Key takeaways
- Site readiness is the first layer of an AI visibility audit: 93 checks, T01 to T93, that apply regardless of which AI product is being measured.
- They answer four questions in order: can bots reach the page, can they read it, can they trust it and can they tie it to the right entity.
- Most checks need real evidence, such as a robots line, a status code, a log entry or an owner-confirmed fact. Without it, the item is reported as partial or unmeasurable, never as passed.
- Fixing this layer helps classic search and AI answers at once, which makes it the best place to start.
Why site readiness comes first
AI answer systems do not invent their knowledge of your business from nothing. They draw on pages they could reach, content they could read, facts they found consistent and identities they could resolve. When any of those steps fails, the result is not a polite “no information”. It is an answer assembled from other sources: a directory listing, an old article, a competitor, or a namesake.
That is why the audit begins with the website itself. Surface measurements are interesting, but they are hard to act on if you do not know whether the site layer is healthy. A missing citation can have a technical cause that takes an hour to fix, or a strategic cause that takes months. The site layer lets us tell which one we are looking at.
The 93 checks are numbered T01 to T93. They are grouped on this page into eleven themes so that they are easier to navigate, but the numbering is the official order in the catalog. Every check has a stable ID that reports and re-tests refer to.
The four questions every site check serves
1. Can crawlers reach the page?
Access covers robots.txt, bot-specific rules for search crawlers such as OAI-SearchBot, Googlebot and Bingbot, whether a content delivery network or firewall lets verified bots through, and whether a login wall or form hides the main content. The distinction between a rule that says “allowed” and a server that actually answers a real bot is central. Robots permission is something you can read in a file. Real access is something you have to verify against logs and official bot identities.
2. Can they read the page?
Reading covers HTTP status codes, redirect chains, canonical addresses, noindex directives, snippet limits, rendering, mobile layout, and whether decision-critical text arrives in the initial HTML or only after scripts run. A page that returns a friendly 200 status but shows an empty shell to a bot is a common cause of silent invisibility.
3. Can they trust the page?
Trust covers a named author or reviewer, primary sources, dated prices, honest scope for offers, careful language instead of absolute promises, and original data presented with its method. It also covers the basics of safety and quality: HTTPS, accessibility and readable structure.
4. Can they tie it to the right entity?
Identity covers brand names, legal entities, branches, people and their roles, structured data, sameAs links, addresses, phone numbers and hours. It matters most for local businesses, professional services and multi-brand groups, where a confused identity produces answers that are fluent, confident and wrong.
How the checks are evaluated
Each check is reported in one of four states: completed, partial, unmeasurable or not applicable. The evidence rules are strict.
- Observable facts are collected by crawling the site directly: status codes, headers, HTML, structured data, links and sitemaps.
- Owner-supplied facts are used where only the owner knows the truth: access logs, current prices and their dates, the person responsible for updates, the correct address of each branch, and which pages are meant to exist.
- Provider data is used where a search product exposes it, such as Google Search Console URL Inspection (T68) and Bing URL status (T69). It is stored with its date.
- Not-applicable decisions are made in the open. T90 records why an item does or does not fit the sector and page type, so the evaluation scope is visible.
The order of work matters too. Access checks come before content checks, because a page that cannot be reached makes later findings moot. Identity checks come before surface measurements, because answers about a business are only trustworthy if the business is clearly defined.
The complete list of 93 site-readiness checks
Every record below belongs to the site layer. The groups are for navigation only. Use the record ID when you discuss a finding with your team or with Smartise.
