All articles

Published 3 March 2026 · Jeremy Agoya

Cannibalization analysis: how to open a second store without eating the first

Overlap is not automatically cannibalization. A practical method for estimating transfer, deciding when it is acceptable, and spotting when a new unit is simply moving revenue.

Every expansion team eventually faces the same question: this site looks strong, but how much of its revenue will come out of the store we already own two miles away? Cannibalization analysis answers it before the lease, not after.

Overlap is the symptom, not the diagnosis

Two trade areas can overlap by 30% and transfer almost nothing if the shared households are already underserved at peak. They can overlap by 15% and transfer heavily if the existing unit draws most of its volume from exactly that band. Start with overlap, then weight it by where the incumbent's demand actually sits.

A workable estimation method

  1. Build drive-time trade areas for the incumbent and the candidate.
  2. Compute the population and spend in the intersection.
  3. Estimate the incumbent's share of intersection demand from its current volume, not from area alone.
  4. Apply a distance-decay assumption: households closer to the new unit switch at a higher rate.
  5. Net the result against incremental demand the new unit reaches that the incumbent never did.

When cannibalization is worth it

Transfer is acceptable when it buys capacity, defence or coverage. If the incumbent turns customers away at peak, a nearby unit converts lost demand into revenue. If a competitor is circling the same corner, a defensive open protects the whole cluster. If the second unit shortens drive times enough to raise total category frequency, the network grows even as the first store dips.

Red flags

Be sceptical when the candidate's incremental population is small, when the intersection contains the incumbent's highest-value households, or when the pro forma only clears hurdle rate by assuming zero transfer. Model a 10%, 20% and 30% transfer case and check whether the decision survives all three.

Reporting it

Present cannibalization as a range with explicit assumptions, tied to a map of the overlap. Property committees approve ranges they can interrogate; they reject single numbers they cannot.

Feedback