Every variable we check.
We read each page the way Google and AI assistants read it and score it on 26 variables. Here is every one: what we measure, why it matters, and how we fix it.
All 26 variables
Content
Whether your pages answer what people ask, in a form search engines and AI assistants can use.
8 variables
- V1
Content quality and topical authority
Depth, originality and clarity of every page, whether claims are backed by examples, numbers or sources, and how many outside sources each page cites.
- Why it matters
- Google and AI assistants favour pages that cover a topic properly and back what they say. It is the heaviest variable for both.
- How we fix it
- Section-level rewrites grounded in facts already on your site, with sources added and nothing invented.
- V2
Search intent match
Whether each page's title, heading, description and body answer what people searching for it want, checked against live search results.
- Why it matters
- Google ranks the pages that give searchers what they came for.
- How we fix it
- Reframing a page around its query, or a new page for an intent no page serves.
- V9
Title tags and H1 headings
That every page has a title of 30–60 characters and exactly one H1.
- Why it matters
- The title and main heading are the first things search engines and AI read to work out what a page is about.
- How we fix it
- A new title, with wordings to choose from, and a corrected heading structure.
- V10
Content extractability
Heading and paragraph structure, whether each section opens with a self-contained answer of 40–120 words, and whether pages declare what they are.
- Why it matters
- AI assistants quote passages that are clearly structured and answer their heading's question at once.
- How we fix it
- Answer-first openings, question headings, a table of contents and stable anchors.
- V11
Author bios and E-E-A-T
Named authors in structured data, a clearly identified organisation, HTTPS and security headers, and links to about, contact and privacy pages.
- Why it matters
- Named authors and a clear organisation are trust signals for Google, and help AI systems attribute what they quote.
- How we fix it
- Organisation and author markup for the real people behind your pages. We never invent an author.
- V13
Meta descriptions
The share of pages with a description of 70–160 characters.
- Why it matters
- Google often shows the description under your title, so it decides whether people click.
- How we fix it
- A new description per page, written from the page itself.
- V17
Content freshness
How recently pages changed, by their structured data, server headers and visible dates; whether those dates agree; and how old the years the copy refers to are.
- Why it matters
- AI assistants prefer recent content on time-sensitive topics, and clear dates show a page is looked after.
- How we fix it
- Updated facts first, then the dates. A date never changes on its own.
- V26
Cross-page topical coherence
Whether your pages share a consistent focus and terminology, with no pages that stray off-topic.
- Why it matters
- A site that stays on a consistent set of topics reads as an authority on them.
- How we fix it
- Missing sections added, overlapping pages consolidated and related pages linked as a cluster.
Technical
Whether search engines can crawl, index and load your pages properly.
11 variables
- V3
Crawlability and indexability
robots.txt access, a readable sitemap, pages free of noindex, canonical tags and HTTPS.
- Why it matters
- A page crawlers cannot reach, or one marked noindex, cannot appear in search or feed an AI answer.
- How we fix it
- robots.txt, meta robots and sitemap changes, each tested against how crawlers read them.
- V4
Core Web Vitals (LCP, INP, CLS)
Loading (LCP), responsiveness (INP) and layout stability (CLS) against Google's thresholds, 70% from mobile and 30% from desktop.
- Why it matters
- Speed and stability are a measured, if modest, Google signal.
- How we fix it
- Image, font, script and caching changes in your code, through a pull request with a preview.
- V5
Mobile usability
A viewport tag on every page, mobile performance, responsive images and HTTPS.
- Why it matters
- Google ranks the mobile version of your site, so pages have to work well on phones.
- How we fix it
- Responsive theme or code fixes, checked across phone sizes before and after.
- V12
Internal linking
How many links your home page and inner pages carry, and how descriptive their anchor text is.
- Why it matters
- Internal links help crawlers find pages and see which ones matter.
- How we fix it
- Links added between related pages, with descriptive anchors, and orphan pages connected.
- V14
Image optimisation
Alt text, lazy loading, set dimensions and modern formats such as WebP, AVIF and SVG.
- Why it matters
- Alt text explains images to search engines; sized, modern images keep pages fast and stable.
- How we fix it
- Alt text written from the page and the image, plus compression and format changes.
- V15
HTTPS and security headers
HTTPS, HSTS and five security headers: CSP, X-Content-Type-Options, X-Frame-Options, Referrer-Policy and Permissions-Policy.
- Why it matters
- HTTPS is a light Google signal, and the headers are basic web hygiene.
- How we fix it
- Header changes at your host or in your code, with a compatibility check and instant rollback.
- V16
URL structure
Clean URLs: no query strings, under 100 characters, lower case, hyphens and a clear hierarchy.
- Why it matters
- Short, readable URLs are easier for search engines and people to understand.
- How we fix it
- URL changes as one bundle: the new address, its redirect, canonical, sitemap and internal links.
- V19
Canonical tags
Whether every page has a canonical tag, and whether it points to itself.
- Why it matters
- Canonicals tell search engines which version of a page is the main one, so duplicates do not split its signals.
- How we fix it
- The right canonical set on each page, checked to be the only one and to match the sitemap.
- V20
Hreflang
Language and region tags, an x-default and at least two languages where a site has them.
- Why it matters
- Hreflang tells Google which language or region version to show to whom.
