Net effect of a new item: cannibalization
Compares the margin gained from a new item with the margin lost on neighboring items after launch (adjusted for the overall trend in control stores). Shows the net effect on the category and the cannibalization share.
At a glance
- What question does it answer?
- Whether a new item widens the category’s sales or takes them from neighbors.
- What data is needed?
- Sales and margin of the new item over the period.
- How to read the result?
- A net effect above zero means the item widens the category. Zero or below means it dilutes the stock. The share of substituted pieces shows what part of the item’s sales came from neighbors; items with a low contribution and high substitutability are candidates to drop. Example on the default values: Margin of the new item — 1.49 m RUB; Margin lost on neighbors — RUB 376,040; Net effect on the category — RUB 961,960; Cannibalization share by margin — 25.3%.
Why and how to reason
A new item rarely brings all its revenue “on top”: part of the buyers switched from neighboring styles. So a novelty’s value is measured by the net effect: the novelty’s margin minus the margin the neighboring items lost, minus additional costs.
Neighbor losses are computed not by a simple before-after comparison but with a correction for the general trend: if sales in control stores without the novelty grew 3%, the neighbor’s expected sales = before x 1.03. The shortfall relative to expected is multiplied by the unit margin.
Example. The novelty sold 620 units at a margin of 2,400: RUB 1.488 m. Neighbors: A expected 515, sold 420, shortfall 95 x 2,100 = RUB 199.5 k; B 61.4 x 2,500 = RUB 153.5 k; C 12.8 x 1,800 = RUB 23 k. Loss RUB 376 k. Net effect 1,488 - 376 - 150 k costs = RUB 962 k. Cannibalization 25.3% of the novelty’s margin, 27.3% of units substituted.
When to use it
It comes in handy when assessing launches of new items and assortment extensions, to decide whether to add similar styles. Assumption: neighbors are chosen correctly (close in price and function), control stores really are without the novelty. It does not account for the effect on traffic and image: a novelty can bring new customers. The period after launch must be long enough. Define neighbors before launch, not after.
How to read the result
Look at the net effect and the cannibalization share. If the share is about a quarter, the novelty takes part of the sales but remains profitable. A share of 70-80% is a sign that you duplicate what exists. Substituted units show how many buyers came from neighbors. Decision: continue with styles with a positive net effect, withdraw duplicates, and in each new launch define in advance whom it should replace and whom complement. Compare the novelty with the alternative: what would have happened had you put the same money into strengthening the bestsellers. The effect of a novelty should be judged against the best alternative use of the budget, not zero.
Common mistakes
Typical trap: comparing “before” and “after” without control stores; including neighbors from another price group; ignoring seasonality; computing the novelty’s margin at full price when discounted. Also: not including launch costs. Also: ignoring that a novelty can displace a neighbor with a better margin.
What data is needed?
- Sales and margin of the new item over the period.
- Sales of 2–5 neighboring items before and after launch over periods of equal length.
- Optional: the trend in control stores without the item and extra costs.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| New item sold in the period | pcs | 620 | Sales of the new item over the same period as you look at the neighboring items. |
| Margin per piece of the new item | ₽ | 2,400 | Price excl. VAT minus cost. |
| Extra costs of the new item (optional) | ₽ | 150,000 | Cost of capital in stock, storage, unsold leftovers, display. Empty: zero. |
| Trend in control stores without the item (optional) | % | 3 | How sales of the same neighbors changed where the item is not sold. Empty: zero trend. |
How it is calculated
| Step | Formula |
|---|---|
| Expected neighbour sales | sold before × (1 + trend of control stores) |
| Margin lost | Σ (expected − sold after) × unit margin |
| Net effect | new item sold × margin − loss on neighbours − extra costs |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Margin of the new item
- 1.49 m ₽
- Margin lost on neighbours
- 376,040 ₽A negative figure means neighbours grew: a “halo” of the new item.
- Net effect on the category
- 961,960 ₽
- Cannibalisation share by margin
- 25.3 %
- Share of substituted pieces
- 27.3 %What part of the new item’s sales was taken from neighbours.
Neighbours: expected and actual
| Item | Sold before | Expected with trend | Sold after | Lost, pcs | Margin lost |
|---|---|---|---|---|---|
| Style A | 500 | 515 | 420 | 95 | 199,500 ₽ |
| Style B | 380 | 391.4 | 330 | 61.4 | 153,500 ₽ |
| Style C | 260 | 267.8 | 255 | 12.8 | 23,040 ₽ |
The new item expands the category: the net gain is positive. Check that it is not caused by seasonal growth or advertising.
Colours substitute easily, cuts less so: do not average them. Count margin, not revenue, and compare with stores without the item, otherwise the season’s trend will be credited to the item.
Nuances and limits
- Do not count without the trend: the season will be credited to the item.
- Colors substitute easily, cuts less so.
- Look at margin, not revenue.