Skip to content
Leadee
Methodology

Every AI crawlability check, explained

Leadee's AI crawlability report scores a site against 170 checks, catalog version 2026-10-28, covering crawlability, server-rendered content and schema markup so AI engines can read and cite a page. Each entry below explains what it detects, why it matters, and how to fix it, and links back from the matching issue in your own report.

Updated

AI access

AI search bots allowed in robots.txt

access.robots_search_bots

Severity
high
Effort
S
Weight
1
Who can fix
site admin

Blocking an AI search or answer bot (e.g. OAI-SearchBot, PerplexityBot) in robots.txt prevents your site from appearing in that engine's AI-powered answers entirely.

How to fix: Remove the Disallow rule for the listed bots, or scope it so it no longer covers the whole site.

AI training bot policy

access.robots_training_bots

Severity
info
Effort
S
Weight
0
Who can fix
site admin

Blocking a training-only bot (e.g. GPTBot, CCBot) is a legitimate policy choice and does not remove you from AI search answers — this is informational, not a score deduction.

How to fix: No action required; this reflects your own robots.txt policy choice.

robots.txt is healthy

access.robots_txt_health

Severity
high
Effort
S
Weight
1
Who can fix
site admin

A robots.txt that errors, is oversized, or redirects away from the site is often treated by bots as a full block, even if the rules inside would otherwise allow crawling.

How to fix: Serve robots.txt as a plain 200 response, under 500 KiB, on the same domain.

No firewall block on AI bots

access.waf_blocks_bot

Severity
critical
Effort
S
Weight
1
Who can fix
site admin

A WAF or CDN can block a bot's user agent before your server or CMS ever sees the request — robots.txt alone cannot fix this, and the bot cannot read your site regardless of your robots.txt policy.

How to fix: Allow-list the affected bot's user agent at your firewall/CDN.

No AI-blocking meta robots tags

access.meta_robots

Severity
high
Effort
S
Weight
1
Who can fix
content editor

A noindex, nosnippet, noai, or noimageai directive stops AI systems from reading, quoting, or indexing the page, even if robots.txt allows crawling.

How to fix: Remove the blocking directive from the meta robots tag unless it is intentional.

No AI-blocking X-Robots-Tag header

access.x_robots_tag

Severity
high
Effort
S
Weight
1
Who can fix
developer

An X-Robots-Tag header has the same blocking effect as a meta tag but is set at the server level, so it is easy to leave in place by accident after a staging-to-production move.

How to fix: Remove the blocking X-Robots-Tag header from the affected responses.

Crawled pages return healthy status codes

access.http_errors

Severity
high
Effort
M
Weight
1
Who can fix
developer

A page that errors or silently shows a 'not found' message while returning 200 wastes crawl budget and cannot be cited, since there is no real content to read.

How to fix: Fix the underlying error, or return a real 404/410 instead of a 200 'not found' page.

Clean redirects

access.redirect_chains

Severity
medium
Effort
S
Weight
1
Who can fix
developer

Inconsistent http/https or www/non-www redirects add extra hops and can split how AI crawlers canonicalize the same page across multiple URLs.

How to fix: Redirect straight to the final https + canonical host in a single hop.

HTTPS is configured correctly

access.https

Severity
medium
Effort
S
Weight
1
Who can fix
site admin

A site without working HTTPS, or without HSTS, is less trusted by crawlers and browsers alike and risks serving mixed/insecure content.

How to fix: Serve the site over HTTPS with a valid certificate and a Strict-Transport-Security header.

No rate limiting during a polite crawl

access.rate_limited

Severity
medium
Effort
M
Weight
1
Who can fix
site admin

If a single polite request per second triggers rate limiting, well-behaved AI bots are also likely to be throttled or blocked during normal crawling.

How to fix: Raise the rate limit (or allow-list known bot user agents) so a slow, polite crawl is never throttled.

Rendering

Server-rendered text

render.ssr_text

Severity
high
Effort
L
Weight
1
Who can fix
developer

AI crawlers typically do not execute JavaScript, so a page that only renders content client-side looks blank to them and cannot be read or cited.

How to fix: Make sure the key headings and paragraphs are present in the initial server response, not only after client-side rendering.

Main content is identifiable

render.main_content

Severity
medium
Effort
M
Weight
1
Who can fix
developer

Without a clear main-content landmark, AI crawlers struggle to separate the page's real content from repeated navigation and footer boilerplate.

How to fix: Wrap the primary page content in a <main> or <article> element.

Main content is visible HTML text

render.hidden_content

Severity
low
Effort
M
Weight
1
Who can fix
developer

If the main content only lives in a JSON hydration blob, a long data-* attribute, or an iframe, a bot reading raw HTML text sees none of it.

How to fix: Server-render the main content as visible HTML text, not only inside a JSON blob or iframe.

Discovery

Sitemap

discovery.sitemap

Severity
medium
Effort
S
Weight
1
Who can fix
site admin

A valid sitemap is the fastest way for crawlers to discover every page on the site, and a fresh lastmod tells them what to re-check first.

How to fix: Publish a valid /sitemap.xml listing your pages, and reference it from robots.txt.

Sitemap accuracy

discovery.sitemap_quality

Severity
medium
Effort
S
Weight
0.6
Who can fix
site admin

A sitemap listing dead URLs or copy-pasted lastmod dates wastes crawl budget and makes it harder for crawlers to trust the freshness signal.

How to fix: Remove or fix sitemap entries that 404/redirect, and keep lastmod dates accurate per URL.

llms.txt

discovery.llms_txt

Severity
low
Effort
S
Weight
1
Who can fix
content editor

llms.txt gives AI agents a quick, structured summary of your site and key pages; it is a minor signal but costs little to get right.

How to fix: Publish /llms.txt following the llmstxt.org format: an H1 title, a one-line summary, and link sections.

Canonical tags

discovery.canonical

Severity
high
Effort
S
Weight
1
Who can fix
developer

A missing or conflicting canonical tag leaves AI crawlers unsure which URL is authoritative, which can split or dilute citations across duplicate URLs.

