Kuroma
Unique Value Proposition
What this check reads on your page
Only numerals and symbols are read by this factor, never vocabulary, so it grades a page the same in any language. It splits the page into text blocks of at least four words and looks in each for a checkable particular: a percentage, a money amount, a multiple such as 3x, a measured magnitude with a unit such as ms, GB or km, or a thousands-separated count. It scores the share of blocks that carry at least one, the number of blocks that set two different figures against each other, how many distinct kinds appear, and how many named specifics the page commits to, such as a version, model or standard written as letters and digits.
Our verdict on the evidence
Deliberately down-weighted Unique-wording signals test near null; the real phenomenon is competitive redundancy, which a single-page crawl cannot measure deterministically. Weight halved accordingly.
GEO: Generative Engine Optimization (arXiv 2311.09735)
The score cannot see your competitors, so it cannot tell you whether the figures on the page are ones they also publish; what it can tell you is whether the page commits to any at all. A page that says fast, affordable and trusted is indistinguishable from every other page saying so, and an engine choosing between them has nothing to choose on. One that says 61 ms, 99.9% and 12,000 users has given the engine three things to match a question against and three things a rival would have to equal.
What it is worth
17% of AI Readiness, which is 5.1 points of the overall 100.
How audited pages score on it
Based on 30,738 audited pages.
The median page scores 0, and the middle half of pages score between 0 and 15.
- Good (75 to 100)512 (2%)
- Moderate (50 to 74)1,290 (4%)
- Needs work (25 to 49)3,223 (10%)
- Critical (0 to 24)25,713 (84%)
94% of audited pages score below 50 on it.
Where it fails most and least
Industries that fail it most
- Passenger Transport / Transit: 99% of 138 audited pages fail it
- Gaming: 98% of 817 audited pages fail it
- Manufacturing / Industrial: 97% of 2,970 audited pages fail it
- Streaming / Video: 97% of 756 audited pages fail it
- Online Learning / Course Platform: 97% of 228 audited pages fail it
Industries that fail it least
- E-commerce Platform: 81% of 154 audited pages fail it
- AI / Machine Learning: 86% of 477 audited pages fail it
- Retail Tech / POS: 87% of 100 audited pages fail it
- Digital Health / Health Tech: 88% of 128 audited pages fail it
- Mental Health / Wellness: 89% of 117 audited pages fail it
How to fix it
- Put a measured number in the sentences that describe what you do: a price, a percentage, a latency, a volume.
- Set one figure against another in the same sentence: before and after, yours against the alternative.
- Vary the evidence: a price, a percentage, a magnitude with a unit and a count each answer a different question.
- Name the version, model or standard you actually work with, written as letters and digits with no space between them (GPT-4, RFC9110), instead of describing it in general terms.
What has changed in this factor
- Rubric version 2026-08-10
- Rubric version 2026-07-07: weight in its category moved from 35% to 17%