Revlon used Amazon’s Dynamic TV Creative to make a streaming campaign more responsive to where viewers appeared to be in the shopping journey. The system kept the core video asset intact while changing interactive elements such as the format, headline, and call to action.
The campaign is a useful test of a larger shift in connected TV. Personalization is moving beyond audience selection and into the ad experience itself. For marketers, that creates a new opportunity to connect brand storytelling with commerce signals, but it also makes measurement and creative governance more demanding.
Table of contents
Jump to each section:
- What Revlon changed in streaming TV
- Why personalization is moving into the ad format
- The measurement case needs a control group
- What marketers should know about AI-personalized TV
What Revlon changed in streaming TV
Revlon’s June campaign combined Amazon Interactive Video Ads with Dynamic TV Creative. Viewers still saw the brand’s main video, while Amazon adapted the layer around it according to shopping signals and the action a viewer appeared more likely to take.
That distinction matters. Revlon did not hand the entire brand story to a generative system. The company preserved a carefully produced asset featuring its brand ambassadors, then allowed the platform to vary selected components around that asset.
The model turns one piece of creative into several possible routes. A viewer ready to buy may receive a more direct action, while someone earlier in consideration may be invited to learn more without leaving the streaming experience. Amazon describes the setup as a one-click activation after the advertiser uploads an asset and opts into interactive optimization.
The most interesting part of the test is not the number of variations. It is the separation between what the brand fixes and what the platform can change.
That separation gives marketing teams a clearer governance question. Instead of asking whether AI should make the ad, they can decide which parts of the ad are allowed to respond to data and which parts must remain consistent.
Why personalization is moving into the ad format
Digital advertising has long personalized who receives an impression. Dynamic TV Creative pushes the same logic into what happens after the impression appears, using Amazon’s commerce signals to adjust the interactive experience around the video.
The common assumption is that streaming personalization requires producing many distinct commercials. Revlon’s test points to a different reality: the expensive brand asset can remain stable while a narrower decision layer changes. The strategic implication is that creative scale may depend less on generating more footage and more on defining controlled fields that can vary safely.
Adaptive creative is becoming a rules problem before it becomes a production problem.
This also changes how brand and performance teams meet. Revlon wanted the emotional reach of a video format, but it also wanted the campaign to move viewers closer to a commercial action. Interactive overlays provide that bridge without forcing every viewer into the same next step.
The approach carries a familiar platform tradeoff. Amazon can connect viewing, shopping, ad delivery, and reported outcomes inside one environment. That integration can reduce friction, but it also gives the platform substantial influence over the signals used to personalize the experience and the metrics used to judge it.
For marketers, convenience should not replace clarity. Teams still need to know which variables changed, what audience logic informed those changes, and which results came from the interactive format rather than from Amazon’s broader media and commerce ecosystem.
The measurement case needs a control group
Amazon shared strong results from the campaign, comparing some outcomes with non-interactive ads and return on ad spend with category benchmarks.
9.8x higher click-through rate, 2x higher branded search, and 2.6x higher return on ad spend Amazon reported these results for Revlon’s interactive campaign versus non-interactive ads or category benchmarks.
Those figures make the test commercially relevant, but they do not settle the causal question. Click-through and branded-search comparisons may indicate that the interactive experience prompted more action. A category benchmark for return on ad spend is a weaker substitute for a matched control because campaign timing, audience mix, product demand, and media placement can all affect the result.
A higher response rate can show that an ad invited action effectively. It does not automatically show that personalization created incremental demand.
The next useful test would isolate the adaptive layer. Revlon could compare the same core video across matched audiences, with one group receiving a fixed interactive treatment and another receiving the dynamically selected version. That would help distinguish the value of interactivity from the additional value of AI-led personalization.
Measurement design matters more as platforms combine targeting, creative adaptation, delivery, and attribution. When one system performs every step, marketers gain operational simplicity but lose some independent points of comparison. Strong results should therefore lead to a better experiment, not to the end of scrutiny.
What marketers should know about AI-personalized TV
Revlon’s campaign shows how AI can make streaming creative more responsive without replacing the asset at the center of the brand story.
Lock the non-negotiables. Decide which visuals, claims, ambassadors, and narrative elements cannot change. Give the system a defined range of variation rather than an open creative brief.
Test the adaptive layer. Compare dynamic and fixed interactive treatments using matched conditions. This reveals whether personalization adds value beyond the basic ability to click, shop, or request more information.
Inspect the decision signals. Ask which shopping and behavioral inputs determine the format or call to action. A simple activation for the buyer can still depend on complex platform logic behind the screen.
Connect media to product content. Revlon’s marketing leadership also emphasized that product detail pages need to answer both human questions and long, specific AI queries. The promise of a responsive ad weakens if the product information behind it is incomplete or inconsistent.
The deeper shift is that television creative is starting to behave more like a configurable product. A brand can preserve one recognizable story while allowing selected elements to respond to context, intent, and commerce data.
That gives marketers more control in one sense and less in another. They can define the boundaries of variation, but the platform increasingly decides which version a viewer receives and helps evaluate whether it worked.
The durable advantage will not come from switching on personalization first. It will come from designing creative systems whose variable parts, evidence standards, and decision rights are clear enough to improve over time.