How to fix: Add exactly one canonical tag per page, pointing to the preferred URL on your own domain.

Unique titles and descriptions

discovery.duplicates

Severity
medium
Effort
S
Weight
0.6
Who can fix
content editor

Duplicate or missing titles and descriptions make it harder for AI systems to tell pages apart and to summarise each one accurately.

How to fix: Give every page a unique <title> and meta description that summarise that specific page.

hreflang alternates

discovery.hreflang

Severity
low
Effort
S
Weight
1
Who can fix
developer

A one-way or malformed hreflang tag confuses search and AI engines about which localized URL to show, which can surface the wrong language version or none at all.

How to fix: Make every hreflang alternate reciprocal and use a valid ISO 639-1 (optionally -REGION) code or x-default.

Structured data

Organization schema markup

schema.organization

Severity
high
Effort
S
Weight
1
Who can fix
developer

Organization or LocalBusiness JSON-LD tells AI systems who you are, so answers that mention your business can attribute it correctly.

How to fix: Add an Organization or LocalBusiness JSON-LD block to the homepage with name, url, and logo.

Valid structured data

schema.validity

Severity
medium
Effort
S
Weight
1
Who can fix
developer

JSON-LD that fails to parse or is missing required fields is ignored by AI systems, so the markup provides no benefit even if it is present.

How to fix: Fix malformed JSON-LD and add any missing required fields reported for each schema type.

Schema type matches page content

schema.page_types

Severity
medium
Effort
S
Weight
1
Who can fix
developer

Service, Product, and FAQPage markup help AI systems match the right content to the right type of question, improving citation accuracy.

How to fix: Add the schema type that matches the page's URL path or content (Service, Product, or FAQPage).

Schema matches on-page contact details

schema.consistency

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

When schema says one phone number/address and the page shows another, AI systems can't tell which one to trust.

How to fix: Update whichever source (schema or visible page text) is out of date so both match exactly.

Structured data found

schema.no_markup

Severity
high
Effort
M
Weight
1
Who can fix
developer

Without any JSON-LD, microdata or RDFa, search and AI engines have to guess at what your page is about instead of reading it directly from markup.

How to fix: Add JSON-LD structured data for your page's main entity (Organization, Product, Article, ...).

No deprecated schema.org terms

schema.deprecated_term

Severity
low
Effort
S
Weight
0.2
Who can fix
developer

A deprecated type or property still parses, but schema.org recommends a replacement — using the old term risks it being dropped from future validators/consumers.

How to fix: Replace the deprecated term with the replacement schema.org names in the issue's evidence.

No pending schema.org terms

schema.pending_term

Severity
info
Effort
S
Weight
0
Who can fix
developer

A 'pending' term is still an extension proposal — most consumers already support it, but it is not guaranteed to stay unchanged.

How to fix: No action required — informational only. Keep an eye on schema.org's pending-term status if you rely on this property.

@id references resolve

schema.id_unresolved

Severity
low
Effort
S
Weight
0.1
Who can fix
developer

When one JSON-LD block references another node by @id (e.g. a Product's brand pointing at '#organization'), that node should exist somewhere on the same page — otherwise the reference is dead.

How to fix: Add the missing node (often by combining blocks into one @graph), or point the @id at the node that does exist.

One consistent Organization/LocalBusiness

schema.duplicate_entity

Severity
medium
Effort
S
Weight
0.2
Who can fix
developer

More than one Organization/LocalBusiness block with a different name or url makes it unclear to AI engines which one is really you.

How to fix: Keep one Organization/LocalBusiness block per page (use @graph with @id to link it from other blocks instead of repeating it).

RDFa stays within the Lite subset

schema.rdfa_unparsed

Severity
info
Effort
S
Weight
0
Who can fix
developer

This validator only reads the RDFa Lite attributes (vocab/typeof/property/resource/prefix); full RDFa features (rel, rev, datatype, inlist, about-chaining) are not parsed, so any schema.org data expressed that way is not checked here.

How to fix: Prefer JSON-LD for new markup — it is easier to validate and is what most AI engines read first. If you keep RDFa, stick to the Lite subset.

Content

Content is structured for AI citation

content.citability

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

AI answer engines favor passages that open with a direct answer, include concrete numbers, and are organized under clear headings, so unstructured content is less likely to be quoted.

How to fix: Add clear headings, an FAQ section, and specific numbers, and rewrite weak passages to lead with a direct answer.

Service/product pages have enough content

content.thin_pages

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

Pages with very little text give AI systems little to work with, so they are less likely to be summarized or cited accurately.

How to fix: Expand thin service/product pages to at least a few descriptive paragraphs.

Heading structure is clean

content.headings

Severity
low
Effort
S
Weight
1
Who can fix
content editor

A missing, duplicated, or out-of-order heading structure makes it harder for AI systems to understand a page's outline and extract the right section.

How to fix: Use exactly one H1 per page and keep heading levels in order (H2 before H3).

About, Contact, and business details are present

content.trust_pages

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

About and Contact pages, plus visible phone/email/address details, help AI systems confirm a business is real and verify who they are citing.

How to fix: Link to an About page and a Contact page from the homepage, and publish a phone number, email, or address.

Passages open with a direct answer

content.answer_first

Severity
high
Effort
M
Weight
1
Who can fix
content editor

AI engines quote passages that answer the implied question in the first sentence — a filler opener gets skipped.

How to fix: Rewrite the first sentence of each flagged passage to directly answer its heading's question.

Target question is answered on the page

content.target_question

Severity
high
Effort
M
Weight
1
Who can fix
content editor

If no passage covers most of the target question's terms, AI engines have nothing to quote when a buyer asks it.

How to fix: Add or rewrite a passage that directly answers the target question using its own key terms.

Claims include concrete numbers

content.statistics

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

A vague claim ('fast', 'many customers') is less citable than a concrete number ('ships in 2 days', '4,000 customers').

How to fix: Add a specific number (price, count, percentage, or timeframe) to each flagged passage.

