A good project retrospective presentation answers four questions: what did we set out to do, what actually happened, why the difference, and what will we keep or change next time. Nine slides cover it: background and goals, results against goals, a timeline, what worked, problems and root causes, lessons, and actions, with a one-sentence conclusion on the cover and an appendix for detail. Keep it blameless: trace causes to systems and process, not people. With peak-season campaigns such as Singles' Day, Black Friday and Cyber Monday just behind most retail teams, now is when these decks get built.
Retrospective vs year-end review vs status report
The three get mixed up, but they answer different questions:
| Retrospective / post-mortem | Year-end review | Status report | |
|---|---|---|---|
| When | After a project, launch, campaign or incident | End of year | On a cadence, or at milestones |
| Core question | Why did we get this result, and what changes next time? | What did we achieve, and what is next year's plan? | Where are things, and what needs deciding? |
| Focus | Goals vs actuals, timeline, root causes, lessons, actions | Results, gaps, plan, resources | Status, risks, asks |
| Audience | The project team and stakeholders | Leadership and the department | A manager or decision-maker |
For annual results, see the year-end review presentation. For ongoing updates, see status report presentation structure.
Four steps: write the doc, then build the deck
A solid retrospective follows four steps:
- Restate the goals. Pull up the goals as they were set at kickoff, including the key assumptions. Do not quietly revise them to fit the result.
- Compare results. Put each goal next to the actual, calculate the gap, and mark it met, close or missed.
- Explain why. Where things went well, identify what is repeatable. Where they did not, trace from symptom to root cause.
- Capture lessons. Turn causes into rules, checklists or process changes, then into specific actions.
Write these four steps as a doc before you build slides. If the thinking is fuzzy in the doc, no slide design will fix it.
A 9-slide structure
For a talk of about 15 minutes:
| Slide | Page | What goes on it | Time |
|---|---|---|---|
| 1 | Cover and headline | Project name, dates, the conclusion in one sentence | ~0.5 min |
| 2 | Background and goals | Why we did it, the quantified goals, key assumptions | ~1 min |
| 3 | Results vs goals | Goal, actual, gap, color-coded by status | ~2 min |
| 4 | Timeline | Key milestones and decisions | ~2 min |
| 5 | What worked | Two or three repeatable practices, each with evidence | ~2 min |
| 6 | Problems and root causes | Two or three issues: symptom, trigger, root cause | ~3 min |
| 7 | Lessons | Lessons written as rules, checklists or process changes | ~2 min |
| 8 | Actions | Action, owner, due date, definition of done | ~2 min |
| 9 | Appendix | Detailed data, channel or stage breakdowns, raw logs | As needed |
The one-sentence conclusion on slide 1 does the heavy lifting. "We broadly hit target; growth came from early content investment, and stockouts at peak were the main loss" tells the room what every later slide is proving. Write each title as a conclusion too; see one idea per slide.
Example: a peak-season campaign retrospective
The Singles' Day marketing plan in our gallery ends with the metrics its retrospective should use: GMV, ROI, new-customer acquisition cost, add-to-cart rate and return rate. Once the campaign closes, that plan is the starting point: goals, phases and channel budgets are all there to compare against. The same approach works for a Black Friday or Cyber Monday campaign.
Here are sample titles for the nine-slide structure. The numbers are placeholders:
| Slide | Conclusion-style title |
|---|---|
| Results vs goals | GMV reached X% of target; new-customer share met; repeat rate missed by X points |
| Timeline | Pre-sale add-to-carts beat forecast; the hero product sold out on day X of the peak |
| What worked | Moving creator content into the warm-up phase lifted search traffic share to X% |
| Problems and root causes | The stockout was not total inventory; reorder triggers were set on everyday sales |
| Lessons | Size peak inventory from pre-sale cart data, with a separate alert line for hero products |
| Actions | Recalculate reorder thresholds from pre-sale data X days before the next peak (owner: supply chain; due: TBD) |
Every slide answers "why" and "what next," not just "what happened."
Writing the root-cause slide
This slide is the heart of the deck and the one most often written too shallow.
