Every check explained
The SEO, GEO and AEO checks the site audit runs (about 60), what each one looks for and why it matters.
The site audit runs about 60 checks. They are listed here the way the app groups them under "Every check": by pillar, then by topic. Each row says what is checked, how serious it is, why it matters and how to fix it. The audit tab itself is described on Audit.
How serious each issue is
| Level | What it means |
|---|---|
| "Critical" | Blocks search or AI on its own, or costs the most. Fix these first. |
| "Warning" | Holds rankings or citations back. |
| "Info" | A smaller gain, or something to confirm is intended. |
"(whole site)" marks checks that only run in a whole-site audit, because they compare pages with each other or read the whole sitemap.
"Search" (SEO)
Can Google crawl, understand and rank your pages? 45 % of the score.
Page basics
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| Title tag | Critical if missing · Warning if under 30 or over 60 characters | The title is the clickable headline in Google and the first thing search engines and AI read. Too short wastes it; too long gets cut off. | Give each page a unique title of about 50–60 characters that names the page, then your brand. |
| Meta description | Warning if missing or over 160 characters · Info if under 120 | The text under your link in Google. Without one, Google writes its own; too long, it is cut. | Write 140–160 characters with a clear benefit and a reason to click. |
| Duplicate titles and descriptions (whole site) | Warning for titles · Info for descriptions | Pages with the same title compete for the same searches and Google often shows only one. Repeated descriptions are ignored. | Give each page a title and description about its own topic. |
| H1 heading | Critical if missing · Warning if more than one | The H1 tells Google and AI what the page is about. Several H1s split that signal. | Keep one clear H1 near the top; turn the others into H2. |
| Heading order | Info · Warning if more than 3 skips | Skipping levels (H2 to H4) blurs the outline search engines and AI read to pull out answers. Headings in menus and footers are ignored. | Use headings in order: H1, then H2, then H3, with no gaps. |
| Image alt text | Info · Warning if more than 5 images | Alt text is how search, image search and screen readers understand pictures. | Add a short description to every meaningful image. |
| Page language (lang) | Info | Without <html lang>, search engines and screen readers can misread the page language. | Set <html lang="…"> to the page language, for example lang="en". |
| hreflang x-default | Info | Checked when a page declares several language versions. x-default says which page to show people whose language you don't target. | Add an x-default hreflang pointing to your main page. |
| Open Graph tags | Warning if all missing · Info if some are missing | og:title, og:description, og:url and og:image control how your link looks when it is shared. | Add the four tags to every page, usually through your SEO plugin. |
Indexing
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| Canonical tag | Warning if missing, invalid or pointing to another domain · Critical if pointing to the homepage · Info if pointing to another page | The canonical tells Google which address is the real one. A wrong one can drop the page from search or credit another address. | Make each page's canonical point to its own address, unless the page really is a copy. |
| noindex | Critical on the homepage (score capped at 20) · Warning if the page is in your sitemap · Info otherwise | noindex (meta robots or X-Robots-Tag) hides a page from Google. Often a staging setting left on at launch. Legal and utility pages are not flagged. | Remove noindex from pages that should rank; take the others out of your sitemap. |
robots.txt and sitemap
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| robots.txt | Warning if missing · Critical if it blocks the whole site (score capped at 20) · Info if it has no Sitemap line | robots.txt tells crawlers what they may read and where your sitemap is. "Disallow: /" under "User-agent: *" removes you from search. | Publish robots.txt at your domain root, blocking only paths you want hidden, with a line Sitemap: https://yoursite.com/sitemap.xml. |
| XML sitemap | Warning if none is found | The sitemap lists every page you want indexed, so Google finds pages your links might miss. | Generate sitemap.xml (most CMSs and SEO plugins do) and reference it in robots.txt. |
| Sitemap entries (whole site) | Warning if addresses return an error (Critical above 5) or don't respond · Info if they redirect | The sitemap should list only live, final addresses. Errors and redirects waste Google's crawl time and blur which address is the real one. | Remove dead addresses or restore the pages; replace each redirect with the address it lands on. |
Structured data
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| Structured data (JSON-LD) | Warning | The page has no JSON-LD. Schema earns rich results and tells AI what the page is. Not flagged on legal and utility pages. | Add JSON-LD for the page type: Organization, Article, Product, FAQPage… |
| Schema fields (8 types) | Critical if a required field is missing (Warning for Article and BlogPosting) · Info if more than half the recommended fields are missing | Checked on Organization, LocalBusiness, Article, BlogPosting, FAQPage, BreadcrumbList, Product and SoftwareApplication. The row names the missing fields. | Add the fields listed (the row shows an example), then test the page in Google's Rich Results Test. |
Required fields per schema type: Organization: name · LocalBusiness: name, address · Article and BlogPosting: headline, datePublished · FAQPage: mainEntity · BreadcrumbList: itemListElement · Product: name · SoftwareApplication: name.