Numbers are backed by a source

content.sources

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

A number with no source is harder for an AI engine to trust and cite confidently.

How to fix: Name or link the source for each number that currently has none.

Passages are a citable length

content.passage_length

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

AI engines struggle to quote a passage under ~40 words (no context) or over ~300 words (too much to extract).

How to fix: Split long passages, or add context to very short ones, to land in the 40-300 word range.

Headings match how customers ask

content.question_headings

Severity
low
Effort
S
Weight
1
Who can fix
content editor

A heading phrased the way a buyer would ask it ('How much does X cost?') is far more likely to be matched to that question by an AI engine.

How to fix: Rewrite flagged headings as the question or topic a customer would actually type.

Comparisons use lists or tables

content.lists_tables

Severity
low
Effort
S
Weight
1
Who can fix
content editor

AI engines extract structured comparisons (lists, tables) far more reliably than the same facts buried in prose.

How to fix: Convert the flagged comparison passage into a list or table.

Brand/product named explicitly

content.entity_explicit

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

A passage quoted out of context needs the brand/product name in it — 'we' alone doesn't tell an AI answer who 'we' is.

How to fix: Replace a 'we/our' reference in each flagged passage with the actual brand or product name.

Content is readable

content.readability

Severity
low
Effort
S
Weight
1
Who can fix
content editor

Very dense, long-sentence writing is harder for both readers and AI engines to parse and summarize correctly.

How to fix: Shorten sentences and simplify wording until the Flesch score is at least 30.

No unexplained jargon

content.jargon

Severity
low
Effort
S
Weight
1
Who can fix
content editor

Internal jargon ('leverage synergies') adds no citable information and can make a passage sound vague rather than concrete.

How to fix: Replace or explain each flagged jargon term in plain language.

Content has a visible update date

content.freshness

Severity
low
Effort
S
Weight
1
Who can fix
content editor

AI engines favor content that looks current — a missing or stale update date makes a page look unmaintained.

How to fix: Add a dateModified (schema and/or visible 'Updated on' text) and keep it current.

Speed

Fast server response time

speed.ttfb

Severity
medium
Effort
M
Weight
1
Who can fix
developer

A slow time-to-first-byte reduces how much of a site AI crawlers can fetch within their time budget and can signal poor server health.

How to fix: Reduce homepage time-to-first-byte to 800 ms or less with caching and a CDN.

HTML response size is reasonable

speed.html_weight

Severity
low
Effort
S
Weight
1
Who can fix
developer

Very large or uncompressed HTML responses slow down crawling and increase the chance an AI crawler times out before reading the full page.

How to fix: Keep HTML responses under 1 MiB and enable gzip/br compression for responses over 50 KB.

Real-user Largest Contentful Paint is fast

speed.cwv_field

Severity
low
Effort
M
Weight
1
Who can fix
developer

Field data (real visitors, via Chrome UX Report) measures what people actually experience, unlike a single lab test.

How to fix: Improve LCP: optimize the largest above-the-fold image/text block, reduce render-blocking resources.

Prompt coverage

Priority buyer questions have an answering page

coverage.prompt_unanswered

Severity
high
Effort
M
Weight
1
Who can fix
content editor

These are the buyer questions closest to a purchase decision — if no crawled page or passage covers one, an AI engine has nothing of yours to cite when a buyer asks it.

How to fix: Add or expand a page that directly answers the flagged question, using its own key terms.

Every buyer stage has at least one answering page

coverage.stage_gap

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

Buyers ask AI different kinds of questions at each stage of a purchase — a stage with zero covered prompts means the site has nothing for AI to cite at that point in the journey.

How to fix: Add at least one page or section written for the missing buyer stage.

Pricing or quote information is on the site

coverage.pricing_info

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

Price is one of the most common buyer questions AI engines answer — a site that never mentions price or a quote process leaves AI engines to guess or stay silent.

How to fix: Add a pricing range, starting price, or a clear "how to get a quote" section.

Location-specific content matches the site's stated location

coverage.location

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

Buyers asking AI for a local or regional supplier expect location-specific proof (service area, shipping, local reviews) — without it, AI engines can't confirm the site serves that area.

How to fix: Add a page or section naming the service area, shipping destinations, or local presence.

The site has citable proof (case studies, certifications, reviews)

coverage.proof

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

AI engines favor answers backed by verifiable proof — a site with no case study, certification, or review content gives an AI engine nothing concrete to cite when a buyer asks "is this reliable".

How to fix: Add a case study, certification detail, or genuine review/testimonial section.

Early-stage buyer questions are answered somewhere on the site

coverage.faq

Severity
low
Effort
S
Weight
1
Who can fix
content editor

Problem/awareness-stage questions ('why', 'what is') are often the first thing a buyer asks AI — a site with nothing answering them misses an easy citation opportunity.

How to fix: Add an FAQ section or intro paragraph answering the flagged early-stage questions.

Fan-out coverage

Rule-based sub-queries are covered by the page

fanout.subquery_uncovered

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

These sub-queries are Leadee's simulation of what an engine might search while researching an answer to the prompt — gaps here are ideas for content to add, not proof of what an engine actually searched.

How to fix: Add a passage addressing each uncovered sub-query, in its own words.

Buyer attributes (price, lead time, certification, MOQ, warranty) are named

fanout.attribute_missing

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

Buyer attribute questions (price, lead time, certification, MOQ, warranty) are common sub-queries an engine's research might cover — a page silent on one gives an AI engine nothing to quote for it.

How to fix: Name the missing attribute explicitly on the page (even a range or "contact us" is better than silence).

The page compares itself to an alternative

fanout.comparison_missing

Severity
low
Effort
S
Weight
1
Who can fix
content editor

Comparison sub-queries ('X vs Y', 'alternatives to X') are common while an AI engine researches an answer — a page with no comparison content gives it nothing of yours to weigh against the alternative.

How to fix: Add a short, honest comparison section against a named alternative.

Technical SEO

Title and meta description length

seo.title_meta

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

A missing or oversized title/description gets truncated or rewritten in search results and AI summaries, so your own wording never reaches the reader.

