Measuring the true added value of each channel
A marketing team runs paid search, email, social media, and content marketing simultaneously. Attribution models credit each channel based on touch order, but the real question is: what conversions happen because of each channel that wouldn't happen without it? This is incrementality. A customer who was going to search for the product organically and convert anyway, but happens to see a paid search ad first, is not incremental. The paid ad didn't drive behavior, it just claimed credit.
Incrementality is measured through holdout testing: randomly prevent the channel for a portion of customers (stop showing them ads, pause email, etc.) and compare conversion rates. If conversions drop 8% when paid search is paused, paid search is 8% incremental. If conversions don't drop when email is paused, email is 0% incremental (customers would convert anyway via other channels).
Cannibalization and the mix problem
Some channels boost each other (paid search ads drive brand awareness that helps email open rates). Others cannibalize: a paid ad reaching a customer who would have clicked organic search takes credit but adds no conversion. If a customer was worth 100 conversions in organic search, and paid ads cannibalize 10 of those (reaching the same customer but getting paid credit), the net incrementality of paid is 10, not 100. Mapping incrementality per channel reveals where to spend and where to cut. A channel with high incrementality and low cost-per-conversion is a clear winner. A channel with negative incrementality (cannibalizes more than it drives) should shrink.
Incrementality also shifts by saturation. Early in a campaign, paid search is highly incremental (reaching new audiences). After market saturation, it cannibalizes organic and direct traffic. Mix and volume matter: the optimal strategy might be high spend on paid search until it hits diminishing returns, then reallocate to channels with lasting incrementality.