- How we fix it
- Language tags added as a set, so every version points to every other.
- V23
XML sitemap
A sitemap that parses, with last-modified dates and no errors.
- Why it matters
- A dated sitemap helps search engines find and re-crawl your pages efficiently.
- How we fix it
- A generated sitemap, kept in step with your pages and listed in robots.txt.
- V24
Pagination handling
Next and previous links and canonical tags on paginated series.
- Why it matters
- Correct pagination lets search engines follow a series of pages to the end.
- How we fix it
- Pagination markup and canonicals across the series.
Markup
The structured data and tags that describe your pages to search engines, AI assistants and social sites.
3 variables
- V6
Structured data (Schema.org)
Which schema types each page carries against what your kind of site needs, valid FAQ markup, an Organisation with sameAs links, and whether the markup is in the page's HTML before any JavaScript runs.
- Why it matters
- Structured data tells search engines and AI systems exactly what a page and a business are.
- How we fix it
- Typed JSON-LD inserted or replaced, validated, and checked in the HTML the page serves.
- V21
Open Graph and Twitter Card
Open Graph and Twitter Card tags on your home page and across your pages.
- Why it matters
- They control how your links look when shared on social platforms and in messages.
- How we fix it
- Share tags set per page, with the image checked to load.
- V22
Breadcrumb navigation and schema
BreadcrumbList markup on every page.
- Why it matters
- Breadcrumbs show where a page sits in your site, which gives AI systems its category.
- How we fix it
- Visible breadcrumbs and matching markup, added together.
Reputation and AI access
Signals from beyond your own pages, and whether AI crawlers can reach you at all.
4 variables
- V7
Backlink profile
An estimate from public signals: the domain's age, its PageRank and the sites that link to it.
- Why it matters
- Links from other sites still count for Google.
- How we fix it
- Outreach you approve, sent for you and tracked. We never promise a link will be placed.
- V8
Brand mentions and entity citations
Whether your business has a Wikidata entry and a Wikipedia article, how many authoritative profiles your sameAs links name, and listings on review and directory sites.
- Why it matters
- AI assistants recommend brands they can identify as a distinct entity, confirmed by sources they trust.
- How we fix it
- One consistent identity across your site, your schema and the profiles you own.
- V25
Local SEO signals
LocalBusiness markup and a complete name, address and phone number; for local businesses, whether they match everywhere. It weighs five times more for a local business.
- Why it matters
- For a local business, a complete and consistent name, address and phone number is the base of local search and local AI answers.
- How we fix it
- Your details made consistent on your site, your Google Business Profile and directories.
- V27
AI crawler visibility
Whether robots.txt lets the AI crawlers of 8 providers in, across 13 crawler names, and how much of each page is readable without running JavaScript.
- Why it matters
- ChatGPT, Claude and Perplexity can only use pages their crawlers may fetch, and they do not run JavaScript.
- How we fix it
- robots.txt rules per AI crawler, set the way you choose, and server-rendered content where pages need it.
How the scores work.
Every audit gives each variable a score, and each page its findings. This is how they add up.
- 26
- variables, each scored 0–100
- 4
- tiers, weighted 5, 4, 2 and 1
- 2
- scores: one for Google, one for AI assistants
- 14
- page-level findings, each with a severity
Two scores, out of 100
Each variable scores 0–100. They roll up into a search score for Google and an AI score for AI assistants, so you can see where you stand with each.Weighted by what matters most
Foundational variables count five times as much as marginal ones: weights of 5, 4, 2 and 1 across the four tiers.Fitted to your kind of site
We recognise six kinds of site, from local businesses to online shops, and shift the weights where they differ: local signals count five times more for a local business.Ranked by impact
Every variable below 90 is ranked by how far it falls short times how much it counts. The top seven become your priorities, and the fixes start there.
The findings behind the fixes.
Scores say how a site is doing. Findings say exactly what to change: the page, the tag and the line. These 14 are what our fixes are written from.
- F1Source citations. Citations to authoritative sources per 500 words.
- F2Quantified claims. Specific numbers per 300 words, on pages long enough to need them.
- F3Source age. How old the oldest source a page cites is.
- F4FAQ markup matches the page. Whether FAQ structured data says what the visible page says.
- F5Knowledge Panel. Whether Google knows your business as an entity.
- F6Consistent details elsewhere. Your name, address and phone on third-party profiles.
- F7Promotional language. Marketing superlatives per 1,000 words.
- F8Question headings. The share of section headings phrased as questions.
- F9Table of contents. Anchor links on long pages with several sections.
- F10AI crawler rules. The exact robots.txt line that admits or blocks each AI crawler.
- F11Canonical problems. Missing, duplicated, off-site or wrong canonical tags, page by page.
- F12Page titles. Each title outside 30–60 characters.
- F13Meta descriptions. Each description outside 120–160 characters.
- F14Image alt text. Each image with no alt text.
What we do not score, and why.
llms.txt
We used to score it. We stopped when the studies we reviewed found no measurable effect on how often AI assistants cite a site. We still tell you whether you have one.Rankings we cannot see
Your positions and clicks come from your own Google Search Console, connected in your portal, never from estimates presented as fact.AI assistants added together
ChatGPT, Claude, Gemini and Perplexity answer differently. We track each one on its own and never add them into a single number.