How to fix: Write a unique title (<= 60 characters) and meta description (70-160 characters) for every page.

Mobile viewport tag

seo.mobile_viewport

Severity
medium
Effort
S
Weight
1
Who can fix
developer

Without a viewport tag, mobile browsers render the page at desktop width, so most visitors — and mobile-first indexing — see a broken, zoomed-out layout.

How to fix: Add a responsive viewport meta tag to the page template.

Image alt text

seo.images_alt

Severity
low
Effort
S
Weight
1
Who can fix
content editor

An image with no alt attribute is invisible to screen readers and to AI engines that read page text — any information it carries (a product, a diagram, a quote) is lost.

How to fix: Add a short, descriptive alt attribute to every content image; use alt="" only for purely decorative images.

Indexability conflicts

seo.indexability_conflicts

Severity
high
Effort
S
Weight
1
Who can fix
developer

Listing a noindex page in the sitemap, or canonicalizing to one, sends search and AI engines a contradictory signal about which URL you actually want indexed — wasting crawl budget on a page that will never be shown.

How to fix: Remove noindex pages from the sitemap, and never point a canonical tag at a noindex page.

Links through a redirect

seo.redirect_targets

Severity
low
Effort
S
Weight
1
Who can fix
content editor

Every redirect hop adds latency and a small chance of being dropped by a crawler; linking straight to the final URL is free to fix and keeps internal link equity from leaking through the hop.

How to fix: Update internal links to point directly at the final URL instead of a redirecting one.

Email security

MX record present

email.mx

Severity
high
Effort
S
Weight
1
Who can fix
site admin

Without an MX record, mail servers have nowhere to deliver messages addressed to this domain at all.

How to fix: Add an MX record pointing at your mail provider.

SPF is configured correctly

email.spf

Severity
high
Effort
S
Weight
1
Who can fix
site admin

SPF tells receiving mail servers which servers are allowed to send mail for this domain; missing it (or ending in +all) makes it easy for spammers to forge mail that appears to come from this domain.

How to fix: Publish a single SPF TXT record ending in -all or ~all, with at most 10 DNS lookups.

DMARC is enforcing with reporting

email.dmarc

Severity
medium
Effort
S
Weight
1
Who can fix
site admin

DMARC tells receiving servers what to do with mail that fails SPF/DKIM; p=none only monitors (does not stop spoofed mail reaching inboxes), and without rua no one sees the reports.

How to fix: Publish a DMARC TXT record with p=quarantine or p=reject and a rua reporting address.

DKIM signing key found

email.dkim

Severity
medium
Effort
S
Weight
1
Who can fix
site admin

DKIM lets receiving servers verify that a message was not altered in transit and really came from an authorized sender; without it, SPF/DMARC alone give a weaker signal.

How to fix: Enable DKIM signing with your mail provider and publish its selector's TXT record.

MTA-STS is configured

email.mta_sts

Severity
low
Effort
M
Weight
1
Who can fix
developer

MTA-STS tells sending mail servers to require a valid, encrypted connection when delivering mail to this domain, preventing a network attacker from downgrading delivery to plaintext.

How to fix: Publish the _mta-sts TXT record and a valid policy file at /.well-known/mta-sts.txt.

TLS-RPT is configured

email.tls_rpt

Severity
low
Effort
S
Weight
1
Who can fix
developer

TLS-RPT reports back when another mail server fails to make an encrypted connection to this domain, which is the main way to notice an MTA-STS misconfiguration before mail starts bouncing.

How to fix: Publish a _smtp._tls TXT record with a rua reporting address.

BIMI logo record found

email.bimi

Severity
info
Effort
S
Weight
0
Who can fix
site admin

BIMI lets a supporting mail client show your verified logo next to your mail in the inbox, a minor trust signal that costs little to add once DMARC is enforcing.

How to fix: Publish a default._bimi TXT record pointing at a square SVG logo.

Social profiles

Social profiles discovered

social.discovered

Severity
info
Effort
S
Weight
0
Who can fix
content editor

AI engines and buyers cross-check a brand's social presence; profiles linked from the site's own schema, header, footer or rel="me" links are the only ones Leadee can verify without logging in or guessing a URL.

How to fix: Link your company's social profiles from the site (footer, header, or Organization JSON-LD sameAs).

Social profiles are live

social.broken

Severity
high
Effort
S
Weight
1
Who can fix
content editor

A profile that 404s or redirects to the network's homepage or a login wall reads as abandoned or fake to both buyers and AI engines.

How to fix: Fix or remove the broken social profile link.

Social profile names match the brand

social.name_mismatch

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

A profile title that doesn't mention the brand name makes it harder for AI engines to confirm the profile and the website describe the same company.

How to fix: Rename the profile to match the brand name used on the website.

Social links are listed in schema

social.not_in_schema

Severity
medium
Effort
S
Weight
1
Who can fix
developer

AI engines and search crawlers primarily read structured data, not footer HTML — a profile linked only in the footer is far less likely to be picked up as part of the brand's entity.

How to fix: Add the profile URL to the Organization JSON-LD sameAs array.

Recommended social networks are present

social.missing_network

Severity
low
Effort
S
Weight
1
Who can fix
content editor

Buyers researching a company expect to find it on the network most relevant to its industry (e.g. LinkedIn for B2B) — not having one is a missed trust signal, not a hard error.

How to fix: Create and link a profile on the missing recommended network.

shop

Store platform detected

shop.platform

Severity
info
Effort
S
Weight
0
Who can fix
site admin

Recognising the store platform lets AI shopping agents use the right integration path; this check is informational and carries no score weight.

How to fix: No action required.

Product schema markup

shop.product_schema

Severity
high
Effort
M
Weight
1
Who can fix
developer

Without complete Product JSON-LD, AI shopping agents cannot reliably read price, availability, or shipping details, blocking agentic purchase flows.

How to fix: Add Product JSON-LD with name, image, description, SKU/GTIN, brand, price, currency, availability, shipping, and return policy.

