Five things I’d already be doing with LLMs if I were a CCO

1. 1. I’d look at the media through new eyes (and I’d be in for a few surprises)

      Throughout my professional life, I’ve ranked media outlets using two criteria: prestige and reach. Appearing in a major publication was, almost by definition, a win. I don’t plan to throw that criterion in the bin; it remains valid. But I’d add a new, somewhat uncomfortable question if I’m trying to influence LLMs: does this outlet actually influence what the AI says about me?

      I say uncomfortable because the answer doesn’t always line up with the traditional hierarchy. We’d start seeing second-tier outlets holding huge weight in model responses, whilst certain flagship publications barely feature depending on the search query. Reordering the media landscape with that dual filter (the traditional one plus the new one) would be my first duty. Uncomfortable, yes. Essential, absolutely.

   2. I’d sit communications and SEO/GEO at the same table (and I wouldn’t let them leave)

      This is one of the hardest changes, because it doesn’t depend on technology, but on company habits. Influencing AI models requires corporate communications and positioning specialists (traditional SEO and its cousin, GEO) to stop working in their respective corners.

      Why am I obsessed with this? Because I’ve seen the risk of not doing it. If the positioning team optimises without speaking to the person guarding the brand’s reputation, it’s easy to end up winning the commercial battle while losing the reputational one: mentions go up, sales go up, and out of nowhere, the brand narrative gets eroded. My job as CCO wouldn’t be to choose between one team and the other, but to ensure that no decision regarding this channel is made without both of them in the room. It sounds obvious. Anyone who has worked in a mid-sized business knows it isn’t.

      1. I’d fight for a budget to make mistakes early

         Now for the ugly part: this costs money, and you have to defend it in front of a board that will ask you about ROI. Here I’ll be honest: in such a new channel, knowledge isn’t bought ready-made; it’s built by trial and error. There is no off-the-shelf methodology, today’s learnings will expire in a few months, and every model family behaves differently.

         So, I’d go to the executive committee without promising certainty, presenting a simple pitch: I need a budget allocation to monitor what models are saying about us, team time to interpret it, and room to experiment without guaranteed immediate returns. Hard to get approved? Very. But the alternative (not investing) isn’t free either: it means accepting that others will learn first and claim the territory. In an emerging space, whoever moves first gains a real advantage. And I’d much rather explain an invoice for experimentation than justify a year from now why we arrived dead last.

      2. I’d clean house (and clean up other people’s houses, too)

         This is a far from simple front, but it is equally vital. Models construct their answers out of whatever they find: your website, your press releases, articles from years ago, outdated product pages. If what they find is old, incomplete, or contradictory, the AI can feed it back to you as “truth” and “current affairs” about your brand.

         As a CCO, I’d dedicate a chunk of effort to something thoroughly unglamorous: auditing and purging the content that already talks about me. Not just publishing new stuff (we do plenty of that already), but reviewing the old: data that is no longer accurate, messages we’ve moved away from, third-party information riddled with errors. Think of it as reputational hygiene. The house describes itself to the AI, so what’s inside had better be tidy.

      3. I’d become obsessed with giving the AI reasons to cite me (with good stories)

         I’ll finish with the one that motivates me most, because the previous four are about managing and defending, whereas this one is about going on the offensive. Instead of waiting for the AI to find me, I’d focus on creating reasons for it to cite me: proprietary data, research, figures nobody else has, authoritative content published in relevant outlets. Models love solid, verifiable, well-considered sources; turning your brand into one of them is the cleanest way to gain visibility without relying on what third parties say.

         But be careful not to get bogged down in the mechanics, because here lies the part I care about most. Getting cited by AI doesn’t mean abandoning what has always made communications great: the narrative. A brand that simply drops isolated data points without a thread or coherence feeds fragmented, forgettable responses. A brand that builds a solid, consistent, and relevant story over time presents itself better to people… and also to models, which ultimately learn from those very same stories.

         So, I wouldn’t switch my strategic faith: I’d keep betting on meaningful narratives, messages that don’t contradict one another, and core themes I champion for years. The difference is that now, alongside winning over a journalist or a client, that good story needs to be clear and consistent enough for a machine to understand it, remember it, and repeat it properly.

Five moves, none of them particularly wild. All under the premise of analyzing, measuring, and trying to manage what LLMs say: reorder sources, bring teams together, fight for budget, clean house, and give the AI good stories to tell. What’s interesting is that almost all of these are things communications already knows how to do; it’s just that now they must be done with a new, synthetic, and highly influential reader in mind—one that summarizes and recommends brands to millions of people.

I’ll repeat that I’m not a CCO. But if I were, I wouldn’t wait for the dust to settle. I’d start today, even if it meant making mistakes along the way, because in this channel, latecomers don’t find a free spot waiting for them: they find it taken. And reclaiming a conversation that others have already shaped is far more expensive and laborious than having been there from the start.