A storefront can look finished to every shopper and still give an AI assistant almost nothing to read. A person reads the design, while an assistant reads the code and structured data underneath it. Margareta Petrovic, who founded Visibility Mesh, made that case in her Forbes Business Council column of August 4, 2026. The evidence came from our study of 216 live Shopify storefronts.
The column on the Forbes Business Council site is written for business owners and leaves the technical detail out. This page explains the study behind it and the order of fixes it points to.
Why does a polished storefront fail an AI assistant?
A shopper reads photography, layout and persuasive copy. An assistant reads none of that the way a person does. It checks whether its crawler is allowed in and whether the brand uses one name everywhere. It looks for structured data on each product with a price, availability and an identifier. It also looks at whether questions are answered in sentences it can lift, and whether reviews and mentions point back to the same business.
On a store where those things are missing, the owner sees the hero image, while the assistant finds a Product block with no price and a description made mostly of adjectives.
What did the study behind the column find?
We scanned 229 live Shopify storefronts with our own engine in June 2026. The engine could not read 13 of them, so it excluded them instead of scoring them as zeros, and 216 stores received a score. The median store scored 46 out of 100. Two thirds scored below 50, and the highest scoring store reached 71.
Only 27 of the 216 stores, about one in eight, passed all five layers at half marks or better. The weakest layer was Entity Integrity, the brand and product data that assistants read, and 72% of stores failed it. The full method is on the State of AI Visibility 2026 study page.
A later benchmark of 245 Shopify storefronts, read on August 22, 2026, came out at the same median of 46. That reading has its own page.
Which fixes come first?
Fix the five layers in this order. If the crawler cannot get in, nothing you fix further down will be read, which is a lot of careful work for an audience of nobody.
- Open the front door. Check yourdomain.com/robots.txt for Disallow rules under the AI crawlers. Product text should also appear in View page source, before any JavaScript runs. On Shopify, robots.txt is edited in Online Store > Themes > Edit code, in the robots.txt.liquid template.
- Settle the identity. Go to Settings > Store details and use one exact business name. The homepage source should carry one Organization block, with links to the profiles you own.
- Fix the product data. Every product page needs one Product block with a price, availability and an identifier, and the price must match the one on the page. In Shopify, open Products, choose a product and enter the barcode for each variant in its Barcode field.
- Write the answers. Put a short FAQ on the pages buyers ask about, in visible text, written the way people ask.
- Connect the proof. Reviews and press mentions should name the same business, at the same address, as your own site.
The free scan reads five of your key pages and tells you which step your store starts on. Our report on which problems most often stop a website from being read shows how often each layer failed in both frozen studies.
What can AI not read on a womenswear label with a lookbook homepage?
A womenswear label on Shopify with 120 products and a beautiful lookbook homepage is a common case. Its fabric and fit notes live in a size chart image. Its theme prints one Organization block and an app prints a second one with an older logo. Its robots.txt is the Shopify default, so the door is open.
That store starts at step 2. The owner removes the duplicate Organization block first. After that, the fabric content and fit notes move from the image into the product description as plain sentences. The lookbook stays exactly as it is.
What came next in the column series?
The next column, Before You Buy More Traffic, Fix the Proof, was published on September 23, 2026. It adds a four question test any owner can run on their own website.
Questions readers ask about the column
Does a high Google ranking mean AI can read my store?
A high Google ranking does not prove that AI can read your store. Rankings depend partly on a site's history and links. Legibility depends on what the crawler can read on the page today, so a store can pass one test and fail the other.
Is being unreadable to AI a Shopify problem?
Shopify ships good technical defaults, such as a sitemap and canonical tags. The gaps usually come from themes, apps and edits made over time. They add duplicate markup, hide text in images or load content only with JavaScript.
How long does the first fix take?
Fixing robots.txt rules and a duplicate Organization block is usually a matter of hours. Correcting Product markup across a whole catalog takes longer, since every template and app that prints markup has to agree.
Where can I read the column itself?
The column is on the Forbes Business Council site, published on August 4, 2026, and its title begins "Your E-Commerce Storefront May Look Perfect". The link at the top of this page opens it.
Sources
- Margareta Petrovic, Forbes Business Council column on storefronts AI cannot read, Forbes Business Council, August 4, 2026, read November 9, 2026.
- The State of AI Visibility 2026, Visibility Mesh, 216 Shopify storefronts scored in June 2026.