How the Ruescan score is calculated
By Rues · Published · Updated · Scoring version 2.5
The Ruescan score is a number from 0 to 100 and it comes only from signals that could really be measured. This page explains the rules in plain words. What each signal measures is written in the signal glossary.
Version number
The scoring rules carry a version number. Today's version is 2.5: 275 signals in 10 categories. When the rules change the number goes up, and every report states the version it was calculated with. Two scores calculated with different versions cannot be compared directly; to compare, use two measurements made with the same version.
A signal gets one of five states
| State | What it means | Effect on the score |
|---|---|---|
| Solid | Measured, and what we look for is in place. | Counts as 1. |
| Partial | Measured, and what we look for is only half met. | Counts as 0.5. |
| Open | Measured, and what we look for is missing or wrong. | Counts as 0. |
| Not measured | The measurement could not be made: the page was cut off, a source did not answer or a quota error happened. | Left out of the score, and not a penalty. |
| Not applicable | The signal does not fit this site. Example: app links on a site that has no app. | Left out of the score. |
Why a signal that cannot be measured is left out
Counting something we could not measure as open would punish a site unfairly. Counting it as solid would give false comfort. So a signal that cannot be measured is taken out of the calculation and the report says "not measured" in plain words. When the share of measured signals is low, the score is marked "provisional"; the threshold is written below.
The score is built in three steps
Category score
Each signal counts as solid (1), partial (0.5) or a gap (0). The category score is the weighted average of the signals measured in that category, from 0 to 100. The weight of a signal sets its importance: Urgent 5, Important 3, Moderate 2 and Polish 1.
Site score
The site score is the average of the measured category scores, taken by their own weights. The site is recognised first (for example a shop or a local business), and signals that do not fit it, such as app links, are left out.
Upper limit
A few serious errors put an upper limit on the score; this is called a gate. For example, if the home page does not open at all, the score cannot come out high. The caps of the gates are in the table below.
Category weights
Each of the 10 categories takes as much room in the site score as its weight. The larger the number, the more the category counts.
| Category | Weight |
|---|---|
| AI access and readability | 14 |
| Brand brain: identity and entity | 12 |
| Content and answer blocks | 12 |
| Technical SEO | 14 |
| Speed and code | 14 |
| Structured data | 10 |
| Security and headers | 8 |
| Sharing | 5 |
| Language and hreflang | 4 |
| App links | 7 |
Gates and their caps
If one of the errors below is measured, the site score cannot come out above the cap. Some gates also cap the related category.
| Gate | Measured error | Site score at most | Category score at most |
|---|---|---|---|
K1 | The home page does not answer with success | 20 | - |
K2 | The home page is closed to indexing (noindex) | 30 | - |
K3 | All AI search bots are blocked | 49 | AI access and readability: 20 |
K4 | Fewer than 50 words in the server HTML (content depends on JavaScript) | 59 | Content and answer blocks: 30 |
G1 | robots.txt returns a server error | 70 | - |
K5 | No HTTPS | 49 | - |
The score also comes with a band name
The report labels the score with one of five bands: 0-34 Critical, 35-59 Weak, 60-79 Needs work, 80-91 Good, 92-100 Exemplary.
Provisional score
If the measured weight is below 80% of the weight that applies, the score is marked "provisional" because it rests on few measurements. The mark shows when many signals could not be measured, when robots.txt returned a server error, or when the home page opened only on a retry.
Signals of the information type
Some signals only inform. They do not count toward the score and do not produce a finding; they appear in the report as notes. The llms.txt file is an example. The glossary states the type of each signal next to its entry.
Evidence level
Every finding carries an evidence level. "Observed" means the measurement was seen directly in HTTP, DNS or HTML. "Inferred" means the result rests on a heuristic rule or on context. When the source behind a fix is an official document, the glossary links to it.
What we do not claim
- The score makes no prediction about rankings, traffic or being mentioned in AI answers.
- Ruescan reads only the public face of a site; it does not look at pages behind a login.
- The names and details of the 26 signals unique to Ruescan are not written; only their number is. What every other signal measures is open in the signal glossary.
Who prepares it and how it is updated
The scoring is prepared by Rues. When the rules change the version number goes up and this page is updated. If you spot a mistake, write to info@ruesandora.com.