Why it matters: With no freshness signals, AI assumes your content is stale. ChatGPT and Perplexity both weight recency in their citation algorithms. Competitors who show recent updates will be cited instead of your static pages.
How to fix: Add dateModified and datePublished to your JSON-LD schema, and display a visible 'Last updated' date on the page. Update content quarterly at minimum. AI heavily favors fresh, maintained content.
Only 1 distinct schema type (BusinessEntity) — insufficient for AI visibility
Why it matters: Single-schema sites get minimal AI coverage. AI models need multiple data signals to confidently recommend your store.
How to fix: One schema type is not enough. Add at minimum: Organization (identity), Product (commerce), BreadcrumbList (navigation), WebSite+SearchAction (discoverability), FAQPage (Q&A answering).
Organization schema found but no sameAs links
Why it matters: sameAs links tell AI that this is the same entity across the web. It's how Google builds Knowledge Panels and how AI attributes information to your brand.
How to fix: Add sameAs array to your Organization schema linking to Wikipedia, LinkedIn, Facebook, Instagram, and other official profiles. This creates a Knowledge Graph entity for your brand.
Why it matters: ChatGPT, Perplexity, and Google AI Overviews pull FAQ answers verbatim. FAQ schema is one of the fastest ways to get your content cited by AI.
How to fix: Add FAQPage schema to every product page and category page. Include 3-5 Q&As per page covering: product specifications, shipping, returns, usage instructions. Format: question (full sentence) + answer (75-150 words).
Why it matters: AI assistants use breadcrumbs to understand product categorization and site structure. Without it, AI can't contextualize where products fit in your catalog — e.g., 'Running Shoes' under 'Sports > Footwear > Running'.
How to fix: Add BreadcrumbList schema reflecting your category hierarchy: Home → Category → Subcategory → Product. Each item needs name and URL.
Why it matters: SearchAction enables Google's sitelinks search box and allows AI shopping assistants to search your catalog directly. It's how AI agents find specific products in your store.
How to fix: Add WebSite schema with potentialAction: SearchAction. Define your search URL template so AI assistants and Google can search your store programmatically.
Content doesn't start with a strong summary paragraph
Why it matters: AI assistants extract content from the first 100 words to generate summaries. Most AI systems read top-down, so answer-first pages get cited more often.
How to fix: Place your most important information in the first 100 words of the page. Use the BLUF method (Bottom Line Up Front): start with what the product IS and why it matters, then elaborate.
14 lists found — consider adding comparison tables
Why it matters: AI models are biased toward extracting data from HTML lists and tables. Perplexity and ChatGPT pull bullet points and table data with higher accuracy than paragraph text.
How to fix: Add comparison tables to your content. Use <ul>/<ol> for feature lists, specifications, and benefits. Use <table> for product comparisons, pricing tiers, and specifications. AI extracts structured content much faster than paragraphs.
OG tags complete but no sameAs entity links
Why it matters: AI needs to confidently identify your brand entity. Open Graph and sameAs together create a cross-platform identity that AI models recognize.
How to fix: For full entity clarity: 1) Complete Open Graph tags (og:title, og:description, og:image, og:type), 2) Add sameAs in Organization schema linking to Wikipedia, LinkedIn, Facebook, and other profiles.
Why it matters: AI models weight author expertise heavily. Pages from identified experts get cited more than anonymous content. This is especially important for product guides, reviews, and advice content.
How to fix: For content pages (blog, guides, about): add Article/BlogPosting schema with author property linking to Person schema. Include the author's jobTitle, credentials, and social profiles.
Why it matters: Product feeds power Google Shopping, Bing Shopping, and increasingly AI commerce. Without a structured product feed, AI agents can't efficiently index your full catalog for product recommendations.
How to fix: Create a Google Merchant Center / product feed (XML or CSV). Expose it at a consistent URL and reference it in your sitemap. AI shopping assistants and comparison engines use product feeds for catalog discovery. Tip: validate your Heureka / Zboží / Glami / Merchant feeds for free with Mergado's audit.
Why it matters: Heureka.sk and Heureka.cz are the dominant price comparison engines on the SK/CZ market. Without an XML feed you're not visible on the main marketplace. No generic audit tool checks this.