- Separate three layers. The symptom is what you saw, the trigger is the event that caused it, and the root cause is the system that let it happen. Stop at the trigger and it will happen again.
- Keep asking why. Take a high return rate: returns are high → most cite "not right for my skin type" → the product page does not say which skin types it suits → the launch checklist has no "who is this for" review step. Now the action has somewhere to land.
- Separate controllable from uncontrollable. Platform rule changes or weather can be noted, but keep them apart and spend your time on what the team controls.
- Systems, not people. Replace "X did not reorder in time" with "reorder rules do not distinguish peak from normal days." The first invites defensiveness; the second invites a fix.
The actions slide: no owner, no change
Actions are the part most often lost after a retro. Put them in one table:
| Action | Owner | Due | Done when |
|---|---|---|---|
| Add pre-sale cart data to the peak inventory model | Supply chain lead | X weeks before next peak | Model live and tested in one mid-size campaign |
| Add a "who is this for" item to the launch checklist | Merchandising lead | TBD | Checklist updated; all new product pages complete |
| Daily stand-up retro during peak events | Project manager | Before the next peak | Written into the campaign playbook |
Three to five actions are enough, each with a name, a date and a definition of done. Open the next retrospective by checking whether the last one's actions were completed.
Common mistakes
A diary. Listing what happened each day hides what mattered. Mark key moments on a timeline and move detail to the appendix.
Only wins. The value of a retro lives in the problems and lessons. A victory lap teaches nothing for next time.
Moving the goalposts. Comparing results with revised goals defeats the purpose. Show the original goals and explain any changes separately.
Shallow causes. "Poor execution" and "communication issues" are not root causes. Keep digging.
Blame. Once causes point at people, discussion turns into defense.
Orphan actions. "Improve collaboration" with no owner or date will reappear, word for word, at the next retro.
Turn your retro doc into slides with AI
Once the doc is written, AI can handle slide-making. Our AI slide studio produces high-quality, editable decks:
- Bring your text. Organize the doc around the four steps, take numbers from system reports, and mark gaps as "[TBD]." Open the AI slide studio and paste it, or upload a .docx, .pdf, .txt or .md file, or import a web page by link. PDFs need selectable text; run OCR on scanned files first.
- Edit the outline. The AI drafts a slide-by-slide outline first, and the outline uses no credits. Rewrite titles as conclusions like the ones above and add or remove slides to match the nine-slide structure. The deck follows the outline you approve.
- Pick a style and generate. A clean, data-forward style suits a retrospective, and a brand kit holds your logo and footer. Each slide costs 2 credits, and the total is shown before you generate.
- Export and check. Download an editable PPTX, check every number and swap in your real charts in PowerPoint or WPS; for chart choices, see data visualization in slides. Or export a PDF to send as pre-reading. Every generated slide carries a small "AI-generated" footer label, as Chinese regulations require.
The AI turns your doc into well-structured slides. Accuracy of the data and judgment about root causes stay with the team.
FAQ
How many slides should a retrospective presentation have?
Usually 8 to 12, and about 9 works well. In a 30-minute retro, present for 15 minutes and discuss for 15, at one to three minutes per slide. Put detailed data and channel breakdowns in an appendix and pull them up during discussion.
What is the difference between a retrospective and a project summary?
A summary reports outcomes: what was done and how it went. A retrospective also reconstructs what happened, explains why, and turns that into practices to keep and concrete actions to change. That is why the root-cause, lessons and action slides cannot be skipped.
How do I keep a post-mortem from turning into a blame session?
Lead with facts and data, then interpretation. Write causes at the level of process, systems and information rather than individuals. Give each action an owner and a date instead of assigning fault. The facilitator can state the ground rule up front: we are fixing the system, not judging people.
Can I turn my retrospective doc into slides directly?
Yes. Upload a .docx, a PDF with selectable text, a .txt or a .md file, paste the doc, or import a web page by link. Run OCR on scanned files first. The AI drafts a slide-by-slide outline first, which uses no credits; match it to the 9-slide structure, then generate and export an editable PPTX.