Cannibalisation: when a launch moves its own volume

A new item's sales are easy to count. Whether those sales are additional to the brand, to the category, or to the retailer is a different question, and it is the one that decides whether the launch was worth the shelf space it took.

Cannibalisation is the portion of a new item's volume that came from existing items rather than from new demand. It is normal, it is expected, and it is routinely left out of launch reporting.

Where does the volume come from?

Three possible sources: a competitor, the category's growth, or the brand's own existing items. The first two are gains; the third is a transfer. A launch report that counts only the new item cannot distinguish them.

The honest measurement looks at the brand's total volume before and after, in stores that took the new item, against stores that did not.

What does a clean measurement need?

A comparison group. Stores that stocked the new item and stores that did not, matched on the characteristics that matter, over the same period. The difference in brand total between the two groups is the incremental effect.

Without a comparison group, a launch in a growing category shows a brand total that rose, and the rise gets attributed to the launch.

Is cannibalisation always bad?

No, and treating it as a failure produces bad decisions. A launch that cannibalises while defending shelf space from a competitor, or that moves a shopper to a higher-margin item, can be a good commercial outcome with high cannibalisation.

What matters is that the number is known. A launch justified on total new-item sales and quietly cannibalising most of it is a decision made on a number that does not exist.

How does shelf space complicate it?

Because a new item usually displaces something. In a small-format store the shelf is genuinely fixed, so a listing is a substitution rather than an addition, and the correct comparison includes what left.

That is one of the distinctive features of measuring in the independent channel. In a large-format store an additional item may be additional; in a bodega it almost never is.

What about the retailer's view?

A retailer cares about category total, not brand total. A launch that grows a brand by cannibalising a competitor is a success for the brand and neutral for the retailer, and a launch that grows the category is valuable to both.

Knowing which conversation you are in determines which number to bring. Bringing brand-level incrementality to a category review is a common and avoidable mismatch.

How long should the window be?

Long enough to pass the trial period. Early weeks of a launch are disproportionately trial, and trial volume is not repeat volume. A cannibalisation estimate taken in week four measures curiosity.

Repeat behaviour settles over a longer horizon, and it is the settled level that tells you what the item is worth.

What should be agreed before launch?

The comparison group, the window, the measure, and what result would trigger a delist. Agreeing the delist criterion in advance is the part that is almost never done and the part that makes the measurement consequential.