Links and site structure
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| Dead-end page | Critical | The page has no links to other pages of your site: crawlers and visitors stop there. | Add links to related pages and a clear next step. |
| Few internal links | Warning (fewer than 3) | Internal links pass authority and help Google and buyers find related pages. | Add links in the text to related pages and to key pages such as pricing or contact. |
| Generic link text | Info (over 30 % of the page's links) | "Click here" or "read more" tells Google nothing about the page linked to. | Use link text that names the destination topic. |
| Broken internal links (whole site) | Warning | Links to pages that return an error lead visitors to dead ends and pass no value. | Point each link to a live page, or remove it. |
| Orphan pages (whole site) | Warning | No audited page links to this page, so Google rarely finds or ranks it. Only checked when the audit covered the whole site. | Link to it from a related page, a hub page or the menu. |
| Unreachable pages (whole site) | Info | Other pages link here, but none of them can be reached from your homepage either. Only checked when the audit covered the whole site. | Connect this group of pages to your menu or a hub page. |
| Pages too deep (whole site) | Info (4 clicks or more from the homepage) | Deep pages are crawled less often and receive less authority. | Bring key pages within 3 clicks of home through hub pages, breadcrumbs or related links. |
| Missing or thin site sections (whole site) | Warning if a core section is missing · Info otherwise | Your addresses are compared with a model for your kind of site (SaaS B2B, Local Service Business, E-commerce or a general website): pricing, contact, guides, service pages… A recommendation, not a defect. | Create the pages named, at addresses like the ones the row suggests, and link them from your menu or a hub page. |
Speed
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| Core Web Vitals: LCP, INP, CLS | Warning above Google's "Good" level · Critical at "Poor" | Real-visitor speed from Chrome over 28 days, reported once for the site. Good: LCP 2.5 s, INP 200 ms, CLS 0.1. Poor: 4 s, 500 ms, 0.25. Only for sites connected to Search Console with enough Chrome visitors. | LCP: compress images and preload the main one. INP: break up long scripts and defer the rest. CLS: reserve space for images, embeds and fonts. |
"AI Answers" (GEO)
Can AI assistants reach your site, trust it and quote it? 35 % of the score.
AI access and quotable content
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| AI crawlers blocked in robots.txt | Warning | Your robots.txt stops AI assistants from reading your site, so they cannot quote it. Bots checked: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-Web, anthropic-ai, PerplexityBot, Google-Extended, Applebot-Extended, CCBot. | Allow the bots of the assistants you want to be cited by, unless blocking them is your choice. |
| llms.txt | Info | llms.txt gives AI assistants a short, curated map of your most important pages. | Publish an llms.txt at your domain root listing your key pages, one line each. |
| Definition sentences | Warning | Checked on pages with more than 30 sentences: fewer than 1 in 20 reads like a definition ("X is a …"). AI assistants prefer to quote clear definitions. | Add a short definition near the top: what it is and who it is for, in one or two sentences. |
Who you are
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| Organization schema | Critical if missing on the homepage · Warning if it has no sameAs | Organization schema ties your brand together for Google and AI. sameAs links to your official profiles confirm you are a real, known business. | Add Organization JSON-LD on the homepage with name, logo, url and a sameAs list (LinkedIn, Wikidata, social profiles). |
| Author | Warning if an article has no visible author · Info if the byline has no Person schema or the author has no sameAs | Checked on article pages. Search engines and AI weigh who wrote a page; an unnamed or unlinked author reads as anonymous. | Show a byline, mark the author up as Person in the Article schema, and link their profiles with sameAs. |
| Contact details (whole site) | Critical if no phone or e-mail anywhere · Warning if more than 2 phone numbers, or more than one address in your schema | Name, address and phone prove you are a real business to buyers, Google and AI. Conflicting details confuse them. | Show one consistent name, address and phone everywhere, ideally in LocalBusiness or Organization schema too. |
Proof and trust
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| "0" placeholder figures | Warning | A "0 customers" or "0+ projects" tile reads as no traction. It is usually a counter that was never filled in. | Put the real figure in, or remove the tile until you have one. |
| Case studies (whole site) | Critical if none · Warning if fewer than 3 | Counted from pages under /case-studies/, /customers/, /stories/ and similar. Proof of results convinces buyers and gives AI something to cite. | Publish at least 3 case studies with real outcomes, across different uses or industries. |
| Customer logos | Info | Checked on the homepage: no "trusted by" section or strip of client logos was found. | Add a strip of real customer or partner logos. |
| Security or trust page | Warning | No page at /security, /trust or /compliance. Business buyers look for one before they sign. | Publish a page on how you handle data, security and compliance. |
"Answer Boxes" (AEO)
Do your pages offer questions and short answers that featured snippets can use? 20 % of the score.
Questions and answers
| Check | Level | What it means and why it matters | How to fix |
|---|---|---|---|
| FAQ block | Info | The page has no questions-and-answers block. FAQs are the format featured snippets and AI answers reuse most. | Add 5–8 real buyer questions with answers of 40–80 words, marked up with FAQPage. |
| FAQ without FAQPage schema | Warning | The page has question-and-answer content but no FAQPage markup, so search and AI can't reliably extract it. | Wrap the existing questions and answers in FAQPage JSON-LD. |