Machine-readable catalog API

shop.catalog_api

Severity
high
Effort
M
Weight
1
Who can fix
developer

AI shopping agents browse product catalogs programmatically; an unreachable catalog endpoint blocks that path entirely.

How to fix: Enable the platform's public catalog endpoint and ensure it is reachable without authentication.

Product feed signals

shop.feed

Severity
low
Effort
M
Weight
1
Who can fix
site admin

Without a product feed, AI shopping assistants and comparison engines cannot list your products alongside competitors.

How to fix: Install a product feed plugin/app and submit it to Google Merchant Center.

Shipping/returns/terms pages linked

shop.policies

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

AI shopping agents need shipping and return policies visible to confirm purchase conditions before recommending a store.

How to fix: Create and link Shipping, Returns, and Terms pages from the site footer.

Agentic commerce protocol endpoints

shop.protocols

Severity
low
Effort
M
Weight
0.3
Who can fix
developer

Emerging AI buying agents are expected to use standardised endpoints (UCP/ACP/MCP); early adoption is a competitive advantage, not yet a ratified requirement.

How to fix: Monitor UCP/ACP/MCP specifications and add well-known endpoints as standards mature.

Guest checkout

shop.guest_checkout

Severity
info
Effort
S
Weight
0
Who can fix
site admin

Guest checkout availability affects AI agent purchase flows; not checked in this version of the engine.

How to fix: No action required in this version.

Online product designer

print.designer

Severity
high
Effort
M
Weight
1
Who can fix
site admin

Print shops without an online designer lose customers who expect to customise products (business cards, banners, etc.) before ordering.

How to fix: Integrate an online design tool so customers can personalise products before checkout.

Instant pricing/quote

print.instant_quote

Severity
medium
Effort
M
Weight
1
Who can fix
developer

Without instant pricing, customers and AI agents must contact the business manually to get a quote, slowing the sales cycle.

How to fix: Add a visible price calculator or quantity-tiered pricing table to every product page.

Artwork file upload

print.upload

Severity
medium
Effort
M
Weight
1
Who can fix
developer

Without a file upload option, customers must email artwork separately, slowing order fulfilment.

How to fix: Add a clearly labelled file upload field accepting common design formats.

Bulk/tiered pricing

print.bulk_pricing

Severity
low
Effort
S
Weight
1
Who can fix
developer

Missing quantity discounts make it harder to win bulk orders against competitors who show clear tiered pricing.

How to fix: Display a tiered price table and encode it in Product JSON-LD priceSpecification.

entity

Organization/LocalBusiness entity present

entity.org_block

Severity
high
Effort
S
Weight
1
Who can fix
developer

AI systems need exactly one clear Organization or LocalBusiness block to know who your business is — without it, answers can't attribute anything to your brand.

How to fix: Add one Organization or LocalBusiness JSON-LD block to the homepage.

Business name is consistent

entity.name_consistency

Severity
high
Effort
S
Weight
1
Who can fix
content editor

AI engines lose confidence in an entity when its name differs between <title>, og:site_name and schema markup — pick one name and use it everywhere.

How to fix: Use the exact same business name in schema, <title> and og:site_name.

Open Graph tags present

entity.og_tags

Severity
low
Effort
S
Weight
1
Who can fix
developer

og:title, og:image and og:type help AI systems and social previews describe your page correctly.

How to fix: Add og:title, og:image and og:type meta tags to the page head.

About page linked and descriptive

entity.about_linked

Severity
low
Effort
M
Weight
1
Who can fix
content editor

A linked About page with a real company description (80+ words) gives AI systems the context to describe who you are.

How to fix: Link an About page from the homepage and write at least 80 words describing the company.

sameAs profiles declared

entity.same_as_present

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

sameAs links tell AI systems which social/business profiles belong to this entity, strengthening identity confidence.

How to fix: Add a sameAs array to the Organization schema with links to your official profiles.

sameAs profiles are alive

entity.same_as_live

Severity
high
Effort
S
Weight
1
Who can fix
content editor

A sameAs link that 404s or redirects to a platform's homepage signals a stale or fake profile to AI systems.

How to fix: Remove or fix any sameAs URL that is dead or redirects away from the actual profile.

sameAs profiles match the brand

entity.same_as_match

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

A profile page that doesn't mention your brand name in its title/heading makes AI systems less confident it's really yours.

How to fix: Confirm each sameAs profile's own page shows the brand name clearly.

Wikidata entry found

entity.wikidata

Severity
low
Effort
L
Weight
1
Who can fix
expert

A Wikidata entry with a matching official website (P856) is a strong, independent identity signal AI knowledge graphs use.

How to fix: If your business is notable enough per Wikidata's policy, submit an entry with the official website property set.

Contact details present in schema

entity.contact_point

Severity
medium
Effort
S
Weight
1
Who can fix
developer

Machine-readable contact details let AI systems answer 'how do I contact this business' directly from schema.

How to fix: Add contactPoint, email or telephone to the Organization schema.

Name/address/phone consistent (NAP)

entity.nap

Severity
medium
Effort
M
Weight
1
Who can fix
content editor

Inconsistent name, address or phone between schema and the visible page confuses AI systems and local search alike.

How to fix: Make sure schema NAP fields match exactly what's shown on the homepage, About and Contact pages.

sec

HTTP redirects to HTTPS

sec.https_redirect

Severity
high
Effort
S
Weight
1
Who can fix
site admin

A visitor who types the plain http:// address (or follows an old http:// link) stays on an unencrypted connection instead of being bounced to the secure one, exposing anything they submit to anyone on the network path.

How to fix: Redirect every http:// request to the https:// version of the same URL.

TLS certificate is valid

sec.tls_certificate

Severity
high
Effort
S
Weight
1
Who can fix
site admin

An expired, hostname-mismatched, or untrusted certificate makes every browser show a warning page instead of the site, and breaks any integration that validates the chain.

How to fix: Renew or reissue the certificate so it is valid, matches the hostname, and chains to a trusted root.

