Data report: Does disclosing AI-generated product video ads change brand trust, perceived quality, and purchase intent? A 3-minute consumer test using matched on-brand ecommerce creatives made in Lamina

Will disclosing AI-generated product video ads cut brand trust, perceived quality, or purchase intent?

The supplied evidence has no Lamina-specific, three-minute consumer test that can settle this yet. The strongest available video-ad study found no significant label-driven shift across several brand and creative outcomes; other research shows disclosure can prompt inferences of lower effort and lower quality. Treat disclosure as something to test, not a verdict.

The risk is more specific than a blanket reaction to the words “AI-generated.” In a polished video, a label may leave brand choice and perceived production quality alone while changing whether a shopper sees the product as carefully made or the brand as candid. Those are separate calls. Ask separate survey questions.

For ecommerce teams, the workable position is conditional: get the execution on-brand first, disclose clearly second, then hold every asset at a human review gate. AI generation can deliver the styled product motion, material detail, and campaign variants in the brief. Your test needs to show which disclosure treatment keeps the brand signals that matter to you intact.

What did the existing evidence on video-ad disclosure actually find?

MediaScience found a neutral result on the outcomes it measured: four label formats showed no significant difference from an unlabeled control for brand choice, ad memory, brand recognition, brand attitude, ad liking, or perceived production quality. That should reassure you about a well-made product video. It does not prove that trust in one ecommerce brand, or perceived quality of one SKU, can never shift.

The study tested early text, delayed text, full-duration text, and a full-duration icon. Test placement and persistence; “disclosed” is not one condition. The supplied report does not tell you which treatment works best for Lamina-made creative, or whether the results carry into a short, product-specific shopping context.

Dr. Duane Varan’s comment matters for one reason: the study’s finding depends on creative quality. That is the boundary, not a footnote.

Why might AI disclosure still put quality at risk?

AI disclosure can lower engagement if viewers decide the communication took less effort and the featured product is lower quality. The eight-study research program in the supplied evidence names those two perceptions as the mechanisms. Put both in an ecommerce test; do not bury them inside one favorability score.

Production cues change the outcome. That research says human-involvement cues can amplify the negative effect, while automated-production cues can reduce it; other image-label experiments found a competing pattern, with labels raising novelty and lowering authenticity. A disclosure line belongs to the creative treatment. Pretest the wording.

Do not use click-through rate as a stand-in for these perceptions. Research using matched display-ad visuals ties stronger click performance to AI-created ads that did not appear AI-generated, yet it never measures disclosed video, brand trust, product-quality perception, or purchase intent. It supports disciplined visual polish, nothing more.

How do you run a three-minute matched ecommerce video disclosure test?

Run a randomized, between-subjects test: every participant sees the same Lamina-made product video, and only the disclosure treatment changes. Keep the product, edit, offer, landing-page context, copy, sound, and viewing device fixed. Otherwise you are measuring creative preference, not the label.

Keep the task narrow. A three-minute session can capture an immediate reaction and a handful of decision-useful outcomes; it cannot establish long-term trust, actual conversion, or repeated-exposure effects. Use it to choose the next creative treatment, then check the selected version in live campaign measurement.

Which measures determine whether an AI disclosure treatment is safe to publish?

Publish a disclosure treatment only if it protects brand trust, perceived product quality, and purchase intent together. Strong ad liking does not cover a product-quality decline in ecommerce. Shoppers are judging the item as well as the entertainment.

Keep perceived production quality as a supporting measure, not a proxy for perceived product quality. MediaScience measured the former; the separate disclosure research identifies lower product-quality inference as a possible pathway. Too many ad tests blur that line.

Segment by product type if the sample can support it. The supplied image-label research found novelty’s upside was stronger and authenticity’s downside weaker for utilitarian products. A functional product and a status-led product should not automatically run under one disclosure policy.

What should ecommerce teams do now?

Use AI-generated product video, disclose it where policy or audience calls for it, and validate the treatment with matched creative before you scale spend. Existing evidence permits both a neutral outcome and a perception penalty. Categorical claims would be careless.

Start at the visual standard shoppers already know from your brand: accurate product representation, deliberate styling, readable offer language, and zero stray synthetic artifacts. The available matched-ad research backs polished, on-brand execution as the operating priority, though it cannot answer the disclosure question on its own.

A clean test leaves a decision record you can reuse. Creative, legal, and media teams get an answer on whether one label format works for one product category and audience. That beats asking respondents if they like AI in the abstract.