AI readiness checker
Can AI agents sell your product?
Paste a product page. We read it the way a shopping agent does and show what it can tell a shopper, what it cannot, and how to fix each gap.
Free, no sign-up. We read the page once and keep nothing.
- Crawler access
- Product data
- Price and stock
- Delivery and returns
Example report
haldenaudio.example/products/arc-wireless-headphones
Halden Audio is a made-up store. Paste your own product page above to replace this report.
Agent readiness
An agent can find this product, but it has to guess or stay silent on some of the answers shoppers ask for.
- 11 ready
- 4 to improve
- 2 missing
- 1 not applicable
What an agent can tell a shopper
Is this in stock, and what does it cost?
Arc Wireless Headphones costs $189.00, is in stock, and is sold by Halden Audio.
- Product
- Arc Wireless Headphones
- Brand
- Halden Audio
- Price
- $189.00
- Stock
- in stock
- Product code
- ARC-BLK
- Rating
- 4.6 from 212 reviews
- Seller
- Halden Audio
Can an agent open the page?
3 of 5 ready
AI and search crawlers13 of 15 allowedImprove
What we found
robots.txt blocks 2 of 15 crawlers from this page: GPTBot, ClaudeBot.
Why it matters
An agent can only recommend what its crawler was allowed to read. One Disallow line can hide your whole catalog from an assistant.
What we looked for
robots.txt rules for this page, for 15 crawlers: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User, Google-Extended, Googlebot, Bingbot, Applebot, Applebot-Extended, Amazonbot and meta-externalagent.
How to fix it
Remove the Disallow rules that cover product pages for the crawlers you want to be found by. Blocking a training crawler and allowing a search crawler from the same company is a valid choice: they are separate names.
User-agent: OAI-SearchBotAllow: /products/Allowed in resultsReady
What we found
No noindex directive on the page or in its headers.
Why it matters
A noindex directive tells search engines and the agents built on them to leave the page out, even when they can read it.
What we looked for
The robots meta tag and the X-Robots-Tag response header.
One address per productImprove
What we found
No canonical link. An agent may treat each variant address as a different product.
Why it matters
Colour, size and tracking parameters create many addresses for one product. A canonical link tells an agent which one to keep and cite.
What we looked for
A link rel=canonical tag that points to this site.
How to fix it
Add a canonical link in the head of every product page, pointing to the clean product address.
<link rel="canonical" href="https://yourstore.com/products/arc-headphones">Readable without scripts240 wordsReady
What we found
240 words are in the HTML before any script runs.
Why it matters
Many AI crawlers do not run JavaScript. What is not in the first HTML response does not exist for them.
What we looked for
The words in the HTML the server sends, before any script runs. 150 words or more is enough to describe a product.
Page title and summaryReady
What we found
Title, 38 characters: Arc Wireless Headphones | Halden Audio
Why it matters
The title and the meta description are the headline and the one-line summary an agent shows when it cites your page.
What we looked for
A title of 15 to 70 characters and a meta description.
Does it know which product this is?
4 of 6 ready
Product dataReady
What we found
A schema.org Product block is on the page.
Why it matters
Agents do not guess from the layout. They read a block of structured data that states what the page sells, in a format every assistant understands.
What we looked for
A schema.org Product or ProductGroup in JSON-LD. Microdata in the HTML counts in part.
Product nameArc Wireless HeadphonesReady
What we found
Arc Wireless Headphones
Why it matters
The name is how an agent matches your product to what the shopper asked for.
What we looked for
Product.name, at most 150 characters.
Description31 wordsReady
What we found
A description of 31 words.
Why it matters
The description is where an agent finds the answers: material, size, fit, what it works with, who it is for.
What we looked for
Product.description of at least 20 words.
Images1 imageImprove
What we found
1 image(s) listed. Three or more lets an agent show the product from more than one side.
Why it matters
Assistants show a picture next to a recommendation. With no image in the data, the product is shown without one or not at all.
What we looked for
Product.image with at least 3 images.
How to fix it
List every product image in the data, the main one first. Use full addresses that load without a cookie.
"image": ["https://yourstore.com/arc-front.jpg", "https://yourstore.com/arc-side.jpg"]BrandHalden AudioReady
What we found
Halden Audio
Why it matters
Shoppers ask for brands by name. Without a brand field an agent cannot filter or confirm it.
