AI Advertising Survey Puts the Customer’s Own Reputation Behind the Referral Question

A consumer may like the product and still hesitate to put their name behind the brand. A new survey suggests that when shoppers discover AI created an ad only after buying, that hesitation can extend to trust, repeat purchases and referrals.

ContentEngine surveyed 446 U.S. adults about a simple change in timing: whether an advertisement disclosed its use of AI before a purchase or revealed it afterward. Every respondent imagined buying an unplanned product and being satisfied with it. The difference came after the ad’s role in the decision had been established.

Respondents who learned about the AI use later were less likely to trust the advertiser or buy from it again. They were also more likely to feel misled. The referral finding moved in the same direction, although it was not statistically conclusive: 67.7% of the later-discovery group said they would recommend the advertiser, compared with 73.1% of those told upfront.

That gap matters because a referral asks consumers to spend some of their own credibility. Recommending a brand to a friend, colleague or family member is a small public endorsement. A consumer may accept an AI-generated ad for a personal purchase and still decide that the brand has not earned a place in someone else’s consideration set.

The survey’s clearest results concerned trust and repeat purchase intent. Referral intent showed a smaller difference, and its confidence interval crossed zero. After the survey’s multiple-comparison adjustment, the recommendation result was not statistically significant.

That limitation should stay attached to the finding. The data do not establish that delayed AI disclosure reduces referrals across the population. They do show a directional pattern that fits the larger trust results:

A recommendation has a different cost from another purchase. If a consumer buys a product again, the decision affects that consumer. If the consumer recommends it, the brand becomes part of that person’s relationship with someone else. A later discovery about undisclosed AI can make the original recommendation feel harder to defend.

For advertisers, that creates a reputation issue beyond immediate conversion. A campaign can persuade a person to click and buy, yet leave that person unwilling to attach their own name to the campaign or the brand behind it.

A ContentEngine survey of 446 U.S. adults suggests that when consumers learn about an advertisement’s use of AI only after making a satisfactory purchase, they may trust the advertiser less and feel less inclined to buy from it again. Referral intent moved in the same direction, though that result was not statistically conclusive. The findings support a practical baseline for advertisers: disclose material, consumer-facing AI involvement before the decision point, explain what AI did, and keep a human accountable for the final ad.

The research used two separate online samples of 223 U.S. adults. One group read a scenario in which the advertisement clearly stated before purchase that its words and images had been created entirely using AI. The second group read that the advertisement persuaded them to buy, that they were satisfied with the product, and that they later learned the ad had been created entirely using AI without disclosing it.

Both scenarios described an unplanned purchase. Both described a satisfactory product experience. That design removed an easy explanation for distrust: dissatisfaction with what the consumer bought.

The difference was the information surrounding the decision. In the upfront scenario, the consumer could account for AI before acting. In the later-discovery scenario, the consumer had already acted before learning about it, and the wording explicitly said the ad had not disclosed the AI use.

That second detail matters. The comparison tested disclosure timing along with the reaction to nondisclosure. It cannot isolate the effect of timing from the effect of being told that the brand had withheld the information. The later-discovery wording may have directly increased feelings of being misled.

Still, the setup reflects a real concern for advertisers. Information provided after a purchase cannot help a consumer evaluate the purchase in advance. A later notice may explain production history, but it arrives after the persuasive work is finished.

Trust was 76.7% in the upfront-disclosure group. That figure combines respondents who said they would trust the advertiser a great deal or a fair amount. In the later-discovery group, the figure fell to 65.0%.

The difference was 11.7 percentage points, with a 95% confidence interval from -20.0 to -3.3 points. The adjusted p-value was .040. Repeat purchase intent followed a similar pattern:

Feeling misled moved in the opposite direction. Fifty-seven percent of respondents who knew about the AI before purchase said they felt misled a lot or some. That rose to 70.0% among respondents who learned about the AI afterward, a 13.0-point difference.

These results suggest that satisfaction does not settle the relationship between a consumer and an advertiser. A buyer can think the product worked and still revise their view of the brand’s conduct. The product answers one question: Was the purchase worthwhile? The disclosure answers another: Did the brand give the consumer information that seemed important before the decision?

The survey does not prove that every delayed disclosure will cause a 12-point trust decline. Its sample was nonprobability, and participants responded to a hypothetical scenario. It does show that a successful transaction can coexist with a damaged view of the advertiser.

The most important lesson is not that consumers reject AI advertising. It is that late disclosure changes the meaning of transparency: information provided after the sale cannot help someone evaluate the decision beforehand. Advertisers should treat disclosure timing, materiality and human review as connected parts of brand trust rather than as compliance details added after creative production.

A disclosure shown before a purchase gives consumers a chance to include AI use in their judgment. Some may care a great deal. Others may care very little. The important point is that the decision belongs to the consumer with the relevant information available.

A disclosure shown afterward has a narrower role. It can document what happened, but it cannot restore the choice that should have preceded the purchase. That is why delayed disclosure can feel less like transparency and more like a correction to the brand’s record.

The distinction becomes sharper when an advertisement uses AI for central creative elements. The survey described an ad whose words and images were created entirely using AI. That is a more substantial use than a tool resizing an image, generating background variations or helping a human writer brainstorm copy.

The International Advertising Bureau’s AI Transparency and Disclosure Framework V2 takes a risk-based, materiality-driven approach to when and how AI involvement should be disclosed. That approach gives advertisers room to distinguish minor production assistance from AI involvement that materially shapes what consumers see and hear. The ContentEngine survey points toward an additional practical rule: when the AI role is central and consumer-facing, the disclosure should appear before the consumer acts.

