The Rubric · sixteen checks · two points each
What the checks check, in plain words.
The Pulse Check reads a website the way a search engine or an AI assistant does, with no scripts running, and asks sixteen questions of what it finds. This page lists them: what each one looks at, why it matters, and what green, amber and red mean. The same sixteen run on this site, in your browser, every time it loads.
The sixteen checks.
Grouped the way a machine meets them: first whether it can read you at all, then whether it knows who you are, then whether it can pass you on.
- 01One topic. The page says what it is about, once, at the top. Green: Exactly one h1. Amber or red: No h1, or more than one. Search engines and assistants have to guess the topic, or watch the page argue with itself.
- 02Outline. The headings read like a table of contents, in order. Green: Heading levels never skip. Amber or red: A level is skipped, h2 straight to h4. A machine loses the structure.
- 03Words the machine gets. What is really on the page before any scripts run. Green: 300 or more words in the HTML itself. Amber or red: Under 80 words, or an empty shell that scripts fill in later. Most assistants read it as blank.
- 04Browser tab. The tab title says what and where. Green: A title of 15 to 65 characters. Amber or red: Missing, or too long to show in full.
- 05Search summary. The one-line summary a search result shows is written by you. Green: A description of 50 to 160 characters. Amber or red: Missing. Search engines will invent one.
- 06One address. There is one official address for this page. Green: A canonical link is set. Amber or red: Missing. www and non-www, http and https can all compete as separate pages.
- 07Introduces itself to machines. The page tells machines what kind of thing it is. Green: Structured data (JSON-LD) that parses, naming three or more entities. Amber or red: None, or data that does not parse. To a machine this page is text with no identity.
- 08Who you are. A machine can find your business name and a way to reach you. Green: A business entity with a name and an address or an email. Amber or red: No business entity, or one with no way to reach it.
- 09Named people. A real person stands behind the page. Green: At least one named person in the entity data. Amber or red: Nobody named. Assistants weigh pages with a named author or founder more heavily.
- 10Answers questions. The page answers the questions people actually ask, in a form an assistant can lift. Green: An FAQ entity with questions and answers. Amber or red: Question-shaped headings but no entity, or no questions at all.
- 11Language declared. The page says what language it is in. Green: A lang attribute on the html element. Amber or red: Missing.
- 12Works on a phone. The page tells phones how to display it. Green: A viewport meta tag. Amber or red: Missing. Phones render it as a shrunken desktop page.
- 13Pictures described. Every picture has words a machine can read instead. Green: Every image has alt text, or there are no images. Amber or red: Images with no alt text.
- 14Fonts downloaded. Fonts a visitor has to download before the words appear. Green: None. Text renders instantly. Amber or red: One or more font sources. Each is a download before the text can render.
- 15Links go somewhere. Every link leads to a real page. Green: Every link has a destination. Amber or red: Links that point at #, nothing, or javascript:.
- 16Shares well. When someone pastes your link in WhatsApp or LinkedIn, it shows a title and summary. Green: Open Graph title and description present. Amber or red: Missing. The link shows bare, or with whatever the app guesses.
How the score works.
Two points for green, one for amber, none for red. Sixteen checks, so the most a page can score is 32. The bands are deliberately blunt.
- 28+Working for you. Fix the ambers and it is done.
- 20+Mostly working, with real gaps. The reds are costing you.
- 12+Working against you more than for you.
- <12Invisible to most of the machines that matter.
How this site grades itself.
The rubric on our home page is the same instrument pointed at ourselves, with the checks that only make sense for our own pages swapped in: that our FAQ answers match the ones we give machines, that both founders are properly named, that the opening paragraph is readable. Sixteen rows, 32 points, and the one we cannot pass yet is named.
The Nudge rubric · this site16 rows · 32 points
- Structureone h1
- Outlineno skipped heading levels
- Served words600 or more words in the HTML
- Titlenames Nudge and Johannesburg
- Description50 to 160 characters
- Canonicalset
- Entitiesnine or more in the entity data
- FAQ mirrorthe answers given to machines match the answers on the page, word for word
- Speakablethe two marked sentences exist on the page
- Peopleboth founders with job title and surname
- Linksno link points at #
- Contrastthe opening paragraph measures at least 4.5:1, computed from the live colours
- Typeno webfont downloaded
- Motiona reduced-motion rule exists
- Readingthe opening paragraph reads under grade 8
- Client proofa named client with a result we are allowed to publish
What the checks cannot see.
- What does a Pulse Check score mean?
- Each of the sixteen checks scores two points for green, one for amber and none for red, so the most a page can score is 32. Above 27 the site is working for you. Below 11 it is invisible to most of the machines that matter.
- Does the Pulse Check use AI?
- No. Every check is a function on the HTML your site serves. The same code runs on this page. An opinion would be a guess; a check is evidence.
- What does it miss?
- Whether you rank, what an assistant actually says about you against your competitors, whether your story is true, and whether the business underneath works. Those need people. That is the Signal Audit.
A check is evidence. A score is a claim.
That is why every row shows what it measured. Run it on your site, then on a competitor.
Run the Pulse Check