How Reception scores a site
How confidently machines can identify a business from its site, 0-100: the weighted mean of five layer scores (entity 30, structured data 30, content 20, Open Graph 10, technical 10).
The five layers, and why they weigh what they do
| Layer | Weight | What it measures |
|---|---|---|
| Entity identity | 30 | Whether name, address, phone, email and web address are one identical string everywhere they appear. |
| Structured data | 30 | Machine-readable JSON-LD that states who the business is, what it offers and what its pages answer. |
| Open Graph | 10 | The tags that decide what a shared link looks like. |
| Content | 20 | Whether pages answer first, link to each other, are current and carry no template filler. |
| Technical | 10 | Headings, https, robots, sitemap and metadata present before any script runs. |
What one problem costs
The points one finding subtracts from its layer: a severity floor plus (base minus floor) times the share of scanned pages affected.
That second half is why the same fault costs different amounts on different sites. One page missing a heading and every page missing a heading are the same fault with very different consequences, and the score follows the consequence rather than counting faults.
| Severity | Cost | What it means |
|---|---|---|
| Critical | 15-20 points | Machines cannot tell who the business is or what it sells. |
| High | 7-10 points | Damages trust or structure across several pages. |
| Medium | 4-6 points | Costs visibility on one surface. |
| Low | 1-4 points | Housekeeping. |
The four grades
| Grade | Range | What it reads as |
|---|---|---|
| Weak reception | 0-39 | Found, but not understood. |
| Fair | 40-64 | Understood on some pages, not others. |
| Strong | 65-84 | Consistent identity, most pages fully readable. |
| Clear | 85-100 | Every page says the same thing to every machine. |
The checks themselves
Reception runs 66 individual checks across the five layers. They are the scanner’s internals rather than vocabulary worth publishing one by one, and every finding in a report names the check it came from alongside the evidence that triggered it — the page it was found on, the element, and the value that was there.
Nothing in a report is modelled or estimated. If a site cannot be reached, or blocks the scanner, the report says so and shows no score at all rather than a zero — an unreachable site has no score, and zero is a score.
Questions
- How is the Reception score calculated?
- It is a weighted mean of five layer scores, each out of 100. Entity identity and structured data carry 30 points each, content 20, Open Graph 10, and technical 10. The weighting reflects what actually prevents a machine from identifying a business: inconsistent contact details or missing structured data break identification outright, while a missing social preview image costs reach rather than recognition.
- What do the four grades mean?
- Weak reception is 0 to 39 — the site is found, but not understood. Fair is 40 to 64 — understood on some pages and not others. Strong is 65 to 84 — consistent identity, with most pages fully readable. Clear is 85 to 100 — every page says the same thing to every machine. The grade is a reading of the score, not a separate judgement.
- How much does a single problem cost my score?
- Each finding has a severity band and subtracts points from its own layer, not from the total directly. Critical findings cost 15 to 20 points and mean machines cannot tell who the business is or what it sells. High costs 7 to 10 and damages trust or structure across several pages. Medium costs 4 to 6 and loses visibility on one surface. Low costs 1 to 4 and is housekeeping.
- Why does the same problem cost different amounts on different sites?
- Because the cost scales with how widespread the problem is. A finding subtracts a fixed floor for its severity band, plus a share of the remaining points based on how many of the scanned pages it affects. One page missing a heading and every page missing a heading are the same fault with very different consequences, and the score reflects that rather than counting faults.
- What are the five layers Reception scores?
- Entity identity is whether name, address, phone, email and web address are one identical string everywhere they appear. Structured data is machine-readable JSON-LD stating who the business is, what it offers and what its pages answer. Content is whether pages answer first, link to each other, stay current and carry no template filler. Open Graph is the tags deciding what a shared link looks like. Technical is headings, https, robots, sitemap and metadata present before any script runs.
- Does Reception guess at anything it cannot see?
- No. Nothing in a report is modelled or estimated. Every finding carries evidence taken from a real page — the page it was found on, the element it was found in, and the value that was there. If a site cannot be reached or blocks the crawler, the report says so and shows no score at all, rather than reporting a zero that would read as a judgement.