Promotion post-mortem: plan, actual and verdict
Matches plan and actual on six indicators, takes into account the baseline method and what was measured, and gives a “repeat, change or stop” verdict with a list of measurement errors. Do it 2–4 weeks after the promotion ends, when the dip is visible.
At a glance
- What question does it answer?
- Repeat the promotion, change it or stop it, and what we measured badly.
- What data is needed?
- The main goal of the promotion and the baseline method.
- How to read the result?
- By default, of six indicators with a 5% tolerance three are on plan (volume, revenue, share of new customers); margin is 6.3% below plan, returns and costs are above. Incremental profit is +30,000, but cannibalization was not measured: the verdict is “change”, not “repeat”. Example on the default values: Verdict — 3 of 6 indicators on plan · change; Indicators met — 50%; Incremental profit — RUB 30,000; Measurement flaws — 1.
Why and how to reason
A promotion post-mortem exists to learn from facts and not feelings. Plan and actual are compared on six indicators and the tool counts how many goals were reached within tolerance. A plus indicator (volume, revenue, margin, share of new) is met if the actual is not below plan by more than tolerance; a minus indicator (returns, costs) is met if the actual is not above plan by more than tolerance. In the example tolerance is 5%: volume 680 against 600 (met), revenue 2.5 against 2.4 million (met), margin 750 thousand against 800 thousand (−6.25%, not met), share of new 31% against 25% (met), returns 14% against 10% (not met), costs 95 thousand against 80 thousand (not met). That is 3 of 6, 50%. Then measurement quality is counted: cannibalization was not measured, so the gain may have been taken from neighboring items. The conclusion depends on both factors.
When to use it
Do it 2–4 weeks after the promotion, when the dip is visible, and before deciding to repeat. State the main goal (clear stock, attract customers, raise revenue): the conclusion is assessed against it. It does not fit a promotion without a plan: nothing to compare with. Honestly mark what was measured: baseline method, dip, cannibalization, overlap with other promotions. If the promotion ran in selected stores, the result may be biased.
How to read the result
Read the conclusion together with the list of measurement errors. Repeat is justified when most goals are met, including the main one, and there are few measurement errors. Change means there is an effect but depth, term or costs are wrong. Stop when profit does not cover costs and the main goal is not reached. Here 50% are met and there is one measurement error: change, and measure cannibalization before repeating. Write down what to change and set a review date. It is worth keeping the post-mortem next to the plan of the next promotion and reread recorded errors at the start of planning: repeating known misses is the commonest cause of weak promotions.
Common mistakes
The usual mistakes: writing the post-mortem right after the promotion without waiting for the dip; using a plan drawn up after the promotion; hiding measurement errors so the conclusion looks cleaner; comparing with last year’s baseline without adjustment. A conclusion without recorded lessons is useless: save it for the next promotion. Do not change the tolerance after the promotion to make indicators pass.
What data is needed?
- The main goal of the promotion and the baseline method.
- Plan and actual on six indicators: volume, revenue, margin, share of new customers, returns, costs.
- What was measured: the dip, cannibalization, overlap with other promotions, self-selection; incremental profit from “Promotion result”.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Main goal of the promotion | Clear leftover stock | One promotion, one goal: with two goals the effect is hard to assess. | |
| How the baseline was estimated | Seasonal and trend adjusted | How you estimated the baseline: by season, by control or otherwise. The more reliable the method, the higher the trust in the verdict. | |
| Tolerance of deviation from the plan | % | 5 | An indicator counts as met if it is not worse than the plan by more than the tolerance. A rule of thumb, replace with yours. |
| Incremental profit of the promotion (optional) | ₽ | 30,000 | From “Promotion result” or a control. Without it the verdict “repeat” is not given. Optional. Empty: the field is ignored (counted as zero in sums). |
| Post-promotion dip | Measured | Whether you measured the sales dip after the promotion. Without it the uplift is overstated. | |
| Cannibalization | Not measured | Whether you measured how many sales the promotion took from neighboring items. | |
| Overlap with other promotions | There was none | Whether other promotions ran in the same weeks: they mix up the effect. | |
| The promotion ran in selected stores or for selected customers | No | If the promotion ran in selected stores or for selected customers, the comparison with the rest is distorted by selection. |
How it is calculated
| Step | Formula |
|---|---|
| Deviation | actual ÷ plan − 1; the sign flips for “lower is better” |
| On plan | higher is better: actual ≥ plan × (1 − tolerance); lower is better: actual ≤ plan × (1 + tolerance) |
| Verdict | repeat: profit > 0, at least two thirds on plan and at most one measurement flaw; stop: profit ≤ 0 and under half on plan; otherwise change |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Verdict
- 3 of 6 indicators on plan · change
- Indicators met
- 50 %
- Incremental profit
- 30,000 ₽
- Measurement flaws
- 1
Indicators
| Indicator | Plan | Actual | Deviation | On plan |
|---|---|---|---|---|
| Volume, pcs | 600 | 680 | 13.3 % | yes |
| Revenue excl. VAT | 2,400,000 | 2,500,000 | 4.2 % | yes |
| Gross margin | 800,000 | 750,000 | −6.3 % | no |
| Share of new customers, % | 25 | 31 | 24 % | yes |
| Returns, % | 10 | 14 | 40 % | no |
| Promotion costs | 80,000 | 95,000 | 18.8 % | no |
Cannibalisation is not measured: part of the uplift was taken from neighbouring items.
Goal: clear leftover stock. What to watch first: leftover in pieces and money against profit, exit price.
Carry the verdict into the promotion calendar: what to repeat, what to change and what gap is needed before the next one. Counting by revenue without margin and picking “successful” stores for the promotion are frequent errors.
Nuances and limits
- Do it 2–4 weeks after the end, when the dip is visible.
- One promotion, one goal.
- Without incremental profit the verdict “repeat” is not given.
- The verdict thresholds are conventional: replace with yours.