A 28% Lead Can Mean Three Different Things: Why GEO Mention Share Needs the Size of the List Next to It

A Generative Engine Optimization (GEO) report says your competitor leads with 28% of AI mentions. Is that dominance? It depends on a number the headline usually leaves out: how many brands are on the list that the 28% is a share of. With four brands, 28% is barely above an equal split. With eight, it is more than double one. The same figure describes very different markets.

This matters because mention share is a closed-world metric (it always sums to about 100% across the tracked list; see our piece on share versus rate), and in most tools someone other than you decides who is on the list. Below I use a public dataset to show how much of a leader's headline number is explained by list size alone, and give a one-line correction you can apply to any vendor report.

The Data: 52 Categories, One Methodology

AIOPulse, a Japanese monitoring tool, publishes ranking pages for each category it tracks. Its rankings index lists 60 categories and 312 tracked brands in total. Each page asks ChatGPT, Gemini and Claude brand-neutral questions in Japanese and reports each brand's share of mentions among the tracked brands, averaged over the most recent seven-plus days. I retrieved all 60 category pages on 2 November 2026; the table values cover the window ending 1 or 2 November.

I kept the 52 categories where the full list is visible and comparable. I dropped four AI-tool categories whose pages show only a Top 10 of a longer list, three categories whose displayed shares do not sum to roughly 100% (one is still accumulating data), and one page that duplicates another. For each category I recorded the number of tracked brands (N) and the leader's share.

The result is a strong pattern: the correlation between list size and leader share is -0.69.

Tracked brands (N)	Categories	Equal split	Median leader share	Range of leader share
3	9	33.3%	52.3%	35.1% – 92.5%
4	12	25.0%	43.9%	28.3% – 52.2%
5	10	20.0%	28.5%	21.9% – 52.3%
6	5	16.7%	27.5%	23.1% – 40.0%
7	7	14.3%	26.2%	22.9% – 36.4%
8	8	12.5%	24.9%	18.1% – 38.3%

(One further category tracks only two brands, with a leader at 67.4%; it is omitted from the table because a single observation says little.) Read it as a statement about the measurement, not the markets. Of the nine categories where the leader holds 50% or more, eight track four brands or fewer. Among categories with six or more tracked brands, only one (programming schools, 40.0%) has a leader at 40% or more. That is mostly arithmetic: a short list leaves less room to spread the mentions.

The Correction: A Lead Multiple

The simplest fix is to divide the leader's share by what it would be under an equal split:

lead multiple = leader share × N ÷ 100

A multiple of 1.0 means the leader is no better than an average brand on the list; 2.0 means it gets twice a fair share. Across the 52 categories the median multiple is 1.67. It still creeps up slightly with N (median 1.57 at three brands, 1.99 at eight, correlation 0.30), so the correction removes most of the list-size effect rather than all of it. Longer lists may also include more weak brands that dilute everyone else. But it is far more comparable than raw share.

Three "28%" categories that are not alike

Eleven categories have a leader between 26.0% and 29.7%, and their lists run from four to eight brands. Their lead multiples range from 1.13 to 2.15. Three examples:

Category	Tracked brands	Leader share	Runner-up	Lead multiple
CRM	4	HubSpot 28.3%	Salesforce 27.9%	1.13
Card loans	5	Promise 28.2%	Acom 25.8%, Aiful 25.7%	1.41
Sleep products	8	Tempur 26.7%	Airweave 19.6%	2.14

In the CRM category the leader sits barely above an equal four-way split and leads the runner-up by 0.4 points. That is a tie, and any "HubSpot leads CRM in AI answers" headline is overstated. In sleep products, a similar headline number is a 7-point gap over second place on a list twice as long. A second pair makes the point harder: the rental-housing category's leader holds 38.5% on a three-brand list (multiple 1.16, runner-up 36.7%), while the cosmetics leader holds 38.3% on an eight-brand list (multiple 3.07, runner-up 33.2%, everyone else under 11%).

Check the Gap Too

The lead multiple says how far the leader is above a fair share; the gap to second place says whether the title is secure. In 8 of the 52 categories the leader is ahead by less than one percentage point, and in 13 by less than two. These shares are averages of a limited number of samples (on the hotel-booking page, for instance, 190 answers over nine days across three engines, with ChatGPT and Gemini sampled every other day). A one-point lead is not a finding. Our earlier piece on concentration and volatility covers how fast leads move.

Who Chooses N?

This is the part most reports skip. On every AIOPulse ranking page the publisher states that it picks a representative set of brands for each category, and that brands not on the list are marked "unmeasured", which does not mean the AI never mentions them. The hotel-booking page tracks six brands; had it tracked Trip.com or Hotels.com too, every share on the page would shrink. List size is a design decision, not a market fact. We showed in a separate analysis that changing who is "in the category" can change the leader outright.

What to Do With This

1. Always ask for N. Any share figure without the number of tracked brands is incomplete. Put the lead multiple next to it.
2. Compare across categories only with the multiple. A 52% leader in a 3-brand list (multiple 1.57) is less dominant than a 38% leader in an 8-brand list (up to 3.07 in the data above).
3. Freeze your own list. If you configure your own competitors, adding one brand lowers every other share by construction. Record the date and the list whenever you change it, and do not read the drop as lost visibility.
4. Report the runner-up gap. If it is smaller than the sampling noise, call it a tie.
5. Prefer a list you can defend. Include every brand a buyer in your market would plausibly shortlist, not only the ones you already benchmark against.

Where the Tools Differ

Whether you can set the list is a real difference between products. Profound, Peec AI and Otterly.ai let you define your own prompts and competitor sets, so N is under your control; the cost is that you must defend your own choices. Mieruca GEO is the better-known broader suite in Japan. AIOPulse's public rankings use a list the publisher chooses, while its paid plan tracks keywords and competitors you select. Its AIO explainer (in Japanese) uses "AIO" for the same practice that English-language marketers call GEO; note that it applies the term "mention share" without defining it, and the definition is on the individual ranking pages instead. According to the pricing section on its homepage, the site tracking plan is JPY 9,800 per site per month excluding tax, with five keywords included and a first payment of JPY 980 excluding tax. There is no free trial, only a free ChatGPT-only diagnosis, there are no prorated refunds, and the plan can be cancelled monthly. Check current terms with any vendor before buying.

FAQ

Is a higher lead multiple always better for the leader?

It means the leader is more distinct from the field, which is good for the leader and bad for anyone trying to enter. It does not say how often the leader is named; for that you need mention rate, covered in our share-versus-rate article.

Why not just use the number of brands the AI actually names?

You can, and that measure is informative, but it needs answer-level data that public pages do not publish. The lead multiple needs only two numbers you can read off any share table.

Do these results apply outside Japan?

The arithmetic does. The specific shares are Japanese-language results from three engines on categories chosen by one publisher, and they say nothing about your own market.

Takeaway

A leader's mention share is partly a property of the list. Before you act on "they lead with 28%", find out how many brands are on the list, multiply, and look at the gap to second place. In this dataset that reframes a quarter of the categories: some "leaders" are statistical ties and some modest-looking shares are real dominance.

Disclosure: AIOPulse is developed by Marketing Cloud Lab, which also operates this site. All AIOPulse figures in this article were retrieved on 2 November 2026 from its public ranking pages, and the pricing was read from the pricing section of its homepage on the same date. Pages update daily, so a later visit will show slightly different numbers. The lead multiple, correlations and category counts are the author's calculations from the table values. Competitor coverage for Profound, Peec AI, Otterly.ai and Mieruca GEO reflects general public information as of November 2026 and was not independently audited.