknown | The missing source or unanswered question | A zero used in place of missing data |
| Decision | Keep, change, test, or stop, with owner and date | A vague wish for more coverage |
Reconcile the fictional Solstice record before discussing it
The fictional final report starts with six unique coverage candidates. Five are evidence-ready and included in the headline output count. One in-scope mention remains in the missing-proof appendix. Of the five included items, three accurately carry the core rainwear message and two do not. Audience response, awareness change, and sales effect were not measured.
The separate send ledger contains eight recipient decisions: two cancelled before dispatch, one ready but unsent, and five sent. Four of the five sent products are recorded as delivered, while one delivery remains unknown. Two delivered sends connect to observed coverage; two have no observed coverage at the review date. Those send facts provide operational context, but they do not show that gifting caused either publication.
| Layer | Reconciled total | Permitted conclusion |
|---|---|---|
| Coverage candidates | 6 | Five ready plus one missing-proof row |
| Evidence-ready coverage | 5 | Five observed outputs meet the stated rule |
| Core message present | 3 of 5 | Three included items contain the agreed message |
| Recipient decisions | 8 | Two cancelled, one ready, five sent |
| Sent products | 5 | Four delivered and one delivery unknown |
| Delivered sends with linked coverage | 2 of 4 | A recorded relationship, not proof that the send caused coverage |
| Audience or business outcome | Not measured | No outcome claim is available |
Complete a decision log while the evidence is visible
Use one row per decision. Name the observation that triggered it, the interpretation being tested, the action, the owner, the due date, the evidence needed at the next review, and the condition that would reverse or refine the choice. Keep unresolved questions in their own rows rather than hiding them in meeting notes.
In the fictional example, three of five included items contain the core message. The next decision is not to declare that the message worked. It is to preserve the concise wording, test whether a tighter briefing note improves its accurate use in the next relevant pitch, and review the resulting evidence. The missing-proof item creates a separate action to recover or close the evidence gap.
| Evidence prompt | Decision | Owner and check |
|---|---|---|
| Three of five outputs contain the core message | Keep the wording; test a shorter briefing block | Campaign owner; compare exact passages at next review |
| Roundup items omit the message | Test a product-first angle only with relevant roundup contacts | PR operator; record acceptance and resulting treatment separately |
| One in-scope item lacks proof | Recover the full item or close it as unverified | Evidence owner; decide before the archive date |
| Two delivered sends have no observed coverage | Review timing and relationship notes before any follow-up | PR operator; do not assume a promised post |
| Audience response was not measured | Choose a feasible source before the next brief | Campaign owner; document method and baseline |
Run a short review that preserves disagreement
Send the frozen inputs before the meeting. During the review, confirm scope and totals first, then discuss patterns, then choose actions. Ask the person making a claim to point to the relevant row or source. Give disagreements an owner and resolution route instead of forcing agreement for a cleaner slide.
Close by reading the decision log aloud. Every action needs an owner, a date, and a future evidence check. Archive rejected ideas with a short reason so they are not reopened without new information. If you work alone, perform the same sequence on two separate passes: evidence first, decisions later. The pause makes it easier to notice when an interpretation has become a fact in your notes.
- Confirm objective, period, inclusion rule, and evidence cutoff.
- Reconcile counts before discussing explanations.
- Label interpretations and preserve dissenting readings.
- Turn each selected idea into an owned decision row.
- Carry unknowns into the next measurement plan.
- Save the frozen report and final decision log together.
Move only the decisions into the next plan
AMEC's planning guidance connects objectives, audiences, activities, and measurement, while its taxonomy treats evaluation as iterative. Use the retrospective to revise the next brief, contact selection, message, evidence practice, and measurement source. Keep the prior facts attached to their original campaign so an old report does not become a living document with changing totals.
cutePR can keep manually entered contacts, campaigns, sends, coverage evidence, qualitative context, and frozen reports connected for one private user. It does not run the retrospective, monitor media, send emails, create team or client workspaces, or measure outcomes automatically. Current report exports use English labels. Whether you use the app or a spreadsheet, the final artifact should be a dated working plan with explicit decisions and remaining unknowns.