Recommendation intent deserves careful treatment because it was not one of the survey’s conclusive results. The 5.4-point difference could reflect a real business effect, random variation, differences between the two samples or some combination of those factors.

Even so, referral behavior is useful for thinking about the cost of a trust problem. Consumers do not recommend brands in a vacuum. They often consider whether the product fits the other person’s needs, whether the brand seems reliable and whether the recommendation could make them look careless.

Delayed AI disclosure can raise that last concern. A person who says, try this brand may later wonder whether the recommendation also vouches for the company’s approach to persuasion. The product may remain good. The ad may have been effective. The brand may still have made the recommender feel uninformed.

That is the concern behind the survey’s referral question. It puts the consumer’s own reputation into the transaction. The brand is asking for more than another sale when it seeks a recommendation. It is asking the customer to lend the brand some trust that the customer has built elsewhere.

Advertisers should avoid treating the nonsignificant result as proof of a referral penalty. They should also avoid treating it as permission to ignore the issue. Trust, repeat purchase and recommendation are connected measures, and two of the three most commercially direct measures showed clear differences in this survey.

Transparency helped, but it did not make the scenario comfortable for everyone. Among respondents told upfront about the AI-generated ad, 57% still felt misled a lot or some. About 62% felt manipulated at least somewhat.

That result has several possible explanations. The scenario involved an ad persuading someone to make an unplanned purchase. Some respondents may have viewed the persuasive act itself as manipulative. The scenario also described complete AI creation, which may trigger stronger reactions than partial or routine AI assistance.

The survey did not ask respondents why they felt misled or manipulated. It therefore cannot identify whether those feelings came from AI use, the unplanned purchase, the ad’s persuasive power or the combination.

For brands, the practical lesson is that disclosure is a starting condition for trust, not a substitute for responsible advertising. A label cannot make an exaggerated claim accurate. It cannot make inappropriate targeting acceptable. It cannot answer whether the creative treatment respects the audience.

The disclosure should be clear enough to mean something. A vague phrase such as made with advanced technology may leave consumers unsure whether AI generated the copy, images, voice, targeting or only a minor production element. A more useful notice explains the role AI played in the material the consumer is viewing.

The survey found broad support for human review. Across both samples, 88.3% said it was very or somewhat important to know that a person had reviewed and approved an AI-generated advertisement before publication.

The two groups were almost identical on this question: 87.9% in the upfront-disclosure group and 88.8% in the later-discovery group. The difference was only 0.9 percentage points, and the adjusted p-value was .768.

That consistency suggests that consumers see disclosure and review as separate responsibilities. Disclosure answers how the advertisement was made. Human review answers who remains responsible for what the advertisement says, shows and implies.

A meaningful review should cover more than spelling and visual quality. It should include factual claims, product demonstrations, pricing, endorsements, audience targeting, depictions of people and the possibility of fabricated or misleading details. Someone with authority should be able to stop publication and document the basis for approval.

The European Commission’s explanation of transparency duties under Article 50 of the EU AI Act also places attention on the organization using the system. Employees acting under the authority and control of a legal person are generally not treated as separate deployers merely because they operate the tools. That principle reinforces a straightforward business expectation: responsibility stays with the organization that commissions and publishes the advertising.

Brands using AI in advertising can turn the survey’s findings into a fairly practical process:

The exact format will vary by channel. A short video may need an on-screen notice and accessible text. A product page may need a nearby explanation of AI-generated images or copy. A social post may need the disclosure in the post itself, where viewers will see it before acting.

The standard should also account for the difference between assistance and authorship. AI that removes a background from a product photograph raises a different question from AI that invents the product scene, writes the claims and generates a synthetic spokesperson. Treating every use exactly the same can make notices either meaningless or unnecessarily broad.

Responsible AI advertising moves transparency and accountability earlier in the consumer journey.

ContentEngine’s results are useful evidence about how consumers may interpret disclosure timing, but they are not a population estimate for all U.S. adults. The 446 respondents came from online opt-in panels. The samples were not randomly selected, not weighted and not randomly assigned between scenarios.

The study also compared two separate groups rather than exposing the same respondents to both conditions. Measured demographic profiles did not differ significantly, but that does not rule out differences in attitudes, technology experience or views about advertising.

The scenario was hypothetical. Participants did not see a real ad, spend money or use a real product. Stated trust and purchase intent can differ from marketplace behavior. The study also used complete AI creation, so the results may not apply to every advertisement that uses an AI tool.

Those limits make the findings narrower, not irrelevant. The survey tested a recognizable consumer experience and found meaningful differences in the measures most directly related to the brand relationship. It also showed why delayed disclosure is hard to defend as a trust strategy: by the time the information arrives, the consumer has already made the decision the information could have informed.

The survey’s referral result should be read with caution, but the question itself deserves more attention. Advertisers often judge AI creative by production speed, click-through rate and conversion. Those measures describe what happens during the campaign. They say less about how consumers talk about the brand afterward.

A consumer who learns about undisclosed AI after a satisfactory purchase may still keep the product and may even use it again. The weaker trust score and repeat-purchase result suggest that the brand relationship can deteriorate before the product experience does. The referral question adds a social dimension: the customer may choose not to put personal credibility behind the advertiser.

That is why early disclosure is a reasonable baseline for brands using AI to create consumer-facing advertising. It gives people the information before the decision, keeps the advertiser’s process open to scrutiny and gives human review a visible place in the chain of responsibility.

The survey does not show that every consumer will reject AI advertising. It shows that consumers are paying attention to when brands tell them and whether a person remains accountable. A purchase can be successful on paper while the customer quietly decides that the brand is no longer worth recommending.