When marketers first start paying attention to what AI assistants say about them, the first question is usually binary: does ChatGPT mention us or not? It's the natural question. We spent two decades asking the same thing about Google's first page. Are we on it, yes or no?
But "are we mentioned" is the wrong question, and it leads to a comfortable kind of self-deception. A business gets named in an AI answer, sees its name in the text, and concludes it's doing fine. Meanwhile it's sitting fifth in a list of seven, and almost nobody reading that answer will ever act on it.
Position is the metric that matters. Presence is just the price of entry.
Why order-of-mention decides everything
Think about how someone actually uses an AI assistant. They ask, "Who's the best HVAC company in Tucson?" or "What's a good commercial cleaning service near me?" The assistant returns a few sentences and a short list. The person reads the top of the answer, picks one or two names that show up first, and moves on. They don't scroll a results page weighing ten options. The format itself discourages it.
This is different from search, and the difference is the whole point. On a Google results page, a user sees ten blue links and scans them. Being fourth or fifth still earns clicks — diminished, but real. In an AI answer, the list is shorter and the framing is more decisive. The assistant isn't handing you options to evaluate. It's giving you an answer. The first name carries the weight of a recommendation. The fifth name reads like a footnote.
We've all done this as humans. Ask someone for a restaurant recommendation and they'll say two or three names. If they rattle off seven, you remember the first two and forget the rest. AI assistants compress the same way. Order is not cosmetic. It's the recommendation.
So a share-of-voice number that only counts presence — were you named, yes or no — tells you almost nothing useful. A business named in 80% of relevant answers might feel great about that figure and still lose, if it's landing fourth or fifth in most of them. Presence without position is a vanity metric.
Weighting position changes the picture
The fix is to weight order-of-mention. Instead of scoring a mention as a flat one point, you score it by where it lands. First mention is worth more than second, second more than third, and a name buried at the bottom of a list is worth a fraction of a name that leads the answer.
You can argue about the exact curve. Maybe first position is worth three times a fifth-position mention, maybe more. The precise weights matter less than the principle: a measurement that treats "named first" and "named fifth" as the same event is measuring the wrong thing.
When you apply that weighting, something useful happens. Two businesses that look identical on a presence basis — both mentioned in most answers — separate sharply. One consistently leads. The other consistently trails. That gap was always there. The flat metric just hid it.
The ranking nobody knew they had
Here's the part that surprises people. When you measure position across enough queries and across the assistants — ChatGPT, Claude, Gemini, Perplexity, and Google AI — a ranking emerges. Not a ranking you set, not one any business signed up for, but a real, consistent ordering of who the assistants name first, second, and third for a given question.
Most businesses have no idea this ranking exists. They've never seen it because they've never looked past the binary. They know their Google Ads position because they pay for it and watch it daily. They know their organic rank because tools have reported it for fifteen years. But the order in which an AI names them against their competitors? That's been invisible. It's been shaping demand quietly while everyone argued about whether they were "in the answer" at all.
And it's not random. Ask the same question across a week and the ordering is fairly stable. Ask it of different assistants and you'll see correlation, with interesting divergences — a business that leads on Perplexity might trail on Gemini. Those patterns are information. They tell you where you actually stand in the one place a growing share of buyers now starts their decision.
Measure it before you do anything about it
I want to be careful here, because the instinct is to jump straight to "so how do I get to first?" That's a real question, but it's the second question. You can't move a number you haven't measured, and you can't measure a ranking you've reduced to a yes-or-no.
The first job is to see it clearly: across the queries that matter to your business, where do you land, on which assistants, against which competitors, and how does that change over time? That's the baseline. It's the honest look at your current trajectory before you start projecting where you'd like to be. This is the gap LLMClarity is built to fill — measuring share-of-voice weighted by position, tracked over time, so you can see the ranking that's been there all along. Plans run from $89 to $800. What you do with that ranking comes after. The measurement comes first.
The businesses that come out ahead in AI-driven demand won't be the ones who simply got mentioned. Everyone in a category gets mentioned eventually. They'll be the ones who knew where they stood, watched it move, and understood that in an AI answer, fifth place and no place are nearly the same thing.