How we compute the median score you're compared with

When your analysis ends, NIMSEO shows three numbers: your score, the median of stores like yours, and the gap between the two. A number you can't check is a number you shouldn't trust, so this article explains where the median comes from — and when we'd rather show nothing.

What a median is, and why not an average

Line up every store by score, from lowest to highest. The median is the score of the store in the middle: half the stores are below it, half above. Unlike an average, it doesn't move because of one spectacular store or one disaster. For a "where do I stand?" question, that's the right number.

What goes into it

Every time NIMSEO analyses a store — from the free check, from a member's dashboard, or a re-check — one anonymised row is added to our observations base. It contains:

Field	What it is	What it is not
Store	A one-way hash of the domain	Never the domain, the URL or any page content
Platform	Salla, Zid, Shopify, Converty, WooCommerce, Wix or "other", as detected	Never your account details
Sector	The sector a member chose in their profile, when there is one	Empty for a check without an account
Scores	GEO, AEO and SEO, plus the status of each of the 24 checks	No evidence, no text from your pages
Date	When the analysis ran	Nothing about who ran it

No e-mail, no name, no visitor data, no IP address goes into the benchmark. The row exists so that your score can be compared — nothing else.

The score we compare

The number on the "your score" screen is the mean of your GEO and AEO scores: can AI find and read you, and can it quote you. It's the same number Nims uses for his verdict. SEO is shown separately in the full report, because Google renders JavaScript and AI crawlers mostly don't; mixing the two would hide that difference.

The rules

1. One row per store. If a store was analysed ten times, only its latest analysis counts. A re-check replaces the older result instead of adding a vote.
2. Last 180 days. Stores not analysed in six months leave the benchmark: platforms change, and an old score would drag the median around.
3. Your platform first, your sector when possible. If you told us your sector and at least 30 stores on your platform share it, the median is that of your platform and sector. Otherwise it's the median of your platform.
4. No median under 30 stores. If fewer than 30 stores on your platform have been analysed, we show how many we have — and no median. Thirty is a low bar statistically, but under it the number would mostly describe luck.
5. The count is always printed. "Based on 212 Salla stores" or "we've analysed 14 Zid stores so far": you always see what the median rests on.

What we don't do

• We don't fill the base with stores nobody asked us to analyse. Every row comes from a real analysis someone ran.
• We don't take "industry averages" from reports or other tools. If it isn't in our base, it isn't on your screen.
• We don't show a countdown, an expiring offer or an estimated percentage. The pressure, if any, comes from the data.

Reading your gap

The gap is simply your score minus the median, in points. On the next screen, each of your three biggest problems says how many points its fix is worth: a check's weight in its score, multiplied by the share of that weight you're currently losing. Fix the first one and your score moves by that much — then the median tells you whether that was enough to pass the stores around you.

It will move

The median changes as stores are added, re-checked or expire. If yours went up without you touching anything, stores around you improved. That's the point of a benchmark: it's a moving line, and the only honest way to show it is to show the data behind it.