An AI Strategy Is Only as Good as the Operating Model Behind It

The tool everyone’s racing to deploy can’t build the coordination it depends on. That part is still the company’s job.

Every AI strategy inherits the shape of the organization running it. That’s the part the demos never mention.

A model is only as good as the operating model around it: who owns what, how anyone knows it’s working, and who has the authority to keep it moving when a team falls behind. Point a powerful tool at a well-run operating model and it compounds. Point it at a siloed one and it doesn’t fix the silos. It runs them faster, and splinters the message into more pieces, more publicly.

One place I keep running into this is GEO, Generative Engine Optimization. It’s the topic on every marketing call, and on almost every one of those calls it lands on a single desk: whoever owns the website, or whoever owns content. The logic feels obvious. GEO sounds like SEO with a new first letter, and SEO was one team’s job. So GEO becomes one team’s job too.

That instinct is the mistake.

GEO isn’t SEO 2.0. Traditional SEO could largely be run from inside one team. The pages, the keywords, the metadata all lived in one place, under one owner. GEO doesn’t live in one place. When an AI answer engine decides what to say about a company, it pulls from everywhere at once: the Google Business Profile, the influencer network, Glassdoor reviews, Reddit threads, the press, the podcasts. The raw material for the answer is scattered across a dozen teams who’ve often never sat in the same meeting. Hand that to one person and they can’t win, not because they aren’t good, but because the job was never one person’s to do.

For years, that kind of fragmentation was survivable. A customer might search a brand, click a few links, and stitch the story together on their own. Now the machine does the stitching, instantly, and hands back a single answer. If sales says one thing, the careers page says another, and Reddit says a third, the AI notices. And so does the buyer.

This is where GEO stops being a marketing quirk and starts looking like the whole AI economy. McKinsey found that more than 80% of companies say generative AI has had no real impact on their bottom line, despite all the spending. The single biggest thing separating the companies that do see a profit impact isn’t a better model or a bigger budget. It’s whether they redesigned how work flows across their teams. Only about one in five have actually done it.

It’s worth sitting with that. The winners aren’t the ones with the best AI. They’re the ones who changed how their people work together around it. Everyone bought the tool. Almost nobody rebuilt the operating model. And the operating model is the whole game.

The good news is that the fix isn’t heroic, and it isn’t more technology. When the operating model actually works on the ground, it comes down to three fairly unglamorous things: clearly defined roles and responsibilities, so someone owns each input and everyone knows who owns what; reporting and analytics, so the team can see whether it’s working instead of hoping it is; and executive support, so the work survives the day-to-day and there’s a place to escalate when one team falls behind on its part.

That’s the whole operating model: who owns what, how anyone knows it’s working, and who has the authority to keep it moving. GEO is simply one place a lot of companies are discovering they never built one.

Any AI strategy will eventually expose whichever of those three is missing. Integration isn’t the thing a company does after the AI works; it’s the thing that makes the AI work at all.

The teams pulling ahead aren’t buying smarter tools than everyone else. They decided, before the tool ever showed up, that the handoff between teams was the product.

Kristen Lee Santos is a go-to-market executive focused on strategy, growth, and delivery. More on how she works: her résumé and the brands she’s building. Find her on LinkedIn.