What the thermometer reads
Every signal below is observable from one public fetch of one URL, with no login and no crawl of the rest of the site. The table is generated from the engine at page load, so it cannot describe a check that is no longer running.
That last part matters more than it sounds. The usual failure mode for a page like this is that it gets written once, the tool changes twice, and eighteen months later the site is publicly describing a methodology it abandoned. Rendering it from the registry means adding a signal to the code adds a row here in the same deploy, and removing one removes it.
Search Visibility
10 signals · 30% of the temperatureCan a search engine reach this page, index it, and work out what it is a result for?
The signals a classic crawler needs before the page is eligible to rank for anything at all. Nothing here is about competition; it is about being in the running.
| Signal | Weight within this reading |
|---|---|
| Open to indexing | 14 of 72 |
| Search crawlers are allowed | 10 of 72 |
| Title tag present and sized | 8 of 72 |
| Meta description present and sized | 6 of 72 |
| Canonical URL declared | 6 of 72 |
| Exactly one H1 | 6 of 72 |
| Sectioned with H2s | 4 of 72 |
| Linked into the rest of the site | 6 of 72 |
| Sitemap published at the site root | 6 of 72 |
| Page-level structured data | 6 of 72 |
AI Discoverability
9 signals · 30% of the temperatureCan an AI answer engine retrieve this page and lift a usable answer out of it?
Answer engines read the raw response and quote the first passage that stands on its own. This reading is about whether there is anything there to quote and whether the crawlers were let in.
| Signal | Weight within this reading |
|---|---|
| AI crawlers are allowed | 16 of 75 |
| Content is in the HTML, not built by script | 14 of 75 |
| Answers before it explains | 10 of 75 |
| llms.txt published | 6 of 75 |
| Question and answer markup | 6 of 75 |
| Machine-readable data of any kind | 8 of 75 |
| At least one heading asks a question | 4 of 75 |
| Extractable structure on the page | 6 of 75 |
| Dated content | 5 of 75 |
Technical Accessibility
9 signals · 20% of the temperatureCan a machine get this page cleanly, quickly, and on the first try?
The physical read. Protocol, redirects, response time, page weight, and the declarations that stop a parser having to guess.
| Signal | Weight within this reading |
|---|---|
| Served over HTTPS | 14 of 60 |
| Reached without a redirect chain | 6 of 60 |
| Responds quickly | 8 of 60 |
| Document is a reasonable weight | 6 of 60 |
| Mobile viewport declared | 6 of 60 |
| Language declared | 6 of 60 |
| robots.txt answers | 5 of 60 |
| Character set declared | 4 of 60 |
| Images carry alt text | 5 of 60 |
Content and Entity Strength
9 signals · 20% of the temperatureDoes this page say what it is, who published it, and why it should be believed?
Depth, authorship, identity, and freshness. This is the reading that decides whether a machine can attribute the page to a real thing in the world rather than to an anonymous block of text.
| Signal | Weight within this reading |
|---|---|
| Enough content to be worth retrieving | 12 of 62 |
| Publisher declared as an entity | 12 of 62 |
| Entity name is consistent across the page | 6 of 62 |
| Identity linked to somewhere else | 6 of 62 |
| Authorship declared | 6 of 62 |
| Open Graph title | 4 of 62 |
| Sharing image declared | 4 of 62 |
| Cites something outside itself | 6 of 62 |
| Updated within the last year | 6 of 62 |
How a signal becomes a temperature
Each of the four readings is scored as a percentage of its own available weight, and the overall temperature is the weighted blend of those four percentages. It is deliberately not one pool of 37 weights. A single pool would let a reading made of many small signals quietly outvote one made of few large ones, and then the four readings would stop meaning what they say.
Weights are ordinal, not scientific. They encode a judgement about how much each signal costs you when it is missing, based on twenty years of watching which ones actually keep pages out of results. A page blocked by robots.txt is not slightly less visible than a page missing an og:image, and the weights say so.
What it does not read
Everything here is observable from one public fetch of one URL, without logging into anything. That is the boundary, and it is a real one. This thermometer does not know your rankings, your traffic, your backlinks, your Core Web Vitals, or how the page renders after JavaScript runs. It does not crawl the rest of the site. It does not judge whether the writing is any good.
Those are real questions and several of them matter more than anything on this page. They are also not thermometer questions. The Visibility Diagnostic covers the site rather than the page, and treatment is where the ones that need building rather than measuring go.