A Recommendation Has to Be Earned: 34% Would Be Very Likely to Recommend the Brand in AI Ad Research

A successful sale can still leave a brand with a trust problem. New survey findings suggest that when consumers learn about AI involvement only after buying, they become less willing to recommend, trust, or buy from that advertiser again.

A brand can make a persuasive advertisement, sell a product a customer enjoys, and still lose ground after the customer learns how the ad was made. That is the central tension in ContentEngine research on AI-generated advertising. Among respondents who saw a scenario in which the advertiser disclosed AI use upfront, 34% said they would be very likely to recommend the brand. The broader recommendation measure, which combined very likely and somewhat likely responses, reached 73.1% in that group. Among people who learned about the AI involvement afterward, the combined figure fell to 67.7%.

The recommendation gap was smaller and less statistically conclusive than the differences in trust, repeat purchase intent, and feeling misled. It still points to an important practical issue: customers decide whether to speak positively about a brand based on more than product satisfaction. They also judge how the brand handled information that could have affected the purchase.

ContentEngine surveyed two separate groups of 223 U.S. adults, for 446 respondents in total. Each group read a short scenario about an online advertisement for a product the person had not planned to buy.

Both scenarios included the same positive outcome:

The timing of the AI information changed between the groups. One group was told before the purchase that the advertisement’s words and images had been created entirely with artificial intelligence. The other group learned after the purchase that AI had created the advertisement and that the ad had not disclosed this fact when it appeared.

Respondents then rated the advertiser on eight measures, including trust, repeat purchase intent, regret, feeling misled, recommendation, feeling manipulated, the importance of AI disclosure, and the importance of human review.

The recommendation results were:

That last detail matters. The survey gives a useful signal about how disclosure timing may affect word of mouth, but the recommendation result carries more uncertainty than the results for trust, repeat buying, and feeling misled.

ContentEngine’s survey suggests that a successful purchase does not guarantee lasting customer advocacy when consumers learn afterward that an advertisement was created entirely with artificial intelligence. Respondents told about the AI use before buying reported higher trust, repeat purchase intent, and recommendation likelihood than those who learned later, although the recommendation difference was not statistically conclusive after adjustment. The findings support a practical standard for advertisers: disclose material AI involvement before the consumer acts, maintain documented human review, and measure relationship outcomes beyond clicks and conversions.

The 34% figure describes respondents who selected the strongest recommendation response: very likely to recommend. That is a narrower measure than the 73.1% top-two figure for the upfront group.

A top-two result can show general openness to recommending a brand. A very likely result shows stronger conviction. That distinction is useful for marketers because many customers who are willing to recommend a company may still need a reason to actively praise it, refer a friend, post a review, or mention it in conversation.

The survey does not provide the full breakdown for each recommendation category in the supplied report, so the 34% figure should not be treated as a complete measure of advocacy. It does, however, frame the headline in a grounded way: a sizable minority expressed strong enthusiasm, while the wider group became somewhat less willing to recommend after discovering undisclosed AI use.

Recommendation also tends to involve a social risk. When a customer recommends a brand, that person is attaching some credibility to the recommendation. If the brand later appears to have withheld a meaningful detail about its advertising, the customer may become more cautious about endorsing it.

That caution can show up in several ways:

The survey did not measure those specific behaviors. It measured stated likelihood. Still, the distinction helps explain why recommendation deserves separate attention from immediate conversion.

The largest business implications appeared in the measures tied directly to the customer relationship.

Among respondents who knew about the AI-generated advertisement before buying, 76.7% said they would trust the advertiser a great deal or a fair amount. That fell to 65.0% among respondents who learned about the AI use afterward, a gap of 11.7 percentage points.

Repeat purchase intent followed a similar pattern:

Feeling misled moved in the other direction. Fifty-seven percent of the upfront group said they felt misled a lot or some. The figure rose to 70.0% among people who learned about the AI use later, a 13-point gap.

Those three results remained statistically significant after the report adjusted for the eight comparisons. The recommendation difference did not meet that same standard after adjustment.

For a brand, the pattern creates a practical warning. A campaign can perform well at the moment of purchase and still weaken the next stage of the customer relationship. The customer may like the product and remember that the ad worked. The later discovery can still change how the customer evaluates the advertiser’s judgment and candor.

The most important lesson is not that every AI-generated advertisement needs the same label or that disclosure automatically creates trust. It is that brands should treat AI transparency as part of the customer experience, not as a post-purchase footnote. Consumers appear to care both about how an ad was made and who was accountable for approving it. Advertisers that disclose meaningful AI use at the point of decision, explain its role accurately, and test trust alongside performance metrics will be better positioned to earn recommendations rather than merely generate transactions.

An AI label shown before a purchase gives consumers information while they are deciding. A disclosure that arrives afterward serves a different purpose. It explains the production process after the decision has already been made.

That timing can affect how people interpret the same fact. Before a purchase, a consumer can decide whether AI involvement matters. After a purchase, the consumer may wonder why the information was withheld and whether other relevant details were also left out.

The later-discovery scenario explicitly said the advertisement had not disclosed its AI use. That wording likely influenced the high rate of respondents who felt misled. The study authors acknowledge this limitation. The survey cannot isolate timing from the experience of nondisclosure because both appeared together in the second scenario.

