Shakaview

The 10-point AI-search checklist for Shopify product pages

AI shopping assistants recommend the products they can read. These are the ten things we check on every product page, why each one matters and how to fix it.

By Shakaview · Updated October 26, 2026

Check your store against it

Free, about ten seconds, no login.

Before the checks

First, let the crawlers in.

None of the ten checks matter if a crawler isn’t allowed to fetch the page, and your robots.txt decides that. These are the crawlers the checker tests against one of your product pages.

CrawlerCompanyWhat it feedsIf you block it
GooglebotGoogleGoogle Search, AI Overviews and AI ModeYour products drop out of these answers.
BingbotMicrosoftBing and CopilotYour products drop out of these answers.
OAI-SearchBotOpenAIChatGPT searchYour products drop out of these answers.
Claude-SearchBotAnthropicClaude’s search resultsYour products drop out of these answers.
PerplexityBotPerplexityPerplexity answersYour products drop out of these answers.
GPTBotOpenAIModel training onlyYour pages aren’t used to train models. Search answers are unaffected.
ClaudeBotAnthropicModel training onlyYour pages aren’t used to train models. Search answers are unaffected.

Shopify’s default robots.txt allows all of them. If yours has been edited, or a firewall setting blocks bots, check that the search crawlers still get through: a firewall can turn them away even when robots.txt lets them in.

The checklist

The ten checks.

Weighted by how much each one changes what an assistant can say about a product. Together they add up to 100.

01 · 20 points

Product markup

The product page’s HTML includes Product (or ProductGroup) JSON-LD with a name, a price and availability.

Why it matters

An assistant answering “running shoes under $150 that are in stock” needs price and stock as data, not as text it has to find in your layout. Most AI crawlers read the HTML your server sends and don’t run JavaScript, so markup a script adds after the page loads may never be seen.

How to fix it

Most current Shopify themes output this already. If yours doesn’t, add Shopify’s structured data filter to the product template.

sections/main-product.liquid
<script type="application/ld+json">
  {{ product | structured_data }}
</script>

02 · 15 points

Product attributes

The product markup lists at least two specific attributes beyond name and price: material, color, size, weight, a GTIN or MPN, or additionalProperty entries.

Why it matters

Shoppers ask for specifics: wool, machine washable, under 300 grams. Attributes in markup are facts an assistant can match and compare. The same facts in a paragraph are guesses. A GTIN lets it match your product with reviews and listings elsewhere.

How to fix it

Fill in each variant’s barcode in Shopify, make sure your theme outputs it as a GTIN, and add the attributes buyers compare on as additionalProperty entries, fed from metafields.

Inside the Product JSON-LD
"material": "Merino wool",
"color": "Natural white",
"gtin13": "0123456789012",
"additionalProperty": [
  { "@type": "PropertyValue", "name": "Weight", "value": "280 g" },
  { "@type": "PropertyValue", "name": "Care", "value": "Machine washable" }
]

03 · 10 points

FAQ markup

The product page has FAQPage JSON-LD with at least three questions and answers.

Why it matters

Questions and answers are the shape assistants answer in. Google no longer shows FAQ rich results for most stores, but the markup still states sizing, shipping, care and returns as plain facts that machines parse reliably.

How to fix it

Write three to five real questions per product or product type, show them on the page, and mark up the same text. Never mark up questions that aren’t visible.

A second JSON-LD block on the product page
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [{
    "@type": "Question",
    "name": "Does it run true to size?",
    "acceptedAnswer": { "@type": "Answer", "text": "Yes. Most people take their usual size." }
  }]
}

04 · 10 points

Title tag

The page title is 30 to 60 characters long and includes your brand name.

Why it matters

The title is the label most systems use for your page, in search results and in the sources an assistant cites. Too short says nothing, too long gets cut, and without the brand a cited product loses who makes it.

How to fix it

In Shopify, open the product and edit its search engine listing: product name, the main differentiator, then your brand.

05 · 10 points

Meta description

The meta description is 80 to 160 characters and written for this product: not missing, not shared with another product and not just the opening of the description.

Why it matters

It is the summary systems show and read first. Left empty, most themes fall back to the start of your product description, which often opens with a story rather than what the product is.

How to fix it

In the product’s search engine listing, write one sentence on what it is and one on why it is different.

06 · 10 points

Image alt text

At least half of the product’s images have alt text that says more than the product name.

Why it matters

Assistants and image search can’t see your photos. Alt text is how they learn what each image shows: the color, the angle, the product in use. It is also how shoppers using screen readers buy from you.

How to fix it

In Shopify, open each product image and describe it: “Natural white runner, side view, on a wooden floor.”

07 · 10 points

Use-case context

The product’s name and description mention at least two kinds of context: who it is for, a gift or occasion, a season, an activity or setting, or a plain “best for”.

Why it matters

People ask assistants by need, not by product name: a gift for a runner, warm boots for a rainy commute. Pages that say who and when a product is for are the ones that match those questions.

How to fix it

Add a short “Best for” line to each description, written the way your customers describe why they bought it.

08 · 5 points · store-wide

Organization markup

The home page has Organization (or OnlineStore) JSON-LD with your name, logo and social profiles in sameAs.

Why it matters

It ties your store to the same brand everywhere else on the web, so an assistant knows your social accounts, your reviews and your store are one business.

How to fix it

Add one Organization block to your theme layout, with links to every profile you keep up to date.

layout/theme.liquid, home page only
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Brand",
  "url": "https://yourbrand.com",
  "logo": "https://yourbrand.com/logo.png",
  "sameAs": [
    "https://www.instagram.com/yourbrand",
    "https://www.tiktok.com/@yourbrand"
  ]
}

09 · 5 points

The product page has BreadcrumbList JSON-LD with at least two levels, such as Home, then Shoes, then the product.

Why it matters

It tells machines where the product sits in your catalog, which matters when someone asks for a category (“men’s running shoes”) rather than a product.

How to fix it

Most themes can show breadcrumbs. Turn them on and mark up the same path, or add the markup on its own.

10 · 5 points

Reviews in markup

The product markup includes individual reviews, not only a star rating summary.

Why it matters

Assistants quote what reviewers say: runs small, great for wide feet. A star average gives them nothing to quote. Some review apps add their markup with JavaScript after the page loads, which most AI crawlers never run.

How to fix it

Ask your review app for server-rendered markup (many have a setting or a metafield sync) and include a handful of recent reviews.

Scoring

How the checker scores.

Each product earns the points for the checks it passes, out of 100. Your store’s score is the average across six products spread through your catalog. Organization markup is read once from your home page and counts the same for every product.

The checker reads the HTML your server sends, not the page after its scripts have run, because that is what most AI crawlers see. Markup that an app adds with JavaScript won’t count, for them or for us.

  • Ready (80 or more): AI assistants can read most of what matters on your product pages.
  • Partly readable (50 to 79): Assistants can find your products but miss facts they need to recommend them.
  • Hard to read (0 to 49): Assistants get the basics at best, and little they could use to recommend you.

See where your store stands.

Run the free checker for a score in about ten seconds, or ask for the free Store Audit and we’ll check every product and write the fixes.

Check my storeAbout the free audit