A brand delivers one 15-second product film in landscape, vertical, and square. The files pass every upload check. Yet the large-screen placement expects sound and a sustained story, the retailer product page is often browsed silently, and the in-store screen gets only a few seconds of a shopper’s attention.
The problem was never a missing aspect ratio. It was treating three viewing conditions as one creative job.
Two developments in late September 2026 make this a procurement issue, not merely an editing preference. IAB has finalized a new way to classify video advertising around consumer viewing experience. Meanwhile, an independent UK review of major retail media networks found that core measurement definitions still vary materially. Brands therefore need portability in two directions: the creative must survive different viewing environments, and the reported outcome must survive a definition check.
IAB’s important sentence: a platform is not an attribute
On September 28, 2026, IAB published A New Standard for Classifying Video Advertising. The standard responds to a market spanning streaming, CTV, social video, gaming, retail media, video podcasts, and public displays, where legacy channel labels overlap and do not consistently describe how an impression is experienced.
The framework has two layers. The top-down layer classifies large-screen communal viewing, personal screens, and public ambient displays. The bottom-up layer adds operational attributes including sound default, skip capability, screen presentation, audience decisioning, device, ad format, screen size, content source, and available measurement signals.
Its most useful production implication is explicit: a platform is not an attribute. One platform can distribute inventory across multiple media types. The same content can appear on a living-room television, a personal phone, and a store screen. The seller may be the same; the viewer’s posture, audio expectation, competing information, and path to action are not.
That changes the brief. “Make an Amazon video” or “adapt this for social” is no longer enough. A production team needs to know:
- Is the viewer intentionally on a product page or moving through a feed?
- Is the experience personal, communal, or public?
- Does meaning survive when sound is off?
- Can the viewer skip, scroll past, or must the unit complete?
- Does the video own the screen or compete with price, reviews, and other products?
- Is the shopping action inside the unit, nearby on the page, or several steps away?
These questions change the shot and edit, not just the export preset.
“Attributed sales” may not mean the same thing twice
Creative portability is only half of the problem. Measurement definitions also change by network.
On September 8, ISBA and MediaSense published the second phase of their Digital Retail Media Study. The review validated more than 200 capability data points across five major UK retail media networks. It found meaningful differences in metric definitions, attribution methods, reporting transparency, buying models, and service levels.
Business Insider reported on September 30 that the audit surfaced five versions of what qualifies as an attributed sale across the five networks. That is a report about a defined UK sample, not proof that every network in every market behaves the same way. It does illustrate why two ROAS figures may not be comparable.
IAB Europe’s Commerce Media Measurement Standards V2.1 attempts to close that gap by refining the measurement funnel, standardizing gross and net sales definitions, formally defining incrementality, expanding new-to-brand timeframes, and setting a 30-day default lookback while still requiring flexible options.
Before a creative team crowns a winning cut, the brand should confirm:
- whether sales cover the advertised SKU, the brand, or a wider category;
- whether exposure or click starts the attribution path;
- the lookback window;
- how cancellations, returns, discounts, and taxes are treated;
- how “new to brand” is defined;
- whether the report shows attributed sales or estimates incremental sales using a counterfactual.
When those conditions differ, optimize within each network first. Do not build a cross-platform leaderboard from labels that happen to share a name.
The ONCE Retail Video Portability Brief
A portable system does not force one master file through every uploader. It separates reusable product evidence from environment-specific storytelling. We recommend five layers.
1. Product Truth Core
Lock the facts that must remain stable everywhere: appearance, real functions, scale, included items, operating steps, limitations, and approved claims. Capture proof shots as reusable source material, not only as moments buried in one final timeline.
2. Viewing Environment Card
For each placement, record screen scale, personal/communal/public viewing, sound default, skip behavior, screen share, and distance to purchase. This description remains useful even when a platform adds a new surface.
3. Capture Modules
Film modular evidence: product actions that work silently, standalone proof shots, compositions safe for multiple crops, clean opening and closing holds, and audio-led material that earns its place when sound is expected.
Shared facts can share a shoot. Different viewing mechanics need different expression. Communal large-screen viewing can support atmosphere and audio rhythm. A personal shopping surface must identify the product and problem earlier. A public display needs an action that reads during a brief pass.
4. Variant Ledger
Every deliverable gets an Asset ID, viewing environment, primary creative variable, aspect ratio, duration, caption policy, sound strategy, action path, and market/rights scope. Change one major variable at a time when the goal is learning; otherwise a performance difference can select a file but cannot explain a decision.
5. Measurement Contract
Define exposure, view, attributed sale, new customer, returns treatment, and incrementality before launch. A production company does not need to become the brand’s attribution vendor. It does need to understand what evidence will decide whether a version is reused, revised, or retired.
Crop, re-edit, or reshoot?
Adding a retail media network does not automatically justify a new production day.
Crop only when the product, proof detail, and motion survive the new frame; audio conditions and viewing pace remain compatible; and the action path does not alter the story.
Re-edit when the factual footage still works but sound default, attention window, skip behavior, or purchase proximity changes. Rebuild the opening, information order, caption density, pace, and call to action.
Reshoot when the proof disappears in the crop, the existing footage cannot communicate silently, vertical composition was never protected, a public screen cannot reveal the product quickly enough, or a new market requires different talent, packaging, language, or compliance treatment.
The budgeting question is not “How many platforms?” It is “Which new viewing tasks cannot the existing evidence perform?”
Buy a portable decision system, not a folder of files
A stronger retail media video contract should cover approved product truths, environment cards, modular source footage, the variant ledger, usage rights, and measurement definitions. Final-file count still matters, but it sits at the end of the system.
This framework does not eliminate attribution uncertainty. IAB’s classification is an industry standard, not proof of universal platform adoption. The ISBA study covers five UK networks, not the entire global market. Those limits are precisely why brands need explicit definitions instead of confidence borrowed from a dashboard.
If the next launch spans three or more retail, social, and large-screen environments, put their viewing conditions and sales definitions on one page before asking for a video count. That page will reveal which assets are genuinely reusable—and which “cheap adaptation” would simply move the cost downstream.
Related: When AI Can See Ads, Storefronts, and Books, Video Needs a Decision Trail