How to fix: Generate a Heureka XML feed at /feed/heureka.xml (or /export/heureka.xml). PrestaShop has a Heureka.cz module, WooCommerce has plugins. Structure: <SHOP><SHOPITEM>...</SHOPITEM></SHOP> with ITEM_ID/PRODUCTNAME/URL/PRICE_VAT/CATEGORYTEXT (required) + EAN/PARAM/DELIVERY_DATE (highly recommended for ranking).
4 paragraphs, avg 24 words — too short for citation
Why it matters: AI extracts individual paragraphs as answer snippets — dense walls of text get skipped. Focused 40-80 word paragraphs are the most citable.
How to fix: Optimal paragraphs for AI citation are 40-80 words. Break long paragraphs into focused, self-contained answer blocks. Each should make one clear point that AI can extract and quote.
No statistical data found in content
Why it matters: Content without data is harder to cite. Perplexity and ChatGPT specifically seek pages with quantified claims and verifiable statistics.
How to fix: Add numbers: pricing comparisons, performance metrics, customer statistics, industry benchmarks. Specific data (e.g., '99.9% uptime', '4.8/5 rating from 2,400 reviews') is what AI quotes most.
Avg section: 122 words — only 0% in 80-200 word optimal range
Why it matters: For Google AI Overviews, 100-150 words per section is the sweet spot — long sections get skipped and very short ones lack substance.
How to fix: Restructure content into sections of 120-180 words between H2/H3 headings. Each section should cover one topic completely. Split sections over 300 words, expand sections under 80 words.
Why it matters: Q&A content format matches how people query AI assistants. Without question-format headings, your content is harder for AI to map to user queries.
How to fix: Add H2/H3 headings phrased as questions your customers ask: 'How much does shipping cost?', 'What sizes are available?', 'How do I return an item?' Follow each with a direct, concise answer.
Canonical points to different URL: https://www.sperky.sk/
Why it matters: AI search engines (ChatGPT, Perplexity, Bing Copilot) use canonicals to deduplicate content. A wrong canonical means AI may cite the wrong page version or ignore this page entirely.
How to fix: Verify this canonical is intentional. AI models cluster near-duplicate URLs and choose one representative page. If canonical points to a different URL, AI will only index that target URL, not this page.
Why it matters: AI shopping agents (Google Shopping, ChatGPT, Bing Copilot) filter by return flexibility. Products with return policy schema get priority placement in AI product comparisons.
How to fix: Add MerchantReturnPolicy schema with: returnPolicyCategory (e.g., MerchantReturnFiniteReturnWindow), merchantReturnDays, returnMethod, returnFees. Link it from Product/Offer via hasMerchantReturnPolicy.
Why it matters: Fulfillment speed factors into AI product recommendations. Complete shipping schema means AI can show 'Free shipping, delivers in 2-3 days' — a strong conversion driver.
How to fix: Add OfferShippingDetails schema with: shippingRate, shippingDestination, deliveryTime (handlingTime + transitTime). AI agents deprioritize products without shipping info.
2/4 signals — missing: @id in JSON-LD, sameAs links (Wikipedia, LinkedIn)
Why it matters: @id creates a persistent entity identifier that connects your schema across pages and platforms, which is how AI builds a verified brand entity.
How to fix: Add: @id in JSON-LD, sameAs links (Wikipedia, LinkedIn). Use @id in JSON-LD to create a unique node identifier. Ensure your brand name is identical in title, schema, and OG tags. Link to Wikipedia/Wikidata via sameAs.
Grade 20 — too complex for broad AI citation
Why it matters: Content above grade 12 readability is harder for AI to extract as clear, concise answers. Pages using clear headers and approachable language are cited more often.
How to fix: Simplify sentences (target 15-20 words average), use common words, break complex ideas into shorter paragraphs. AI extracts content for general audiences — if it's too academic, AI skips it.
53% good, 5 poor, 9 missing
Why it matters: AI uses alt text to understand images for visual search (Google Lens, Bing Visual Search), product recommendations, and accessibility. Quality alt text opens more discovery channels.
How to fix: Improve alt text: use 3-15 descriptive words per image. Include product name, key feature, and context. Bad: 'image1' or 'photo'. Good: 'Red leather wallet with RFID protection — front view'.
Why it matters: Content without citations or expert quotes appears unverified to AI. AI assistants prefer content backed by named sources, data references, and expert opinions.
How to fix: Add 2-3 expert quotes or data citations per major page. Use <blockquote> for quotes and link to authoritative sources (.gov, .edu, Wikipedia, industry reports).