No legacy TLS versions accepted

sec.tls_protocol

Severity
medium
Effort
S
Weight
1
Who can fix
developer

TLS 1.0 and 1.1 have known cryptographic weaknesses and have been deprecated by every major browser and payment-card standard; still accepting them signals an outdated TLS configuration.

How to fix: Disable TLS 1.0 and 1.1, accepting only TLS 1.2 and newer.

HSTS is configured

sec.hsts

Severity
medium
Effort
S
Weight
1
Who can fix
developer

Without a long-lived HSTS header, a browser that has only ever seen the http:// address can still be tricked onto an insecure connection by a network attacker, even though HTTPS is available.

How to fix: Add a Strict-Transport-Security header with max-age of at least 6 months.

Security response headers are set

sec.headers

Severity
medium
Effort
S
Weight
1
Who can fix
developer

Content-Security-Policy, X-Content-Type-Options, X-Frame-Options (or frame-ancestors), Referrer-Policy and Permissions-Policy each close off a different class of browser-side attack (script injection, MIME sniffing, clickjacking, referrer leakage, unwanted device API access).

How to fix: Add the missing security headers listed in the evidence below.

Cookies are configured securely

sec.cookies

Severity
low
Effort
S
Weight
1
Who can fix
developer

A cookie without Secure can be sent over plain HTTP, without HttpOnly it is readable by any script on the page (including an injected one), and without SameSite it can be attached to cross-site requests.

How to fix: Set Secure, HttpOnly and SameSite on every cookie the homepage sets.

No mixed content

sec.mixed_content

Severity
medium
Effort
S
Weight
1
Who can fix
developer

An http:// resource loaded from an https:// page can be tampered with in transit, and most browsers block or warn on it, breaking the resource for visitors.

How to fix: Change every http:// resource reference on the page to https://.

No software version disclosed

sec.version_disclosure

Severity
low
Effort
S
Weight
1
Who can fix
developer

A Server/X-Powered-By header or meta generator tag naming an exact version (e.g. 'Apache/2.4.41') tells an attacker exactly which known vulnerabilities to try, with no need to guess.

How to fix: Remove or generalise the version string in the listed header or tag.

No sensitive files exposed

sec.exposed_files

Severity
critical
Effort
S
Weight
1
Who can fix
developer

A reachable .git directory, .env file, or database/backup file can hand an attacker credentials, source code, or a full site backup with no further effort.

How to fix: Remove the exposed file from the public web root, or block access to it at the server level.

security.txt is published

sec.security_txt

Severity
info
Effort
S
Weight
0
Who can fix
site admin

RFC 9116's security.txt gives security researchers a clear, standard way to report a vulnerability responsibly instead of guessing at a contact address — this is informational and carries no score weight.

How to fix: Publish /.well-known/security.txt with a Contact field.

llms

llms.txt content-type

llms.content_type

Severity
low
Effort
S
Weight
1
Who can fix
developer

AI agents expect /llms.txt to be plain text or markdown — a wrong content-type can make some clients skip or mis-render it.

How to fix: Serve /llms.txt with Content-Type: text/plain or text/markdown.

Key pages are listed in llms.txt

llms.key_pages_missing

Severity
medium
Effort
S
Weight
1
Who can fix
content editor

llms.txt is only useful if it points an AI agent at your most important pages — missing them defeats the purpose of publishing the file.

How to fix: Add the missing key pages to the relevant llms.txt section.

llms.txt stays small

llms.too_large

Severity
low
Effort
S
Weight
1
Who can fix
content editor

llms.txt is meant to be a short index, not the full content — a large file burns an AI agent's context budget before it reaches the pages it links to.

How to fix: Move long content to llms-full.txt and keep llms.txt as a short index of links.

rich

Article rich result eligible

rich.article.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Article rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Article rich result fully enriched

rich.article.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Article item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Breadcrumb rich result eligible

rich.breadcrumb.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Breadcrumb rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Breadcrumb rich result fully enriched

rich.breadcrumb.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Breadcrumb item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Event rich result eligible

rich.event.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Event rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Event rich result fully enriched

rich.event.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Event item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Job posting rich result eligible

rich.job_posting.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Job posting rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Job posting rich result fully enriched

rich.job_posting.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Job posting item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Local business rich result eligible

rich.local_business.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Local business rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Local business rich result fully enriched

rich.local_business.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Local business item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Logo rich result eligible

rich.logo.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Logo rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Logo rich result fully enriched

rich.logo.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Logo item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Product snippet rich result eligible

rich.product_snippet.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Product snippet rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Product snippet rich result fully enriched

rich.product_snippet.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Product snippet item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Merchant listing rich result eligible

rich.merchant_listing.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Merchant listing rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Merchant listing rich result fully enriched

rich.merchant_listing.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Merchant listing item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Review snippet rich result eligible

rich.review_snippet.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Review snippet rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Review snippet rich result fully enriched

rich.review_snippet.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Review snippet item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Software application rich result eligible

rich.software_application.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Software application rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Software application rich result fully enriched

rich.software_application.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Software application item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Video rich result eligible

rich.video_object.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Video rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Video rich result fully enriched

rich.video_object.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Video item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

Recipe rich result eligible

rich.recipe.required_missing

Severity
high
Effort
S
Weight
0.3
Who can fix
content editor

Google only shows the Recipe rich result when every required property for that feature is present — the schema.org markup stays valid either way, but the richer search/AI presentation is lost.

How to fix: Add the required properties listed in the eligibility table for this feature, then re-run the validator.

Recipe rich result fully enriched

rich.recipe.recommended_missing

Severity
low
Effort
S
Weight
0.1
Who can fix
content editor

The Recipe item is eligible, but adding the recommended properties gives Google (and AI engines reading the same markup) more to work with.

How to fix: Add the recommended properties listed in the eligibility table when the data is available.

FAQ markup noted

rich.faq.limited

Severity
info
Effort
S
Weight
0
Who can fix
content editor

Since 2023-09-14 Google only shows the FAQ rich result for well-known, authoritative government and health sites — for everyone else the markup is still valid schema.org and still read by AI engines, it just will not produce the rich result on Google.

