Three Brand Questions to Ask Before Launching an AI Ad

AI has made ads faster and cheaper to produce, so speed no longer sets one advertiser apart. Generic AI output now goes by a nickname, AI slop, and shoppers notice it. What sets a brand apart is whether a shopper sees an ad and recognizes the brand behind it. A tool can’t make that call, which leaves it with the brand. Before an AI-assisted ad runs, the team signing it off needs to know what on-brand looks like and how to check for it.

Source: McDonald’s Netherlands Christmas ad, YouTube, December 2025

Shoppers notice AI-made ads, and they judge what they see

A 2026 Bentley University-Gallup survey of 3,270 U.S. adults found that 79% had seen ads in the past 30 days that looked AI-made. Nearly half (49%) view businesses’ use of AI to create ads negatively, while 19% view it positively.

McDonald’s Netherlands saw how that judgment plays out. In December 2025 it pulled an AI-made Christmas ad from YouTube three days after launch, once viewers called it creepy and soulless. The production company said up to ten specialists spent seven weeks on it. The effort didn’t change the verdict, since viewers only saw the finished ad. A pre-launch check has to take that same view.

Shoppers welcome AI for drafts and push back on AI-made people

The same survey shows where the line sits. Three in four Americans (75%) say it is acceptable for businesses to use AI to help employees brainstorm ideas or create early drafts of ads. That holds when the business clearly discloses the use. A smaller majority (53%) accept it for the final text, images or videos. But 62% call AI-created people or voices unacceptable, even when the business discloses that AI was used.

Read together, the numbers suggest AI is welcome behind the scenes. It is much less welcome when it stands in for the people and voice a brand presents. Picture a skincare brand that builds one idea around a nightly routine and lets AI resize it for every placement. A synthetic spokesperson delivering that idea asks the shopper to trust someone who doesn’t exist.

AI tools can’t tell whether an ad looks like your brand

Ads made with similar prompts and little brand input can start to look alike. David Droga, vice chair of Accenture and founder of Droga5, put it this way in his Advertising Hall of Fame interview:

“Infinite sameness, disguised as infinite freedom, is not a renaissance. It is a loop.”

Say a jacket brand runs an ad promising “effortless style for every season.” That line could sell any jacket. An ad that names the waxed cotton shell and the rust color regulars know from the brand’s site gives shoppers something to recognize. A prompt rarely carries everything a brand’s customers expect from it. That knowledge sits with the brand’s own team, so the team has to understand it well enough to check ads against it.

Check every AI-assisted ad against the same three questions

Before an AI-assisted ad goes live in a paid feed, on a product page or in an email, check it against three questions. Whoever signs off, in-house or at an agency, needs to know the brand well enough to answer them. Each question covers a different risk: what the ad promises, who it looks like, and who appears to be speaking.

Run each question on its own, since passing one says nothing about the others. An ad can describe the product accurately and still look like any other brand.

Does what the ad promises match the product?

Show the ad to someone who has never heard of the brand. Ask what they expect the product to do, how big it is, who it’s for and what result it gives. Then check those expectations against the real product. AI-made visuals can add details the product doesn’t have, such as an extra pocket, a roomier sofa or a faster result. A new customer has no brand history to correct that impression.

Does it look and read like the brand shoppers already perceive?

Compare the ad with how customers commonly describe the brand: its look, its tone and what it stands for. Cover the logo and ask whether regulars would still recognize it. A customer service rep, who hears customers describe the brand every day, is a good person to ask. If a competitor could run the ad, it isn’t on-brand yet.

Does it ask shoppers to trust someone who isn’t real?

AI-made faces and voices invite viewers to believe in a person who doesn’t exist. Gallup found 62% of Americans call AI-created people or voices unacceptable, even when disclosed. Testimonials and emotional stories invite the same question, so check them too. If the ad relies on any of these, ask whether a real customer, a staff member or a product shot would do the job.

Write the standard down once

Put the standard in a one-page brand reference: signature colors, product names, phrases the brand uses, examples of its tone, and what the product does and doesn’t claim. That way the check doesn’t depend on one person’s memory, and everyone signing off works from the same standard.

Start with your biggest AI-assisted ad set

Pick the live ad set with the highest spend and AI-assisted creative. Run the three questions on it and note what you changed. Then watch the signals already available: comments, customer service tickets that say the product didn’t match the ad, and returns marked “not as pictured”.

Once the first ad set is checked, move to the next-highest spend and update the brand reference with whatever those signals show.