Cannibalization

The halo effect's unglamorous sibling. Halo is spillover that creates value in another channel; cannibalization is substitution that merely relocates it, with an invoice attached. The two coexist: TV can lift branded search (halo) while paid branded search cannibalizes the organic clicks that lift would have produced anyway.

The mechanism inflates apparent ROAS because the counterfactual is wrong. If 80% of branded-search conversions would have arrived through the organic listing below the ad, the true cost per incremental conversion is five times the reported figure. The same logic applies to paid app installs versus store search, and promo emails versus customers who would have purchased at full price.¹

Can an MMM detect it? Partially. A model with organic traffic or organic revenue as an explicit variable can observe paid brand spend rising while organic conversions fall, and estimate the substitution. But the definitive instrument is the experiment: a brand-search holdout, pausing brand ads in a market subset and watching how much revenue organic recaptures.² This is among the cheapest, highest-value experiments in marketing, which is presumably why so few brands run it.