How to fix: No fix needed — this is a Google policy limit, not a markup error. Keep the FAQPage markup for AI engines and leave it as-is.

HowTo markup noted

rich.howto.deprecated

Severity
info
Effort
S
Weight
0
Who can fix
content editor

Since 2023-09-14 Google no longer shows the HowTo rich result at all — the markup itself is still valid schema.org and still read by AI engines that look for step-by-step instructions.

How to fix: No fix needed — this is a Google policy change, not a markup error. Keep the HowTo markup for AI engines and leave it as-is.

Reviews are independently sourced

rich.policy.self_serving_review

Severity
medium
Effort
M
Weight
0.1
Who can fix
content editor

Google does not show review/rating stars when the entity being reviewed controls the reviews about itself (your own LocalBusiness/Organization reviewing itself).

How to fix: Move review/aggregateRating markup onto an independent review page you do not control, or remove it from your own Organization/LocalBusiness markup.

readiness

AI visibility checker finished every pillar

readiness.incomplete_scan

Severity
info
Effort
S
Weight
0
Who can fix
content editor

The AI visibility checker stops at a 25-second budget so you always get an answer quickly; a pillar that did not finish in time is left out of the score rather than guessed, so the number you see only reflects pillars that actually ran.

How to fix: Re-run the check — the pillars that timed out usually finish on a second try, and every sub-tool's own 24-hour cache means the parts that already finished are not re-fetched.

robots

Sitemap declared in robots.txt

robots.sitemap_declared

Severity
low
Effort
S
Weight
1
Who can fix
site admin

A Sitemap: line in robots.txt is the first place most crawlers (including AI search bots) look for your sitemap — without it, discovery depends entirely on them finding /sitemap.xml by guessing or being told elsewhere.

How to fix: Add an absolute-URL Sitemap: line to robots.txt.

meta

og:image is a usable preview image

meta.og_image

Severity
high
Effort
S
Weight
0.8
Who can fix
site admin

When someone shares your page on LinkedIn, Facebook, X or WhatsApp, the og:image is the picture in the preview. A broken, tiny or badly shaped image makes the share look unfinished and gets fewer clicks.

How to fix: Point og:image at a full https:// URL of a 1200 x 630 px JPG or PNG that loads for everyone.

og:description and og:url are set

meta.og_details

Severity
low
Effort
S
Weight
0.3
Who can fix
site admin

Without og:description, platforms guess the preview text or leave it blank. An og:url that differs from the canonical splits shares and likes across two URLs.

How to fix: Add og:description and set og:url to the same URL as the canonical tag.

twitter:card is set to a valid type

meta.twitter_card

Severity
low
Effort
S
Weight
0.3
Who can fix
site admin

X reads twitter:card to choose the preview layout. Without it, or with a value X does not know, the link can show as a small card or plain text.

How to fix: Add <meta name="twitter:card" content="summary_large_image">.

Each head tag appears once

meta.duplicate_tags

Severity
medium
Effort
S
Weight
0.5
Who can fix
site admin

When a page has two titles or two og:title tags, each platform picks one by its own rule, so previews and search results can show the wrong text.

How to fix: Keep one of each tag; turn off the duplicate in your theme or second SEO plugin.

sitemap

Sitemap is valid sitemaps.org XML

sitemap.xml_format

Severity
high
Effort
S
Weight
0.8
Who can fix
site admin

Search and AI crawlers read a sitemap with a strict XML parser. A file that does not parse, has the wrong root element or the wrong namespace can be ignored in full, so none of its URLs are found from it.

How to fix: Serve well-formed XML with a <urlset> (or <sitemapindex>) root and xmlns="http://www.sitemaps.org/schemas/sitemap/0.9".

Each sitemap is within 50,000 URLs and 50 MB

sitemap.size_limits

Severity
high
Effort
S
Weight
0.8
Who can fix
site admin

The sitemaps.org protocol caps each file at 50,000 URLs and 50 MB uncompressed. Crawlers may stop reading or reject a file over the limit, so the URLs at the end are missed.

How to fix: Split large sitemaps into several files of at most 50,000 URLs and list them in a sitemap index.

lastmod dates use the W3C datetime format

sitemap.lastmod_format

Severity
medium
Effort
S
Weight
0.4
Who can fix
site admin

Crawlers use <lastmod> to decide which pages to read again first. A date in another format is ignored, and a future date tells them the value cannot be trusted.

How to fix: Write each <lastmod> as YYYY-MM-DD, or as a full time with a time zone such as 2026-11-14T09:30:00+00:00.

Every <loc> is a full URL on the sitemap's own host

sitemap.loc_values

Severity
medium
Effort
S
Weight
0.5
Who can fix
site admin

The protocol asks for full URLs on the same host as the sitemap. Relative URLs cannot be followed, and URLs on another host are dropped unless that host vouches for the sitemap in its robots.txt.

How to fix: Use full https:// URLs on the sitemap's own host, under 2,048 characters, each listed once.

Every child sitemap in the index loads

sitemap.child_sitemaps

Severity
high
Effort
S
Weight
0.7
Who can fix
site admin

A sitemap index only helps if every file it lists returns 200 and is a normal sitemap. A broken child hides all of its URLs from crawlers.

How to fix: Fix or remove child sitemaps that do not return 200, and list only <urlset> files in the index.

robots.txt names the sitemap

sitemap.robots_declared

Severity
medium
Effort
S
Weight
0.4
Who can fix
site admin

A 'Sitemap:' line in robots.txt is how crawlers without a webmaster console, including AI crawlers, find your sitemap. Without it they must guess the URL.

How to fix: Add a line 'Sitemap: https://yoursite.com/sitemap.xml' (the full URL) to robots.txt.

Sampled sitemap URLs do not redirect

sitemap.redirected_urls

Severity
medium
Effort
S
Weight
0.5
Who can fix
site admin

A sitemap should list the final URL of each page. A URL that redirects costs crawlers an extra request and tells them the sitemap is out of date.