Query fit and answer quality (T01–T08)
| Record | Check | What it examines |
|---|---|---|
T01 | Target page for the query | Checks whether a page exists that matches the question being asked, and whether the text on that page actually supports the answer. |
T02 | Title and query intent | Compares the page title with the question the page really answers, so readers are never misled by a title that promises something else. |
T03 | Direct answer | Looks at whether the page gives an explicit answer to the question and states the conditions under which that answer holds true. |
T04 | Coverage of decision questions | Records whether the questions customers ask when making a decision in your sector are answered on the site, each one tied to the page that answers it. |
T05 | Branded and unbranded queries | Reports searches that include your brand name separately from searches that do not, and evaluates the two groups on their own terms. |
T06 | Topic overlap between pages | Examines whether pages that try to answer the same question compete with each other unnecessarily, while separating legitimate language or product variants. |
T07 | Discoverability through internal links | Calculates whether the target page can be reached through the site's own links, and whether that path sits inside the crawled scope. |
T08 | Sector-specific decision information | Evaluates whether the information a customer needs for your sector and page type is present, without forcing irrelevant items onto the page. |
Crawler access and permissions (T09–T14)
| Record | Check | What it examines |
|---|---|---|
T09 | robots.txt access | Checks whether the robots file can be read and which allowances or restrictions it defines for the target page. |
T10 | OAI-SearchBot permission | Calculates whether OpenAI's search crawler is allowed to reach the target page and records the exact rule line that decides it. |
T11 | Googlebot robots permission | Calculates whether Google's crawler is allowed to reach the target page under the robots rules. |
T12 | Bingbot robots permission | Determines whether Microsoft's Bing crawler is allowed to reach the target page and the related domain under the robots rules. |
T13 | Real bot access through CDN and WAF | Tries to understand, by comparing against the site owner's access logs, whether a firewall or content delivery network blocks genuine bots. |
T14 | Bot identity verification | Verifies that a bot really belongs to the company it claims to, using official address records instead of trusting the user-agent text alone. |
HTTP responses, indexing signals and sitemaps (T15–T25)
| Record | Check | What it examines |
|---|---|---|
T15 | HTTP response of the target URL | Inspects the status code the target page returns and separately records any content or redirect problem it carries. |
T16 | Redirect chain | Checks the redirects leading to the page for loops, wrong destinations and needlessly long chains, step by step. |
T17 | Canonical target consistency | Audits whether the preferred address the page declares is reachable and consistent with the topic of the page. |
T18 | Old domain migration | Evaluates whether old addresses permanently redirect to the correct new content and whether the move is backed by proof of ownership. |
T19 | noindex and X-Robots-Tag | Looks for directives that keep the page out of the search index and whether those choices agree with one another. |
T20 | Snippet restrictions | Detects directives that fully or partly limit how page content may be shown as a summary, and distinguishes their scope. |
T21 | XML sitemap access | Checks that the sitemap file is reachable, that its address format is correct and whether pages are missing from it. |
T22 | Sitemap last-modified dates | Compares the last-modified dates in the sitemap with real content changes by sampling pages. |
T23 | Forms and login walls | Determines whether a form or login is required to reach the main content and which information sits behind that barrier. |
T24 | Raw HTML versus rendered page | Compares whether important text arrives in the first page load or only through code that runs later, and logs loading problems. |
T25 | Training and content-use preferences | Reports, role by role, the preferences your site sets for bots regarding AI training, search and user-initiated access. |
Structure, readability and accessibility (T26–T35)
| Record | Check | What it examines |
|---|---|---|
T26 | H1 and section hierarchy | Checks whether the main heading and subheadings create a clear, understandable order for the reader. |
T27 | Clarity of the first answer paragraph | Inspects whether the first answer paragraph states the core scope of the question and the conditions that apply. |
T28 | Table headers that make sense | Verifies that tables tie each value to the right row and column headers and to its unit. |
T29 | Access to key information on mobile | Compares desktop and phone views to confirm important information and interactions are reachable without truncation. |
T30 | Units and currency | Checks that numeric information is given with value, unit, currency and scope so that it can be understood on its own. |
T31 | Paragraph and list readability | Finds examples that make reading harder in paragraphs and lists and shows them with the relevant text fragment. |
T32 | Visible FAQ consistency | Compares a frequently-asked-questions block, where present, against the visible questions and answers to confirm they mean the same thing. |
T33 | Breadcrumbs and page relationships | Compares the on-page navigation path with structured data and address structure to confirm they describe the same hierarchy. |
T34 | Images and alternative text | Checks that informative images convey their meaning through alternative text, and that decorative images are handled correctly. |
T35 | Content accessibility | Evaluates accessibility for disabled users through automated findings together with keyboard and reading-order checks. |
Trust, pricing and evidence (T36–T45)
| Record | Check | What it examines |
|---|---|---|
