A team asks for ten ads for a new service. By lunch, there are ten polished options on screen. One sells speed. Another sells peace of mind. A third leads with price. The team can spend the afternoon choosing a layout and still leave the meeting without deciding why a customer would choose the service.
Generative AI can shorten the time between a brief and a finished asset. Brand strategy decides who the campaign is for, what it should say and what makes the message believable. When those decisions are still open, a bigger batch of creative gives the team more versions to review before it has a clear basis for choosing one.
What four AI advertising campaigns reveal about the brief
Coca-Cola and McDonald’s put familiar holiday expectations to the test
Coca-Cola used generative AI to revisit its classic Holidays Are Coming campaign in 2024 and returned to the approach in 2025. The red trucks gave viewers an immediate point of recognition. Some critics took issue with the generated visuals, including unnatural movement and faces. For a brand working with a familiar piece of advertising, the review question goes beyond whether people recognise the truck. What did they look forward to feeling when it appeared? Coca-Cola’s 2024 campaign · Coca-Cola’s 2025 campaign · The Verge’s coverage
McDonald’s Netherlands took a different route in December 2025. Its ad turned familiar holiday mishaps into a case for escaping to McDonald’s. The company removed the video after online criticism. Viewers took issue with both the generated imagery and the ad’s cynical take on Christmas. That’s a reason to review the premise before production begins. In this case, the question was whether the joke would feel like relief from holiday stress to the people watching. The Guardian’s coverage
Heinz and Mango gave AI a specific job
Heinz and its agency Rethink already had an idea worth exploring. The Heinz bottle was recognisable enough that an image generator might draw something like it from a generic ketchup prompt. Many of the DALL·E 2 outputs resembled Heinz bottles, and the team built a campaign around that resemblance. The images made the idea easy to grasp. Their value was as a creative demonstration. Measuring what shoppers associate with ketchup would require a different kind of research. Muse by Clios campaign coverage
Mango Teen gave the technology a different job for its Sunset Dream collection. Mango says it photographed each garment and used those photos to train a model to generate campaign imagery. Its art team then selected and retouched the images. The actual garments gave the team a reference for judging whether an editorial image still represented the clothes customers could buy. Mango’s campaign release
Three questions to answer before you generate brand assets
1. Who are we speaking to, and what are they deciding? A founder in Dubai shortlisting a marketing partner may want to know who handles revisions across time zones. A first-time skincare buyer scrolling Instagram may be checking whether the product fits their routine. Both audiences need reassurance, but they’re looking for different information. Write down the decision your buyer is making and the hesitation that could hold it up. In Search engine optimization, the terms people search for can help you spot the questions your service page needs to answer.
2. What should they remember tomorrow? Choose one message you want someone to be able to repeat after seeing the ad. For a computer repair business, that might be “You’ll know the repair cost before we start.” That gives a designer something to show and a copywriter a clear point to make. A list of services can live on the landing page. In social media marketing, that same repair business could show a sample estimate in a carousel and walk through it in a short video. Each post reinforces the same message.
3. Why should they care, and what makes the claim credible? Find something the buyer can inspect. If the repair business promises upfront pricing, show what its written estimate includes. If an agency promises a simpler handover, show the client what they’ll receive and who will walk them through it. The evidence can shape the creative. An annotated estimate may do more for the message than another image of a technician at a workbench. Put those answers in the creative brief and name the person who approves the final message. Then use AI to explore ways of making the idea clearer. For the repair example, you might test a close-up of the estimate against a short video walking through it. Both give the buyer a way to see how upfront pricing works.