How to fix: Replace each redirecting URL in the sitemap with the URL it redirects to.

Sampled sitemap URLs are their own canonical

sitemap.non_canonical_urls

Severity
medium
Effort
S
Weight
0.5
Who can fix
site admin

A page whose canonical tag points to another URL tells search and AI engines not to index it. Listing it in the sitemap sends the opposite signal.

How to fix: List only canonical URLs in the sitemap, or fix the canonical tag if the page should be indexed itself.

redirect

No redirect loop

redirect.loop

Severity
critical
Effort
S
Weight
1
Who can fix
site admin

A loop sends browsers and crawlers round in circles until they give up. Visitors see an error, and search and AI crawlers cannot read the page at all.

How to fix: Find the two rules that send the URL back and forth and remove one of them.

The chain ends within 10 redirects

redirect.too_many_hops

Severity
high
Effort
S
Weight
0.9
Who can fix
site admin

Browsers stop after about 20 redirects and crawlers often stop sooner; Google's docs say Googlebot follows up to 10. A very long chain may never reach the page.

How to fix: Point the first URL straight at the final page and remove the steps in between.

The chain ends on a working page

redirect.final_status

Severity
high
Effort
S
Weight
0.9
Who can fix
site admin

After all the redirects, the last URL must answer 200. If it answers 404, 500 or nothing, visitors and crawlers reach an error instead of your page.

How to fix: Make the last URL in the chain answer 200, or redirect to a page that does.

Permanent moves use 301 or 308

redirect.temporary

Severity
low
Effort
S
Weight
0.4
Who can fix
site admin

A 302 or 307 says the move is temporary, so search engines may keep the old URL in their index. For a permanent move, such as http to https or a new URL, use 301 or 308.

How to fix: Use 301 (or 308) for redirects that are permanent.

No redirect from https to http

redirect.https_downgrade

Severity
high
Effort
S
Weight
0.8
Who can fix
site admin

A hop from https to http drops encryption for that step. Browsers may warn or block it as mixed content, and anyone on the network can change the response.

How to fix: Point every redirect at an https:// address.

AI crawlers get the same redirects as Chrome

redirect.ua_difference

Severity
high
Effort
S
Weight
0.8
Who can fix
site admin

If an AI crawler is sent somewhere else than a browser, the AI system reads a different page than your visitors see. It may cite the wrong URL, or none at all.

How to fix: Send AI crawlers through the same redirects as a browser.

No meta refresh redirect

redirect.meta_refresh

Severity
low
Effort
S
Weight
0.3
Who can fix
site admin

A meta refresh is a redirect inside the HTML instead of an HTTP status code. Crawlers that only read status codes may not follow it, and it adds a page load.

How to fix: Replace the meta refresh with a 301 redirect on the server.

hreflang

Each version links to itself

hreflang.self_reference

Severity
medium
Effort
S
Weight
0.5
Who can fix
site admin

Google asks each language version to list itself as well as every other version. A set without the self link is incomplete.

How to fix: Add a hreflang link to the page itself, with its own language code.

An x-default version is set

hreflang.x_default

Severity
low
Effort
S
Weight
0.3
Who can fix
site admin

x-default tells Google which page to show visitors whose language you do not cover. Without it, Google picks one for them.

How to fix: Add hreflang="x-default" pointing to your fallback page.

Canonical tags agree with hreflang

hreflang.canonical_conflict

Severity
high
Effort
S
Weight
0.8
Who can fix
site admin

hreflang should point at canonical URLs, and each version should be its own canonical. If a version's canonical points elsewhere, Google may drop it and ignore the hreflang set.

How to fix: Make each language version its own canonical and point hreflang at those canonical URLs.

hreflang links point to live, full URLs

hreflang.target_urls

Severity
high
Effort
S
Weight
0.7
Who can fix
site admin

Google needs full URLs that answer 200 and can be indexed. Relative URLs, broken pages, redirects and noindex pages break the set.

How to fix: Use full https:// URLs of live, indexable pages in every hreflang link.

One URL per language code

hreflang.duplicate_codes

Severity
medium
Effort
S
Weight
0.4
Who can fix
site admin

Each language code should name one URL per set. When one code points to two URLs, Google cannot tell which one to show.

How to fix: Keep one URL per language code and remove the extra links.

HTML, header and sitemap agree

hreflang.source_conflict

Severity
medium
Effort
S
Weight
0.4
Who can fix
site admin

Google reads hreflang from the HTML, the Link header and the sitemap as equal signals. When they disagree, the result is hard to predict.

How to fix: Keep hreflang in one place, or make every place list the same set.

Check catalog questions

Where does this catalog come from?

Every check here is generated straight from Leadee's own AI crawlability report API — the same catalog version and definitions your report uses, not a hand-maintained list that can drift out of sync.

What do severity and weight mean?

Severity is how much a failed check hurts AI crawlability (critical down to info); weight is how much that check counts toward your overall Audit score. "Who can fix" shows whether it needs a developer or just a content edit.

How do I fix a failed check?

Each check's "How to fix" summary explains the fix in plain terms; your own Audit report links straight to the matching entry here from the issue card that found it.

Can GEO guarantee results?

No. AI answer engines change their output frequently, and no vendor — including Leadee — can guarantee a ranking or a mention. Leadee publishes confidence intervals on every metric so you can see how much a result can vary run to run.

How does Leadee measure AI visibility?

Leadee runs your prompt set against each AI engine on a weekly to daily schedule, depending on plan, through official vendor APIs (OpenAI, Gemini, Perplexity, Anthropic, DataForSEO) — never by scraping consumer chat interfaces — and reports mention rate, citation rate, average position, share of voice and sentiment with a 95% confidence interval.

Who do I talk to?

Email support@leadee.ai or use Talk to an expert at https://leadee.ai/contact-sales — the Netbase JSC team behind Leadee replies by email or WhatsApp — or WhatsApp the CEO directly at https://wa.me/84937869689.