You Can't Outadvertise the Algorithm: Field Notes From Cannes 2026

I spent three days on the Croisette this year. I came home convinced of one thing: this is a Marketing Renaissance, not a marketing apocalypse — but only for the brands willing to earn the machine's recommendation rather than buy their way around it.

Łukasz Stebelski

President, P&G Alumni Geneva · Partner, WebQuest Digital

Last year at Cannes, every other conversation began with the same word: "what's next?". This year the word was gone. In 2026 nobody asked if AI would reshape how brands grow, or maybe when. The only question left on every stage and every beach panel was how — and that single shift in tense tells you everything about where we are.

I came away with pages of notes and one organizing idea. The brands that will win the next decade are not the ones with the biggest models or the loudest launches. They are the ones who treat this moment for what it is — a renaissance of the marketing craft, where understanding the consumer deeply and executing with excellence across the entire shopper journey suddenly matters more than it has in twenty years. Here is what I heard, and what I think we should do about it.

Signal 01 — The mindset

AI literacy is not a skill. It's a mindset.

The line that stuck with me on day one came out of a panel on building an AI-capable organization: "AI literacy is not a skill, it's a mindset." I have watched too many companies treat AI as a training module — a course you complete, a certificate you file, a tool you bolt onto the side of the org chart. That framing quietly guarantees failure.

A skill is something a few specialists own. A mindset is something an entire culture carries. The organizations pulling ahead are not the ones that hired the most prompt engineers; they are the ones where the brand manager, the supply planner and the e-commerce lead all instinctively ask, "what does the machine see when it looks at us?" That instinct cannot be delegated to a center of excellence. It has to live in the room where decisions get made.

Signal 02 — The hard truth

You can't outadvertise your way out of the LLM

If you remember one sentence from this piece, make it this one, from Lindsay Barrett: "Humans are the best gateway to the LLM. You can't outadvertise your way out of the LLM."

General AIs and retail AIs read your facts. They base their recommendations on numbers, social proof and trust signals. They will not recommend your product if you don't have a genuine product edge and a winning consumer value equation.

— the consensus I heard repeated, in different words, on three different stages

This is the part that should make every CMO sit up. For thirty years our reflex when growth stalled was to spend our way back — more reach, more frequency, more share of voice. That lever still works on a human in a feed. It does not work on a language model standing between your product and the shopper. The model has not seen your TV campaign. It has read your reviews, your ratings, your specifications, your Q&A and your claims, and it is quietly deciding whether to put you in the answer.

You cannot bid your way into that answer. There is no media plan that compensates for a thin product page, a weak value equation, or proof you never bothered to make legible. The machine rewards substance, and it is ruthless about the absence of it. At WebQuest we call the new scoreboard share of answer — and it is earned, not bought.

Signal 03 — The battleground

The bottom of the funnel is where the future is being decided

One of the sharpest provocations of the week, from a session on the shopper journey: "Everyone is focused on the top of the funnel only." Brands obsess over awareness and the big idea, then hand the product page — titles, images, video, social proof on the retailer — to whoever has a spare afternoon. That used to be survivable. It no longer is.

The bottom of the funnel is where purchase happens, and it is now also where the AI does its reading. PDP AI-readiness — whether your detail page actually answers the question a shopper asks an assistant — has quietly become one of the highest-leverage investments in marketing. It is unglamorous. It does not win Lions. And it is exactly where the next point of market share is hiding.

Signal 04 — The new craft

Content enrichment — and the coming "dead sea of brand AI agents"

If there is a single new discipline emerging from all of this, it is content enrichment. We have to create richer, more multifaceted content than ever before, because LLMs capture the full complexity of human need. The example that landed for me: a shopper doesn't search "smartphone." They tell an assistant, "I need a phone for commuting." Does any of your content reflect that human — the battery anxiety, the podcast habit, the cracked-screen-on-the-train fear? Or does it just list a chipset?

But enrichment comes with a warning I keep repeating to clients. Ten years ago, every brand rushed to build a mobile app, and we created a dead sea of abandoned brand apps nobody opened. In 2026 the temptation is to rush out a branded AI agent for the same reasons — because the competitor did. We need to watch that we don't spend the next two years building a dead sea of brand AI agents. Enrich the content that already sits in the shopper's path before you build a new destination they'll never visit.