Why it matters: WebMCP is how AI agents will interact with your store (search products, add to cart, check availability). Google and Microsoft are co-developing this standard.
How to fix: WebMCP (W3C Community Group standard, Chrome 146+) lets pages declare structured tools for AI agents. Add toolname and tooldescription attributes to <form> elements, or include a <script type='application/webmcp+json'> manifest.
No trust widget (Heureka / Trustpilot / AggregateRating) detected
Why it matters: SK/CZ buyers trust Heureka Ověřeno more than other trust signals. LLMs (ChatGPT, Perplexity) and Google SGE cite shops with AggregateRating schema first.
How to fix: For SK/CZ shops: install the Heureka Ověřeno zákazníky widget (free for verified sellers — https://sluzby.heureka.sk). Alternatives: Trustpilot, Google Reviews with AggregateRating schema. Place the trust widget on product + category + footer.
No AI bot restrictions (allowed by default)
ChatGPT-User and PerplexityBot get the same page as a browser
573 words in raw HTML (3.1% text ratio) — readable by AI crawlers without executing JS
A search with no matches still offers products, suggestions or a way to get help
573 words — rich content for AI analysis and citation
Proper structure: 1 H1 → 6 H2s → 4 H3s — clear content outline
4/6 semantic elements: <nav>, <main>, <header>, <footer>
Answers 4/5 key shopper questions: shipping, returns/refunds, availability, payment
42 contextual internal links per 1,000 words — strong knowledge graph signal
Semantic content containers found
Not present — not required for AI visibility
Why it matters: llms.txt is the emerging standard for AI-readable site descriptions — like robots.txt was for search engines. Without it, AI assistants have no guided overview of your store.
How to fix: Create /llms.txt in Markdown format:
# Your Store Name
> One-line summary of your business and key offerings.
Detailed description paragraph.
## Products
- [Product Category](URL): Description
## About
- [About Us](URL): Company history and values
See llmstxt.org for the full specification.
Not present — not required for AI visibility
Why it matters: llms-full.txt provides AI models with your entire site content in one parseable file. It's the difference between an AI having a brief overview vs. deep knowledge of your products and services.
No Product schema on this homepage — Product schema lives on product detail pages, which this single-URL scan didn't visit
Why it matters: Product schema only belongs on product pages, so its absence on this page is expected — not a defect. Scan a product URL to evaluate it.
How to fix: To audit Product schema, run the scan on a product detail URL. On product pages add: name, description, image, offers (price, priceCurrency, availability), brand, sku, aggregateRating.
A correctly spelled query and the same query with a typo return similarly sized pages (148665 B vs 146373 B) — consistent with typo tolerance
Why it matters: This is a raw measurement, not a verdict, and it is never scored: page size is only a proxy for result count and themes differ. It is here so you can see what we saw.
How to fix: Verify by hand: search for a product, then search again with one letter transposed. If the second search finds nothing, a search engine with fuzzy matching and synonyms will recover those sales.
No review schema on this homepage — review/rating schema typically lives on product pages
Why it matters: Review schema is product-page-specific, so its absence here is expected, not a defect.
How to fix: Scan a product URL to audit review schema. On product pages add AggregateRating (ratingValue, reviewCount, bestRating) and individual Review schemas.
Not present — not required (the ChatGPT-plugins manifest was retired in 2024)
Why it matters: ai-plugin.json enables AI agents to interact with your store programmatically — search products, check prices, process orders. This is the bridge between AI assistants and your e-commerce functionality.
No product feed was sampled — GTIN/EAN coverage could not be verified
Why it matters: GTIN is how an AI shopping agent matches your product to the same product in other shops. Without a feed we cannot measure it — this is not a finding against you.
How to fix: Publish a Google Merchant or Heureka product feed with GTIN/EAN populated, then re-scan to measure coverage.
Not present — optional; SpeakableSpecification has limited assistant adoption
Not present — optional; AI-bot permissions are enforced via robots.txt, not ai.txt
Why it matters: ai.txt is an emerging standard for fine-grained AI permissions beyond robots.txt. It gives you control over how AI uses your content.
63/100 — partly ready (Access 100 · Understand 50 · Discover 50 · Transact 50)
Why it matters: Whether an autonomous shopping agent (a ChatGPT operator, Perplexity, the assistants coming next) can do four things on your shop: get in, read and understand the products, discover the whole catalogue, and act (cart, terms). Almost nobody measures this, and for SK/CZ shops nobody does.