What we looked for
Product.brand.
Product codeSKU ARC-BLKImprove
What we found
SKU ARC-BLK, but no GTIN. Agents cannot match it across sellers.
Why it matters
A GTIN (the barcode number) lets an agent know your product is the same one sold elsewhere, so it can compare offers and trust reviews.
What we looked for
A GTIN, or else an MPN or SKU, on the product or its offer.
How to fix it
Add the barcode number as gtin where the product has one. For your own products, a stable SKU is the minimum.
"gtin13": "5060000000017", "sku": "ARC-BLK"Can it quote the price and the stock?
2 of 2 ready
Price and currency$189.00Ready
What we found
189.00 USD
Why it matters
An agent will not recommend a product it cannot price. A number without a currency is not a price.
What we looked for
Offer.price as a plain number and Offer.priceCurrency as a 3 letter code.
StockIn stockReady
What we found
InStock
Why it matters
An agent that sends a shopper to a sold-out product loses that shopper's trust, so it prefers products that state their stock.
What we looked for
Offer.availability with a schema.org value such as InStock, OutOfStock or PreOrder.
VariantsNot applicable
What we found
No variants declared. That is right for a product sold in one version.
Why it matters
A shopper asks for a size or a colour, not for a product page. Variants in the data let an agent answer "do they have it in medium?".
What we looked for
ProductGroup with hasVariant, each variant with its own offer.
How to fix it
Describe the product as a ProductGroup, name what varies, and give every variant its own price and stock.
"@type": "ProductGroup", "variesBy": ["https://schema.org/color"], "hasVariant": [ ... ]Can it answer delivery, returns and reviews?
2 of 4 ready
DeliveryMissing
What we found
No delivery cost or time in the product data.
Why it matters
"When will it arrive and what does delivery cost?" is asked before almost every purchase. Without this data an agent can only say it does not know.
What we looked for
Offer.shippingDetails with a rate and a delivery time.
How to fix it
Add your delivery cost and your handling and transit days for each region you ship to.
"shippingDetails": { "@type": "OfferShippingDetails", "shippingRate": { "@type": "MonetaryAmount", "value": "0", "currency": "USD" } }ReturnsMissing
What we found
No return policy in the page data.
Why it matters
A clear return window removes the last doubt before a purchase, and agents quote it when it is in the data.
What we looked for
A MerchantReturnPolicy on the offer or on the store's Organization.
How to fix it
State the return window in days, who pays for the return and how it is sent back.
"hasMerchantReturnPolicy": { "@type": "MerchantReturnPolicy", "merchantReturnDays": 30 }Ratingsrated 4.6Ready
What we found
4.6 from 212 reviews.
Why it matters
"Is it any good?" An agent answers from the rating and the number of reviews behind it.
What we looked for
Product.aggregateRating with a rating and a review count.
SellerHalden AudioReady
What we found
Halden Audio
Why it matters
An agent names the store it sends a shopper to. A named seller is also what it checks reviews and policies against.
What we looked for
Offer.seller or an Organization on the page.
How the score works
Four questions every shopping agent asks
The score is the share of 18 checks your page meets, weighted by how much each one decides whether an agent can recommend the product. A check that does not apply to your product is left out, not counted against you.
haldenaudio.example/arc
- GPTBot
- ClaudeBot
- Googlebot
- Response
- 200 OK
- In results
- Allowed
5 checks
- AI and search crawlersWeight 3 of 3
- Allowed in resultsWeight 3 of 3
- One address per productWeight 1 of 3
- Readable without scriptsWeight 1 of 3
- Page title and summaryWeight 1 of 3
9 of 36 points
Can it open the page?
If the page is closed to crawlers or kept out of results, nothing else on it is read. Those two checks count triple.
{ "@type": "Product", "name": "Arc Wireless Headphones", "brand": "Halden Audio", "sku": "ARC-BLK"}6 checks
- Product dataWeight 3 of 3
- Product nameWeight 3 of 3
- DescriptionWeight 2 of 3
- ImagesWeight 2 of 3
- Product codeWeight 2 of 3
- BrandWeight 1 of 3
13 of 36 points
Does it know the product?
An agent matches a request to a named product. Without the data block and a name it has nothing to match. Both count triple.
