Getting mentioned in ChatGPT’s answer feels like a win, but the model can cite you a hundred times and still send the customer to a competitor.
Why Your Brand Gets Mentioned but Not Recommended
Ask ChatGPT to recommend a project management tool and there’s a decent chance your brand shows up somewhere in the answer. A footnote, a “you might also consider” line, a comparison table entry. And then the actual recommendation, the “if I had to pick one” answer, goes to someone else.
This confuses marketing teams because it breaks the old SEO logic. In search, appearing on the page was the win. In AI answers, appearing in the response isn’t the same as being the one the model believes in.
Here’s why: the model isn’t retrieving a list and ranking it the way a search engine ranks pages. It’s generating an answer based on a learned story about your category: who solves what problem, who’s trusted for which use case, who’s positioned as the default. Your brand can be a fact in that story without being the hero of it. You’re mentioned because you exist and you’re indexed somewhere relevant. You’re not recommended because the model hasn’t learned to frame you as the answer.
The Narrative Stack: How AI Reads Your Category
Think of it as layers stacked on top of each other. At the bottom sits raw content: your site, your docs, your product pages. Above that sits third-party commentary: reviews, comparison articles, Reddit threads, analyst posts. Above that sits consensus: the aggregate pattern of how your category gets described across all those sources. And at the top sits the model’s frame: the compressed story it tells when someone asks a question.
Mentions live at the bottom of that stack. Recommendation lives at the top. A brand can dominate the bottom layer (tons of content, tons of citations) and still lose at the top if the consensus layer has settled on a different story. That’s why a brand with less content but a cleaner, more consistent narrative across third-party sources often gets recommended over a brand that’s simply louder.
Sources Shape Frame; Frame Shapes Recommendation
Try this: ask ChatGPT or Perplexity “what’s the best CRM for a 20-person sales team” and then ask it to explain why it picked what it picked. Most people never ask the second question. That’s the mistake.
The explanation usually reveals a small set of sources doing the heavy lifting: a G2 comparison, a widely-cited Reddit thread, a couple of review sites, maybe an analyst report. Those sources built the frame. The frame decided the recommendation. Your brand’s own website almost never shows up in that explanation, because the model isn’t learning your category from you. It’s learning it from what other people said about you.
This is the part most brands miss. You can’t argue your way into the recommendation by publishing more about yourself. You have to shift what the consensus sources say, because that’s what the model is actually reading.
The Mention Trap: Visibility Without Authority
Visibility tracking tools will tell you your mention count went up 30% this quarter. That number feels good in a slide. It also tells you almost nothing about whether you’re winning the recommendation.
Picture two brands in the same category. Brand A gets mentioned in 80% of AI answers about the category, always as one of three options listed. Brand B gets mentioned in 50% of answers, but when the model explains its top pick, Brand B’s framing (the specific problem it solves, the specific customer it’s known for) is what gets repeated back. Brand A has visibility. Brand B has narrative share. If you’re only tracking mentions, Brand A looks like the winner. It isn’t.
This is the vanity metric problem. Presence is countable and easy to report. Whose story the model actually believes is harder to see, but it’s the thing that decides the outcome.
How Narrative Consensus Forms Upstream of Citations
Consensus doesn’t form in one place. It forms across dozens of scattered sources that happen to agree with each other over time. A few review sites describe you the same way. A few forum threads repeat the same comparison. An analyst report frames the category the same way three other posts already have. None of these sources coordinate. But once enough of them converge on the same description, that description becomes the default story the model reproduces.
This is why a competitor can “own” a category framing without ever paying for it directly. They just happened to get described consistently, early, across the sources that carry weight. By the time you notice, the frame is already set and your job isn’t to get mentioned more. It’s to shift what the sources are converging on.
Mapping Your Narrative Position vs. Competitors
Before you can change the story, you need to see it clearly: what frame does the model currently use for your category, which sources are feeding that frame, where does your brand’s actual positioning diverge from what AI says about you, and where is a competitor’s frame simply more consistent or more repeated across the sources that matter.
That gap between how you want to be seen and how the model currently describes you is the actual starting point. Most teams skip this step and jump straight to “let’s get more citations,” which is like trying to win an argument by talking louder instead of changing what’s true.
What to Ship to Own the Frame Before the Model Catches Up
Owning the frame means treating narrative as something you actively shape, not something you hope accumulates. That means identifying the specific sources currently driving your category’s consensus, and deciding which ones need a different story told through them. It means being precise about the one or two things you want to be known for, instead of trying to be mentioned everywhere for everything. And it means checking back regularly, because AI models retrain and consensus drifts. The frame that wins this quarter isn’t guaranteed to hold next quarter.
Mentions are the easy thing to measure and the hardest thing to act on. Narrative share is harder to measure and it’s the thing that actually moves the recommendation.
If you want to see whose frame the model is actually using for your category and where the gap is, that’s what Mavel maps. Get in touch and we’ll show you what the model believes about your brand right now.
