Separate product switching from timing shifts
Peer cannibalization describes sales displaced from other products. Post-promotion payback describes sales shifted from later periods into the campaign. Halo describes additional contribution on complementary items. These are separate components, and their interpretation depends on whether the boundary is a retailer's category, a manufacturer's portfolio or the whole business.
Oracle's promotion-planning documentation treats cannibalization, halo and pre/post effects as distinct parts of promotion analysis. Their existence in a conceptual model is not evidence that a particular pair of products caused each other's movements.
Value lost contribution, not lost turnover
Losing a €30 sale does not always lose €30 of contribution. Deduct the variable costs that would have accompanied that sale. A €30 peer product costing €20 contributes €10; displacing 80 of those units costs €800 of contribution under that benchmark.
The sign can be counterintuitive for loss-making products. Displacing a genuinely negative-contribution sale can improve the total. Keep the signed margin visible instead of automatically forcing every displacement term to be a loss.
Worked example across the full horizon
A promoted item would have sold 1,000 units at €50 with €25 variable cost. It instead sells 1,800 at €40, giving a direct contribution change of +€2,000 before spillovers.
| Additional component | Contribution effect |
|---|---|
| Direct promoted-item change | +€2,000 |
| 120 later units displaced × €25 | −€3,000 |
| 80 peer units displaced × €10 | −€800 |
| Complementary-item contribution | +€200 |
| Net change over chosen boundary | −€1,600 |
Map measured effects without duplication
The post-promotion-payback template records the 120 later units in the workbench. That field values them at the promoted item's regular contribution, producing −€1,000 after later displacement. The free cannibalization calculator can then add peer and halo effects.
If you start the free calculator from the workbench's result after later displacement, enter zero additional later displacement there. Entering the same 120 units in both places would subtract them twice.
| Effect | Where to record | Boundary |
|---|---|---|
| Later units of promoted item | post_promo_dip_units | Valued at its regular unit contribution |
| Peer-product losses | Separate cannibalization worksheet | Use each peer's own unit contribution |
| Halo contribution | Separate cannibalization worksheet | Net contribution, not revenue |
| Unmeasured later effect | Leave field blank | Missing measurement is not observed zero |
Check attribution before adding precision
A peer's decline can also reflect stockouts, seasonality, a competing event or a different customer mix. A later dip can come from calendar effects or availability. Choose a credible benchmark and an appropriate horizon before assigning the difference to the promotion.
Do not sum overlapping campaigns' estimated payback if they claim the same displaced units. A coherent category or customer-level model may be needed when events overlap heavily.
Frequently asked questions
How long should the post-period be? Use a horizon relevant to normal replenishment and the commercial decision. There is no universal number of weeks that establishes payback for every product.
Does blank displacement mean none occurred? No. In the workbench it is excluded from arithmetic and flagged as unmeasured. Record an observed zero only when that interpretation is supported.
Put it to work
Source context
Guide text and examples were prepared independently for MarginTrial. Sources provide the stated methodological or product context. They do not validate the hypothetical commercial outcomes.