$189.00 USD
In stock
3 colors, one offer each
- Black
- In stock
- Cobalt
- In stock
- Slate
- 2 left
3 checks
- Price and currencyWeight 3 of 3
- StockWeight 3 of 3
- VariantsWeight 1 of 3
7 of 36 points
Can it quote an offer?
An agent will not recommend what it cannot price or promise in stock. Price and stock count triple; variants count once.
- Delivery
- $4.90
- Arrives
- 3 to 5 days
- Returns
- 30 days, free
- Sold by
- Halden Audio
- Rating
- 4.6 of 5
4 checks
- DeliveryWeight 2 of 3
- ReturnsWeight 2 of 3
- SellerWeight 2 of 3
- RatingsWeight 1 of 3
7 of 36 points
Can it answer the last doubts?
Delivery, returns and the seller are what shoppers ask just before they buy. Each counts double; ratings count once.
After the report
From a score to a product agents can sell
Most gaps are one block of structured data away. Work through them in this order.
robots.txt
- GPTBot
- Allowed
- ClaudeBot
- Allowed
- Googlebot
- Allowed
- noindex
- Not set
Open the door
Fix crawler rules and noindex first. Nothing else counts until an agent is allowed to read the page.
Product data
- Price
- $189.00
- Stock
- In stock
- Product code
- ARC-BLK
- Brand
- Halden Audio
State the facts
Add the Product data block with name, description, images, brand, product code, price and stock. Copy the samples from each check.
Delivery and returns
- Delivery cost
- $4.90
- Delivery time
- 3 to 5 days
- Return window
- 30 days
- Return shipping
- Free
Answer before they ask
Add delivery cost and time and your return window, so an agent can settle the last doubts for you.
Feed formats
- ChatGPT
- Meta
- TikTok
Or let the store do it
A Tanqory store writes the Product data for every product and variant from your catalog, and builds product feeds in the formats that ChatGPT, Google, Meta, TikTok and Pinterest read. Start your store
Questions
About the AI readiness checker
What does the AI readiness checker do?
It reads one product page the way an AI shopping agent does: the HTML the server sends, the structured data inside it and the site's robots.txt. It then runs 18 checks and shows, for each one, what it found, why it matters and how to fix it.
Does a high score mean AI assistants will recommend my product?
No. The score tells you whether an agent can read and trust the facts about your product. Which products an assistant recommends is decided by each assistant, and no tool can promise a placement. A low score, on the other hand, is a reliable sign that agents are missing or misreading your product.
Which AI crawlers do you check?
Fifteen: GPTBot, OAI-SearchBot and ChatGPT-User from OpenAI; ClaudeBot, Claude-SearchBot and Claude-User from Anthropic; PerplexityBot and Perplexity-User; Google-Extended and Googlebot; Bingbot; Applebot and Applebot-Extended; Amazonbot; and meta-externalagent. We read your robots.txt rules for each of them against the page you paste.
Do you store my address or the report?
No. The page is read once to build the report. The address, the page and the report are not saved or logged, and nothing is added to a mailing list. You do not need an account.
Why is a rating marked as not applicable?
A product with no reviews yet has nothing to publish, and marking up reviews you do not have breaks the rules of every search engine. So a missing rating does not lower the score. Variants work the same way for products sold in one version.
My page shows a price, so why does the check say it is missing?
The price is visible to people but not stated in the page's structured data, or it is added by a script after the page loads. Many AI crawlers do not run scripts, so they never see it. Put the price and currency in the Product data block in the HTML your server sends.
Does it work for stores on any platform?
Yes. It reads public web pages, so it works for any store on any platform, including custom-built ones. It reads one page at a time; check a few of your best sellers to see whether a gap is a one-off or comes from your theme.
What does a Tanqory store do about these checks?
A Tanqory store writes Product structured data for each product you publish, straight from your catalog: name, description, images, brand, SKU or barcode number, and a price and stock for every variant. Delivery cost and time, your return policy and your seller name are yours to set: state them on the product page so shoppers and agents can read them. It also builds product feeds in the formats used by ChatGPT, Google Merchant Center, Meta, TikTok and Pinterest. Connecting a feed to a channel is a step you take with that channel.
Build a store agents can read
Describe what you sell and Tanqory drafts the store, with product data written for shoppers and for the agents that shop for them, so you do not mark up each page by hand.
The checks are signals that AI agents may use to find, read and recommend products. Passing them does not mean an agent will show your product. Check another page