T36 | Sufficient content for a decision | Examines whether the page has what a customer needs to decide, rather than simply counting words. |
T37 | Price scope | Audits whether what is included or excluded in the price, and the conditions under which an offer is valid, are explained clearly. |
T38 | Price validity date | Checks whether the page states the date and period for which a price applies, compared with the owner's reference. |
T39 | Publication and modification dates | Compares the visible and structured publication and update dates with each other and with the real change history. |
T40 | Author and expert review | Looks at whether the identity and competence of the person who wrote or reviewed the content are supported by suitable evidence. |
T41 | Primary evidence and references | Records whether important claims rest on reachable, appropriate primary sources, together with the relevant passage. |
T42 | Does the source support the claim | Compares the claim with the passage in the cited source to see whether it supports it in the same scope and period. |
T43 | Promotional language and absolute promises | Finds unprovable, absolute promises on the page and classifies the relevant text fragments with their evidence. |
T44 | Original decision data | Looks at whether data your own site produced is presented verifiably, including how it was gathered and which decision it supports. |
T45 | Outdated third-party prices | Compares prices on outside sites cited in answers with the owner's current, dated price to detect contradictions. |
Language, identity and entity clarity (T46–T57)
| Record | Check | What it examines |
|---|---|---|
T46 | User language and target language | Assesses whether the page language matches the intended user and market, reporting the confidence of the language detection. |
T47 | Content update owner | Checks whether the person responsible for updating content, their scope and the review process are confirmed by the owner's records. |
T48 | Brand name and commercial identity | Audits whether brands and legal entities with identical or similar names are kept apart through permanent identity information. |
T49 | City and namesake separation | Verifies that businesses sharing a name are told apart by city, branch and location and are not merged by mistake. |
T50 | JSON-LD syntax | Checks whether structured data blocks are syntactically valid and logs broken ones together with their location. |
T51 | Schema type suitability | Compares the structured data type, where present, with the real function of the page and the entity it describes. |
T52 | Schema consistency with visible text | Matches values such as price, stock and organization in structured data against visible content to find conflicts. |
T53 | sameAs link identity | Audits whether links to a business's other profiles are reachable and genuinely belong to the same business or person. |
T54 | Person/expert and organization relationship | Evaluates whether the relationship between a person and an organization is described with the right role and time, without relying on name similarity. |
T55 | Address and phone consistency | Compares addresses and phone numbers given for the same branch in different places to find contradictions. |
T56 | Business profile website address | Checks, with the owner's permission, whether the web address in a business profile points to the correct business, branch and site. |
T57 | Service and organization relationship | Audits whether the entity that provides a service or product, and the branch, are reflected correctly through identity links. |
Technical delivery and performance (T58–T67)
| Record | Check | What it examines |
|---|---|---|
T58 | HTTPS and secure connection | Checks that the site is served over a secure connection, that the certificate is valid and that page resources load securely. |
T59 | HTML response integrity | Verifies that the page content from the server arrives complete, uncut and with the correct content type. |
T60 | Mobile viewport definition | Separately checks whether the page defines its mobile screen settings and whether the real mobile layout works accordingly. |
T61 | LCP / INP / CLS field data | Checks whether page speed and stability data collected from real users exists and whether its coverage is sufficient. |
T62 | Server response measurement conditions | Records how the first server response time was measured, under what conditions and with how many repetitions, with its distribution. |
T63 | html lang definition | Compares whether the page language tag is valid and consistent with the true content language and the target market. |
T64 | hreflang suitability | Tests whether language and region versions link to each other reciprocally and point to the right language targets. |
T65 | www and root domain consistency | Checks that the www and non-www addresses consolidate consistently on one preferred address. |
T66 | Broken internal links | Finds permanently broken links between the site's own pages with the page they appear on and the target address. |
T67 | Conflicts between robots and HTTP headers | Explains whether robot rules and server header directives contradict each other for the same page and bot. |
Search engine data and discovery (T68–T75)
| Record | Check | What it examines |
|---|---|---|
T68 | Search Console URL Inspection | Retrieves and stores dated information about index and canonical status from Google Search Console's page inspection record. |
T69 | Bing URL status | Records which fields Bing's page status data actually returns and the date they belong to. |
T70 | Open Graph share preview | Checks whether the title, description and image shown when the page is shared on social networks are defined correctly. |
T71 | llms.txt experimental file | Records, for information only, whether the site has the experimental file prepared for AI models; it never counts toward a score. |
T72 | Crawl scope and missing page inventory | Compares crawled and known page counts to show pages left outside the scope and the limits of the crawl. |
