Ask an AI the same buying question twelve weeks running and two things happen. A small set of brands is named almost every time. A much larger set drifts in and out. Those are different facts about a business, and every other tool in this category reports them both as “mentioned”.
Where the idea comes from
A SISTRIX study of 82,619 prompts across 17 weeks found that 86.5% of prompts keep a stable core of one to five domains that persist week after week, while roughly 89% of the remaining names rotate.
Read the second half twice. Most of the brands you see in an AI answer are not holding a position — they are passing through. So if you check once, see yourself, and conclude you are winning, there is a good chance you have watched rotation rather than standing.
Named in 46% of cycles, this brand is emerging — climbing, but not yet a position. A tool that only counted mentions would have called week 7 a breakthrough and week 3 a crisis. Neither was either.
How we classify it
Run the same question weekly and count the share of runs in which a brand appears:
- Core — named in at least 70% of runs. A position. Reaching it is a real result and losing it is a real problem.
- Emerging — 40% to 70%. Contested: on the way in or on the way out, and the band worth watching most closely.
- Carousel — under 40%. Present sometimes. Pleasant, and not yet a result you should report to anyone.
Illustrative. One check would have shown all three as “mentioned”. Twelve show which of them the answer actually keeps.
Below 4 observed cycles we give no verdict at all. A rate computed from two data points is noise wearing a percentage sign, and somebody would act on it. That floor is a constraint we chose, not a limitation we are hiding.
Why it costs nothing extra
It falls out of tracking the same prompts over time — the data is already there, and the classification is arithmetic on top of it. No additional engine calls, no additional spend.
So the reason competitors do not report it is not cost. It is that “named in 3 of 12 runs” is a less flattering number to put on a dashboard than “mentioned”. We would rather show you the less flattering one, because it is the one you can act on.
What it changes about the work
- It tells you whether a change worked. A brand moving from carousel to core over six weeks is a result. The same brand appearing once after a campaign is weather.
- It tells you which competitors actually matter. The ones in the core hold ground. The ones rotating through are not your problem.
- It tells you where the ceiling is. A core of one to five names means there are only so many seats, and knowing that is better than assuming the list is open.
The idea is explained in full, with the evidence, in the GEO guide. What you see on it in your own account is the tracking.
Questions people ask
How long before I get a verdict?
Four weekly cycles at minimum. Below that the rate is computed from too few points to mean anything, so we show no classification rather than one you might act on.
Is this a metric other tools have?
Not as far as we have found. Every competitor we have checked reports whether a brand was mentioned. The distinction between a brand named consistently and one named occasionally is the thing this product exists to show.
What counts as a cycle?
One run of one prompt against one engine. A brand's stability is measured per prompt per engine, because being durable on Perplexity and absent from ChatGPT is a real and common situation that a blended score would hide.
Can I be core on one prompt and carousel on another?
Yes, and that is usually the interesting part. It tells you which questions you have actually won and which ones you only occasionally show up on.