Retail Media

Retail media margins come from three layers that behave very differently. On-site sponsored listings are the profit engine, typically clearing 70 to 90 percent contribution margin because the inventory already exists and the auction does the pricing. Off-site programmatic extends retailer audiences across the open web and connected TV at far thinner margins, since media is bought rather than owned. In-store retail media, the newest layer, carries real capital cost in screens, networking and content operations, and only pays back where footfall and dwell time are high enough to justify the hardware.

The strategic question for most retailers in 2026 is not whether to launch a network but which layer to own and which to outsource. Mid-size grocers and specialty chains increasingly run on platforms operated by Criteo, Epsilon or Topsort rather than building demand-side technology themselves, keeping the first-party data asset while renting the auction machinery.

Brand budgets have moved faster than measurement standards. Advertisers now run campaigns across a dozen networks that each define an impression, a view and an attributed sale slightly differently, which makes cross-retailer comparison close to impossible. The IAB and MRC retail media measurement guidelines have narrowed the gap on definitions, but adoption is uneven and in-store measurement remains the weakest link.

Incrementality testing is where sophisticated advertisers are pushing hardest. Last-click attribution inside a retailer walled garden reliably overstates contribution, and holdout-based measurement often shows that a meaningful share of attributed sales would have happened anyway. Retailers that publish honest incrementality data are winning share of budget from those that do not.

Every additional sponsored slot on a search results page raises short-term revenue and lowers the relevance of what shoppers see. The failure mode is gradual: conversion holds, then basket size softens, then repeat visit frequency drops, and by the time the effect is visible in trading numbers the ad load has been normalised internally. Retailers with mature networks now cap ad density by page type and run relevance guardrails as a formal constraint on the auction rather than a marketing preference.

This is also where AI is having the clearest operational impact. Ranking models that blend bid, relevance and predicted long-term customer value let retailers hold ad density steady while raising yield, which is a better outcome than simply selling more slots.

Three developments matter most. Retail media is consolidating onto fewer buying interfaces as agencies refuse to manage dozens of bespoke platforms. In-store networks are moving from pilot estates to measured, audited inventory with third-party verification. And commerce media is spreading beyond retailers into travel, banking and telco, which will compete for the same performance budgets with comparable first-party data.