Every June the advertising industry decamps to the French Riviera for Cannes Lions, and the mood along the Croisette tends to reveal where marketing is genuinely heading. In 2025 the festival ran on pure AI optimism, with agencies and brands promising that generative tools would reinvent creativity almost overnight. In 2026 the tone shifted. Read across the week, as analysts at McKinsey did, the conversation moved from what the technology could do to what it actually delivers, and that change says a great deal about the year ahead for marketers.
The clearest sign was a quiet fatigue with AI as a topic. After two years of dazzling demos, the people who actually run marketing organisations sounded less interested in what a model could produce and more interested in whether any of it moved the business. The phrase that kept surfacing was mediocrity at scale, a blunt description of what happens when everyone reaches for the same handful of models trained on the same patterns. The result is a flood of competent, forgettable work that all looks and sounds alike.
From what is possible to what pays off
The headline shift is a move from possibility to value. Marketers have largely won the argument that AI belongs in the workflow. Roughly 60 percent now use it several times a week, so the novelty phase is clearly over. The harder truth is that only about 10 percent have actually redesigned how their teams work in order to capture real business impact. Most organisations have simply bolted AI onto old processes and are quietly disappointed that the returns feel thin. The festival's real signal was that the gap between adoption and value is now the central marketing problem.
Creativity is the antidote to sameness
If AI floods the market with average output, then genuine creativity becomes the scarce and valuable thing again. This was the first of several shifts on display. The brands drawing a crowd were not the ones generating the most content, but the ones using the tools to sharpen a distinctive idea rather than water it down. The lesson is uncomfortable and freeing at once. When production becomes cheap and close to infinite, the premium moves back to taste, judgement, and a point of view that a machine cannot supply on its own.
The operating model finally matters
The second shift is about plumbing rather than sparkle. The teams pulling ahead are rebuilding their operating model around AI instead of treating it as a side tool. That means rethinking how briefs are written, how work passes between people and machines, how data feeds decisions, and how success is measured. The small group that has done this work is separating from the pack, and everyone else is learning that you cannot buy your way to impact with software alone. The value comes from the redesign, not the purchase.