Signal 05 — The human edge

If everything becomes linear, nobody gets excited

For all the talk of automation, the most quoted moment of my week was about emotion. Toto Wolff, on stage, on why the machine can't have the last word: "Human emotions cannot be replaced. If everything is linear, then nobody gets excited. Formula 1 needs to be good — but also unpredictable."

I think about that line as a discipline, not a comfort. AI is extraordinary at making everything more efficient, more optimized, more linear. But linear is the enemy of memorable. A brand that optimizes away every surprise, every imperfection, every spark of unpredictability optimizes away the very thing that makes humans feel something. Use the machine to remove the friction — and protect, fiercely, the magic it cannot generate.

This connects to a quieter but vital point I heard elsewhere: AI needs to drive business growth, not just effectiveness. Efficiency is table stakes. If your AI program only makes the existing machine cheaper to run, you've automated your way to the same place. The mandate is growth — new demand, new value, new emotion — not a tidier cost line.

Signal 06 — The honest metric

Proxy metrics can be gamed. The consumer goal cannot.

A line from a media-measurement panel that every marketer should tape to their monitor: "We use a lot of proxy metrics to audit results — and proxy metrics can be gamed, by you, by your teams, by your agencies. You have to be always clear on what your ultimate consumer goal is, and make sure you reflect it in everything you do."

In an AI-mediated world this matters more, not less. When the algorithm optimizes toward a proxy, it will hit that proxy brilliantly and miss the point entirely. The only defense is a relentlessly clear definition of the real consumer outcome — and the discipline to keep pointing every model, every dashboard and every agency at that, not at the number that happens to be easy to move.

Signal 07 — The clarity test

Find the one use case that ends the debate

For anyone still stuck in the abstract "is AI good or bad for us" argument, the most useful mental model of my week came via Malcolm Gladwell: when you wonder about the advantages and disadvantages of a technology, identify one use case where the advantage is so exceptional that you stop wondering.

His example moved the room — access to health services, in a world where half the U.S. population can't cover an unexpected medical bill. It reminded me of watching a hearing-impaired person communicate over video conferencing for the first time: a technology that, almost as a side effect, delivered the first real channel for remote human connection they'd ever had. After you see that, you never doubt the human upside again. For your brand, the exercise is the same: stop debating AI in general, and find the one place where it does something genuinely exceptional for your consumer. Start there.

Signal 08 — Beyond the consumer shelf

Even B2B now has a "buyability" score

One last idea worth carrying home, from the team at Bain — the originators of the NPS — who introduced what they call the Likelihood-to-Buy (LTB) score: a structured read on your probability of winning a given B2B deal. It blends "are you on the buyer's Day-One shortlist?" with "how much do they actually trust you?"

What struck me is how asymmetric the stakes have become. Selling a B2B service, you might expect one in five deals to land. But on the buyer's side, a single bad purchase can get someone fired. That asymmetry means trust and a compelling, legible story aren't soft assets — they're the whole game. The same logic that governs whether an AI recommends your shampoo governs whether a procurement committee shortlists your platform: substance, proof and trust, made unmistakably clear.

The verdict: a Marketing Renaissance

Compared to 2025, when everyone was anxiously asking "what's next," the mood in 2026 had changed entirely. The defining theme of the week was not disruption or fear. It was renaissance — a return to the fundamentals of the craft, supercharged by new tools.

Because here is the quiet irony of the AI shelf. It does not reward the brands that shout loudest or spend most. It rewards the brands that understand their consumer more deeply than ever, that build genuine product edge, that prove their claims, and that execute with excellence across the full shopper journey — from the first spark of awareness to the moment an assistant decides to put them in the answer. That is not a new game. That is the oldest game in marketing, finally being judged on its merits again.

You can't outadvertise the algorithm. But you can earn it. That's the work — and I, for one, have never been more optimistic about it.

These notes are drawn from my three daily recaps of Cannes 2026, expanded and connected. With thanks to the speakers and friends who made the week what it was — you know who you are.