T73 | Empty content on a 200 response and soft 404 | Separates pages that answer successfully but are empty or behave like error pages by comparing them with real content. |
T74 | Access to dependent resources of the main content | Establishes whether the styles, code, images and data files needed to display the main content are reachable and which section is affected. |
T75 | Pagination and interaction-dependent content | Tests whether decision content behind later pages, load-more buttons or expandable sections can be discovered. |
Non-text content and entity consolidation (T76–T82)
| Record | Check | What it examines |
|---|---|---|
T76 | Extractability of PDF and document content | Records whether decision-relevant text inside PDFs and similar documents is readable and stays tied to its source address. |
T77 | Text equivalent of video and audio | Checks whether video and audio that carry decision information also have captions or a written transcript. |
T78 | Decision information inside images | Compares whether price, unit or condition details live only inside an image or are also given as visible text. |
T79 | Persistent entity identity and cross-page merging | Audits whether the same business or product is linked with a consistent identity across pages while distinct entities are not mixed. |
T80 | Accuracy of brand, company, branch and seller relationships | Checks whether roles and legal relationships among brand, company, branch and seller are separated in line with reality. |
T81 | Source and scope of reviews and ratings | Separates the source of displayed reviews and ratings, the product or institution they cover, and any conflict with the on-page assessment. |
T82 | Product variant, stock and offer consistency | Compares stock and offer information across site, seller and business sources for the same product option, country and time. |
Commerce, local details and consistency (T83–T89)
| Record | Check | What it examines |
|---|---|---|
T83 | Total cost, taxes and mandatory fees | Compares the total cost to confirm that mandatory taxes, extra fees and the currency are clearly shown before payment. |
T84 | Opening hours and service area | Checks whether day, hour, holiday and service-region information is consistent across profile, owner and web records. |
T85 | Returns, cancellation, warranty and support | Compares whether returns, cancellation, warranty and support questions suited to the page type are answered clearly and match company details. |
T86 | Claim and price parity across language versions | Compares whether the same offer contradicts itself across language versions in terms, price and date, while separating legitimate country offers. |
T87 | Content differences by country and session | Records what content and price the same page shows under different country, cookie and session conditions. |
T88 | Stale content versions | Searches with concrete evidence for an outdated version of the page remaining in cache or in structured data. |
T89 | Duplicates in filter and search URLs | Examines whether filter, parameter and search addresses create needless duplicate pages, separating legitimate variants. |
Scope, governance and decision gaps (T90–T93)
| Record | Check | What it examines |
|---|---|---|
T90 | Fit to sector and page type | Decides, with a reason, whether each item applies to the business's sector and page type and records the evaluation scope. |
T91 | Inventory of external record corrections | Lists incorrect records found on the web with their source and the person responsible for the fix, and tracks request sent and corrected separately. |
T92 | Scan for misleading content shown to bots | Looks at whether users and AI bots are shown different content and leaves suspicious cases for human review. |
T93 | Missing decision pages | Identifies pages and content clusters that do not exist, or remain incomplete, for the questions customers ask at each decision stage. |
Where to start if you have one afternoon
Not every site needs all 93 checks reviewed at once. If you have a limited amount of time, this order gives the best return.
- Read your robots.txt (T09 to T12). Make sure you are not blocking search crawlers by accident, and that your rules for different bots are deliberate. Keep the preference about training crawlers separate from the preference about search (T25).
- Test a real page as a bot (T13, T14, T15). Confirm that a verified crawler gets a 200 response and the same content a visitor sees, not a challenge page.
- Check your canonical and noindex setup (T17, T19, T20, T67). Conflicts between robots rules, meta directives and HTTP headers are among the most common self-inflicted problems.
- Open your sitemap (T21, T22). It should list the pages you want found, with honest last-modified dates.
- View source on a key page (T24, T74). Confirm the answer, the price and the key facts are in the initial HTML.
- Compare your facts (T55, T56, T84). Put your site, your business profile and your top directory listings side by side for address, phone and hours.
Common failures by layer
Access failures
- A security plugin or CDN rule challenges unknown bots and also blocks legitimate ones.
- Robots.txt allows a bot, but a server header or meta tag says noindex or nosnippet, creating a conflict (T67).
- Pages are only reachable after a form submission or login, so the content that matters is not available to crawlers (T23).
Indexing failures
- Redirect chains and loops after a redesign or domain move (T16, T18).
- Canonical tags that point to a different page, a redirected page or a page that returns an error (T17).
- Soft 404 pages that return a 200 status with an empty or error-like body (T73).
- Filter and search URLs that generate thousands of near-duplicate pages (T89).
Content failures
- A heading structure that does not help a reader or a parser find the answer (T26).
- Decision facts such as prices or units locked inside images, PDFs or videos without a text equivalent (T76 to T78).
- Marketing claims that are absolute and unsupported (T43).
