CEOs: Avoid AI Strategy Pitfalls with Informed Decisions

Publicis chairman Arthur Sadoun told investors this week that his firm is now selling directly to CEOs, not CMOs. “This has to start from the top,” he said on the Q3 results call, crediting Sapient, Publicis’s technology arm, with helping win “the two biggest accounts of the year without a pitch.” One of them was PepsiCo’s global media business, awarded on September 2 with no review. Omnicom had held it for roughly three decades, and Barclays put the lost fees at about $100 million a year. Publicis also walked away from Coca-Cola’s global media pitch outside North America. Net revenue rose 5% to €3.53 billion.

Most CMOs will read this as a turf threat. Most boards will read it as efficiency: fewer agencies, no six-month review, faster decisions on AI. Both reactions miss what happened. The AI-anxious CEO is not the story. The story is that the one moment that forced a company to diagnose its own problem has been quietly deleted from the buying process.

The Pitch Was Never About Choosing an Agency

Mark Ritson’s planning triptych has a strict order: Diagnosis, then Strategy, then Tactics. Never reordered. A competitive pitch, for all its waste, is the last place most companies are forced to execute the first step. To brief an agency, someone must write down what the problem is, who the customer is, and what winning looks like. Agencies then compete on the quality of their answer to that written problem.

Remove the pitch and the first document in the room is the vendor’s. The vendor’s proposal now defines the problem, scopes the solution, and prices both. That is Tactics wearing Strategy’s clothes, and it is exactly the sequence error the Practicum exists to prevent.

“AI Anxiety” Is a Symptom, Not a Diagnosis

The first of the Practicum’s seven strategic philosophies is to diagnose the problem type before prescribing. “We need an AI transformation” is not a diagnosis. It is the feeling of being outflanked, and it can sit on top of a product problem, a pricing problem, a sales-capacity problem, or a genuine market-orientation problem. A CEO who feels the anxiety and buys the cure from the party selling it has skipped the question of which disease she has.

Look at where the money sits. Media is 62% of Publicis’s revenue. A transformation engagement is a persuasive way into a media wallet, and I would not assume the sequence runs the other way. That is not a criticism of Publicis. Sadoun is playing a smart hand, and a CEO-level sale is a rational response to a CEO-level worry. It is a criticism of any client that treats the door-opener as the diagnosis.

What Both Sides of the Table Taught Me

I co-founded Bolt Goodly and helped build it into a nine-figure agency, so I sat on Sadoun’s side of that table for years. The clients who got the best work from us were never the ones who moved fastest. They were the ones who handed us the sharpest written problem, because a sharp problem is the only thing that lets an agency be held accountable. When the brief was vague, so was the accountability, and we both knew it.

At Pinnacle the order of operations was the whole game. We cut costs by 60% and built AI workflows that took 55% out of the workload, but the decision that mattered came before any tool was chosen. It was deciding which work should stop existing at all. Automating a process that should have been abolished would have given us a faster version of the wrong organization. Diagnosis made the savings possible, and the technology only executed it.

Who Owns the Diagnosis Owns the Company

The fair argument for Sadoun is real. Pitches burn months and millions, three holding companies now dominate the field, and a CEO who wants fewer, deeper vendors is not irrational. Publicis’s record of new business wins suggests clients are getting something for their trust.

But consolidation changes who holds the pen. When a company is choosing among three agencies instead of thirty, and choosing them at CEO level without a competitive brief, the only internal check left on the vendor’s framing is the CMO. If the CMO is not the company’s diagnostician, the agency is. And whoever writes the problem statement writes the strategy, because every recommendation is downstream of what the problem was said to be.

This is where the conventional CMO response goes wrong. Defending the gatekeeper role, the right to run the review and sign the contract, is the wrong fight, and it deserves to lose. The fight worth winning is ownership of the diagnosis: a written statement of the business problem, the target segment, and the trade-offs the company will accept, authored inside the company before any agency, platform, or AI vendor gets to answer it. A CEO can buy from anyone, with or without a pitch. What a CEO cannot outsource is the sentence that says what is wrong.

The position I would defend in any boardroom: a no-pitch award is not a procurement shortcut, it is a strategy decision, and it is only as good as the diagnosis the buyer brought to it. So when your company’s most consequential AI commitment arrives without a pitch, ask the question the process no longer asks for you: who wrote the problem statement, and was it anyone who gets paid by the answer?