Since generative AI went mainstream, our industry has asked whether AI will make advertising cheaper. The early evidence suggests we have been asking the wrong question.
Look at the tier just below the giant advertisers. Wyndham Hotels told Digiday it now produces 15 times as many social media assets as before, while cutting concepting, production, and approval time by three-quarters. Opella, Sanofi’s former consumer health arm, says it produces 20 times more, mostly variants for different languages and markets. This is the upper middle of the market. The same shift has started in this region, and the pace will accelerate.
The obvious conclusion is that production costs are collapsing. Look closer, and that is only half true. The cost has not disappeared. It has moved.
Opella has an 85-person in-house team working with generative AI. But producing at that scale creates costs elsewhere: platforms, integration, governance and, increasingly, testing. Kantar told Digiday it saw a 17 percent rise in early-stage creative tests between last December and July, and linked it directly to AI production. When you can make anything, you need to know what is worth making.
So the budget line marked “production” shrinks, while the lines marked platform, people, testing, and governance grow. Marketing has not simply become cheaper. Its cost is being redistributed.
When making becomes cheap, choosing becomes the job.
A weak idea used to cost you one weak ad. Today it costs you two thousand of them, perfectly formatted for every placement, in two languages, by Thursday. The cheaper production becomes, the more expensive bad judgment becomes. The scarce resource is no longer the hands that produce. It is the judgment that decides which route deserves to be multiplied before the machine multiplies it.
Notice what these brands are increasingly keeping agencies for. Not the endless volume of variants, but the big, brand-defining work that benefits from an outside perspective. The versions and adaptations are moving closer to the client.
For our region, that should focus minds. Agencies here do far more than adapt; concepting and strategy are core to the work. But a meaningful share of it lives in that middle: adapting campaigns across Arabic and English, across GCC markets and beyond, across cultural and regulatory lines. That middle is exactly what these systems are built to absorb.
For clients: stop measuring AI by cost per asset. That metric rewards volume, and volume without judgment is noise at scale. Measure the quality of the decisions upstream: which idea, which route, which audience, what not to make. Then fund what improves them: strategy, pre-testing and governance.
For agencies: the move from billing hours to pricing deliverables is already underway. But deliverables are getting cheaper to produce too, and the value is shifting to what happens before production begins: the strategy, the idea, the route worth pursuing, and the judgment to kill the ones that are not. Those decisions are becoming more consequential, and our commercial models should reflect that. Price the value of the decision, not just the cost of the deliverable.
Production at scale is becoming a utility. Choosing well is becoming the premium. The agencies and marketing teams that understand this early will not be the ones producing the most. They will be the ones deciding best.
(Ghassan Kassabji is CEO of Impact BBDO Dubai and Chief Growth Officer of the Impact BBDO Group across MENA.)