Even with that limitation, the consumer expectation is straightforward. If AI involvement is material enough for a brand to discuss after the fact, consumers may reasonably expect to hear about it before they act.

The European Union’s AI Act takes a related approach in its transparency rules for certain AI-generated or manipulated content. The European Commission explains that the legal person operating an AI system under its authority, such as an advertising company, remains the relevant deployer in that context. The details of legal obligations depend on the use case and jurisdiction, but the broader direction is clear: responsibility stays with the organization using the system.

The IAB’s AI Transparency and Disclosure Framework V2 also recommends a risk-based, materiality-driven approach to consumer-facing disclosure. That approach leaves room for different labels and placements depending on how AI was used and how much that use could affect a person’s interpretation.

Disclosure was not the only information respondents wanted. Across both groups, 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 result was nearly identical across the two samples:

That consistency suggests consumers separate two questions:

How was this advertisement made?

Who is responsible for what it says and shows?

An AI disclosure answers the first question. Human review helps answer the second. A person reviewing the ad should be responsible for checking claims, images, targeting choices, disclosures, tone, and the possibility that the system produced something misleading or inappropriate.

A small label saying “AI-generated” cannot carry that entire burden. It can tell the audience about the production method, but it does not confirm that anyone checked the result carefully. Brands that publish AI-created ads should be able to identify the human or team responsible for approval, even if that internal process is not displayed in the ad itself.

The survey does not support a simple assumption that early disclosure resolves every concern. Even among respondents told about the AI-generated ad before buying, 57% said they felt misled a lot or some, and 62.3% said they felt manipulated a lot or some.

Several parts of the scenario may explain those reactions. The purchase was unplanned, and the advertisement successfully persuaded the respondent to buy. The ad was also described as having been created entirely with AI, which is a stronger condition than using AI to resize an image, generate several headline options, or assist a human copywriter.

The survey did not ask respondents why they felt misled or manipulated. Their responses could reflect the persuasive nature of the advertisement, the unexpected purchase, the complete use of AI, or a combination of those factors.

That uncertainty has a practical implication. A disclosure should be treated as one part of responsible advertising. It does not excuse exaggerated claims, aggressive targeting, synthetic testimonials, unclear pricing, or creative choices that would feel uncomfortable if consumers understood how they were produced.

Advertisers deciding how to handle AI-generated creative can translate the research into several operating rules.

A disclosure placed in a terms page, post-purchase email, or help center may satisfy an internal checklist while arriving too late for the consumer. The relevant question is whether a reasonable person can see and understand the information before responding to the ad.

Placement will vary by format. A video may need an opening or persistent label. A social post may need a visible caption rather than a buried note. A landing page may need a clear statement near the creative and call to action.

“Created with AI” can mean many things. It may refer to a fully synthetic image, AI-written copy edited by a human, automated audience targeting, or minor production assistance. A useful disclosure should describe the material role AI played instead of using a vague label that leaves consumers guessing.

Human review should cover more than spelling and brand style. A documented review can check:

Click-through rate and conversion rate can miss a delayed trust cost. Campaign testing should include questions about credibility, comfort with the disclosure, willingness to buy again, and willingness to recommend the brand.

A campaign that produces a strong first purchase but lowers repeat intent may be expensive in ways that a platform dashboard does not show. The ContentEngine results reinforce the value of measuring the period after a consumer learns how the ad was created.

If a customer asks whether AI was used, a vague answer can deepen suspicion. A useful response should identify what AI produced, what people reviewed, and how the brand checked the content before publication. The explanation should match the actual workflow.

The findings describe two groups of survey respondents, not the entire U.S. adult population. The samples were nonprobability online panels, were not weighted, and were not randomly selected from all U.S. adults. The two scenarios were also assigned to separate samples rather than randomly assigned to respondents within one experiment.

The scenario was hypothetical. Respondents did not see a real advertisement, spend money, or use a real product. Stated trust and recommendation intent may differ from behavior in a marketplace.

The survey also compared advertisements created entirely with AI. Results may differ when AI plays a smaller role, such as helping a copywriter brainstorm, resizing an image, or producing alternate versions for testing. The report did not measure race, ethnicity, or education, and it did not ask people to explain their answers.

Those constraints call for careful use of the numbers. The research does not prove that every form of AI disclosure will increase recommendations by a particular amount. It does show a consistent concern worth testing in real campaigns: consumers may care about when they learn about AI involvement, and that information can influence how they judge the advertiser after a successful purchase.

The 34% very-likely recommendation figure offers a useful reminder for brands using synthetic creative. A recommendation is stronger than a completed transaction. It asks a customer to lend personal credibility to the company.

The survey’s broader results show why that credibility needs to be protected. Upfront disclosure was associated with higher trust and repeat purchase intent, while later discovery was associated with more reports of feeling misled. Human review received nearly the same level of support as AI disclosure itself.

For advertisers, the operating standard is simple: disclose meaningful AI use before the consumer acts, keep a person accountable for the final work, and judge campaign quality by the relationship that remains after the sale. A product that satisfies the customer is valuable. A transparent process gives that customer a stronger reason to trust the brand and recommend it.