Trust failures
- No author or reviewer, or one whose expertise is not shown (T40, T54).
- Prices without a date, without a scope or without taxes and mandatory fees (T37, T38, T83).
- Reviews and ratings whose source or scope is unclear (T81).
Entity failures
- Different addresses or phone numbers on different pages for the same branch (T55).
- Structured data that disagrees with visible text, for instance a price in markup that is not on the page (T52).
- sameAs links pointing to profiles that belong to a different business (T53).
Delivery failures
- Mixed content or an invalid certificate (T58).
- Missing viewport settings or layouts that hide important information on phones (T29, T60).
- Poor field data for loading speed, interaction and visual stability (T61, T62).
Structured data, identity and generative engine optimization
Structured data is often presented as a trick. It is better understood as a way to state facts unambiguously. The relevant checks (T50 to T57) are therefore less about adding as much markup as possible and more about accuracy. The markup must parse, it must describe the real function of the page, it must agree with the visible text, and its identity links must point to the right places.
The same logic applies to generative engine optimization in general. AI systems are more likely to use a page that states clearly who it is about, what it offers, where and when, and under what conditions. They are less likely to rely on a page that contradicts itself or that leaves the reader to guess. Nothing on this list is exotic; the value comes from doing the basics consistently and being able to prove it. For a broader introduction, see the AI search visibility guide.
What the site checks do not do
It is as important to know the limits. Passing the site layer does not guarantee an AI citation. A site can be perfectly reachable and clearly written and still not be chosen for a given question, because that depends on relevance, competition and each system’s own source selection. The site layer removes avoidable obstacles. The surface layers, described on the ChatGPT and Google AI and Gemini, Copilot and Yandex pages, then measure what actually happened.
Likewise, T71 records whether an experimental llms.txt file exists, but only as information. It never counts toward a score, because no major provider has made it a requirement. T25 reports your preferences for training, search and user-initiated access role by role, without turning a deliberate policy choice into a penalty.
If you want these checks run on your own site, contact Smartise. If you want the context first, start with the the audit overview and the explanation of the 397-check catalog.
Frequently asked questions about site readiness
What are the site-readiness checks?
They are 93 the audit checks, numbered T01 to T93, that test whether crawlers can reach a website, read its pages, trust its content and tie it to the right entity. They apply regardless of which AI product is being measured.
What is the difference between robots permission and real bot access?
Robots permission is what your robots.txt file allows, which you can read in the file. Real access is whether a verified bot actually receives the page through your firewall or CDN, which has to be confirmed with server logs and official bot identity records (T10 to T14).
Does fixing these checks help normal Google search too?
Yes. Crawl access, indexing signals, clear structure and consistent facts help classic search and AI answers alike, which is why the site layer is the best place to begin.
Is llms.txt part of the score?
No. T71 records whether the experimental llms.txt file exists, for information only. It never counts toward a score.
What happens if a check cannot be verified?
It is reported as partial or unmeasurable, with the reason. For example, checking whether a firewall blocks real bots needs the site owner’s access logs. Without them, the item is not marked as passed.
Do all 93 checks apply to every website?
No. T90 records whether each check fits the sector and page type. Price checks, for example, do not apply to a page with no prices, and the reason is written down.
Will passing all site checks guarantee AI citations?
No. Site readiness removes avoidable obstacles, but AI systems choose sources based on relevance, competition and their own selection methods. Surface measurements show what actually happens.
Continue exploring the AI visibility audit
AI Search Visibility Audit
What the audit is, who it is for, what a report contains and how it differs from a rank tracker.
Read more about AI Search Visibility Audit →397-Check Audit Catalog
How 397 records break down, how to read the numbers and the 30 shared measurement rules.
Read more about 397-Check Audit Catalog →ChatGPT, AI Overviews & AI Mode Checks
The 58 checks specific to ChatGPT Search, Google AI Overviews and Google AI Mode.
Read more about ChatGPT, AI Overviews & AI Mode Checks →Gemini, Copilot & Yandex Checks
The 66 checks specific to Gemini, Microsoft Copilot and Yandex (Russia).
Read more about Gemini, Copilot & Yandex Checks →AI Search Visibility (GEO) Guide
A practical, hype-free guide to being found, read and cited by AI answer systems.
Read more about AI Search Visibility (GEO) Guide →Ready to grow with Smartise?
Tell us about your business and goals. We will recommend the smallest set of changes that makes the biggest difference, and tell you plainly what we can and cannot promise.