Retail media is moving offsite. The measurement has not caught up

Onsite retail media keeps the page, audience and transaction close together. Offsite separates them. The ad may run on connected TV, streaming audio or the open web while the retailer holds the purchase data and the advertiser's media team owns the buying decision.

The move offsite is a capacity problem

Retailers cannot keep adding sponsored placements to shopping surfaces without making those surfaces harder to use. The next inventory is elsewhere: connected TV, streaming audio, social platforms and open-web programmatic. That makes offsite a distribution problem as much as an inventory opportunity.

Advertiser Perceptions estimated that US offsite programmatic retail media could exceed $20 billion in 2024, up from $7.5 billion in 2023. eMarketer's forecast was lower because it used a narrower definition, but pointed in the same direction. The disagreement is a useful warning: the category is growing faster than its definitions are settling.

By 2025, eMarketer estimated offsite retail media at $14 billion, around 20 percent of retail media spend. The exact total depends on whether social, CTV and managed-service activity are counted. The strategic movement does not depend on one forecast.

Walmart's acquisition of Vizio and Amazon's expansion of Prime Video advertising made the CTV connection visible. Purchase data is becoming an input to media bought away from a retailer's own site, but the data, inventory and outcome still sit in different systems.

Offsite has different economics

Onsite retail media benefits from high purchase intent and strong margins. Marketing Dive reported estimates of 80 to 90 percent gross margins for onsite activity, compared with 20 to 40 percent for offsite. Those figures vary by retailer and service model, but the direction changes the commercial conversation.

Offsite inventory adds work at each handoff. A retailer passes an audience into a demand-side platform or media partner. The advertiser then has to identify the supply path, understand the segment definition and test whether the resulting sale was incremental.

That is not a reason to avoid offsite. It is a reason to stop comparing every placement using the onsite sponsored-product benchmark.

CTV changes the job

A sponsored product appears next to a search or category decision. A CTV impression appears during a programme, often before the shopper has chosen a product. The creative, frequency rule and success measure have to reflect that distance.

Retail networks can bring purchase-based audiences, household penetration data and a link between exposure and later sales. The audience is not the whole proposition, though. A CTV impression still depends on the creative, programme, frequency and moment in which someone sees it.

Amazon, Walmart and other networks are not equal here. Amazon has its own streaming environment. Walmart bought Vizio to secure more of one. Smaller networks need partnerships, which adds another layer between the data owner and the inventory.

Measurement that travels with the data

ROAS from a retailer's onsite search report cannot carry the whole account once the campaign runs across CTV, audio and the open web. A useful offsite scorecard needs to hold four views together:

1. Delivery quality: reach, frequency, completion, viewability and invalid traffic.
2. Audience quality: who was reached, how the segment was built and how much overlap existed.
3. Commerce effect: incremental sales, household penetration, new-to-brand behaviour and geographic lift.
4. Supply quality: publisher, app, programme, deal path and the proportion of spend that reached working media.

Kroger Precision Marketing's 2025 expansion into programmatic audio, CTV and dynamic creative shows where the market is going. Its managed-service offer pairs purchase-based audiences with retail measures such as incremental return on ad spend, household penetration, uplift, dollar sales and unit sales.

The buying decision

Before moving budget offsite, make the test answer three questions.

• Does the retailer have a credible audience product for the category?
• Can the network expose the supply path and report outcomes beyond clicks?
• Can the brand compare the result with its existing CTV, audio and programmatic activity?

If the answer to the third question is no, the test will produce a retailer report rather than a media decision.

Offsite gives retailers more inventory and gives advertisers another way to use purchase data. It should be judged as a media channel with retail evidence, not as a new name for onsite search.