MR Digital

Your shop has a second storefront. Nobody built it.

The one machines read. ChatGPT, Claude, Perplexity, DeepSeek, Google, Apple, Amazon and a dozen more read your shop already, exactly as it stands. Our job is to put something there they can use.

The check reads your shop the way an assistant does and emails the report in about a minute. Nothing is installed and nothing on your shop changes.

An assistant does not see the page you see

It retrieves the raw text and the data behind it, then answers from that. A size chart inside a tab, a delivery time in a widget, a material shown only in a photograph: none of it exists as far as the answer is concerned. So good shops go unnamed while worse ones get recommended. The shop is fine. It simply cannot be read.

Both products below run the same engine. They read the descriptions the merchant already wrote, pull the comparable facts out of them, and publish those facts in the forms assistants consume. Nothing is invented, and nothing written by hand is ever overwritten.

Three things you were told would fix this

Most of what is sold as AI visibility is a file, a scoreboard, or the same SEO with a new name. Here is what each one measurably does.

1. Classic SEO

Most shop owners assume that ranking well on Google means an assistant will find them. The measurements say otherwise.

12%of the pages ChatGPT, Gemini and Copilot cite rank in Google’s top 10 for the same question
8%for ChatGPT on its own. Four out of five pages it cites do not rank in Google at all for that query
38%of Google’s own AI Overview citations come from the top 10, down from 76% a year earlier

Ahrefs, 15,000 queries across four assistants, July 2025; 863,000 keywords and 4 million AI Overview citations, March 2026.

Google ranks pages. An assistant retrieves passages, then picks the ones it can quote. A page with ten years of backlinks and no stated fact contributes nothing to that answer. A page with no authority and a clear, structured description gets quoted.

2. An llms.txt file

97%of llms.txt files received zero requests from any AI crawler in May 2026Ahrefs, server logs of 137,000 domains

This is what most AI visibility tools actually deliver: one text file at the root of the site, listing the pages. It is cheap to generate, which is why so many generate it. Google has stated it is not required and has no effect on visibility.

Both our products publish one too, linked to a clean text version of every page it names, because a few tools do read it. It is the smallest module here, not the product. A table of contents that points at unreadable pages points at nothing.

3. Asking ChatGPT about your brand

<1 in 100chance that ChatGPT gives the same list of brands twice for the same promptSparkToro, 2,961 runs, 600 testers, January 2026

The other thing sold as AI visibility is a scoreboard: a tool types a handful of prompts into ChatGPT, counts how often you appear, and calls it your share of voice. Even a properly sampled average tells you the outcome on a given day, for a given phrasing, and nothing about the cause.

What can be measured is the step before the answer: which AI crawlers actually fetched your pages, and what they found there. That runs on your own server, it is a count and not a sample, and when it moves you know what you changed.

The market sells a file and a scoreboard. Neither reaches the reason. An assistant either has a fact to quote from your shop, or it does not.

Pick your platform

One engine, two builds, because the platforms are not the same and a port would have been worse than either.

WordPress and WooCommerce

49 EUR one payment

  • Structured data merged into Yoast or Rank Math, never a second Product block
  • Comparable attributes read from your own descriptions, dictionary in your language
  • A plain text version of every page, an AI sitemap and a live llms.txt
  • An FAQ block, alt text, IndexNow and Google Preferred Sources
  • Crawler check, and the AI crawler requests your site really receives
  • Whole catalogue in one pass. 99 EUR covers five sites
See the plugin Live demo shop

Shopify

Free then 99 USD a year

  • The same deterministic extraction, on your Shopify catalogue
  • Structured data that extends the markup your theme already publishes
  • Every value written into your own metafields, so it stays if you leave
  • Comparison tables on collections, built from attributes that actually vary
  • Plain text page per product, a live llms.txt and agents.md
  • Free plan with no expiry: crawler check, coverage score, three products
See the app Shopify App Store

What lands on the page

The same surfaces on both platforms, built from your own words and published on your own domain.

Structured data

Extends what your theme or SEO plugin already publishes. One Product node on the page, never two.

Comparable attributes

Read out of ordinary prose with a dictionary you control, in any language. An item with no attributes never makes a comparison table.

An FAQ block

Built only from questions your shop can actually answer. Where there is no real answer, nothing is published.

Plain text pages

Every product at its own address, without menus, scripts or layout. This is the page a crawler is offered.

A live llms.txt

Generated on request from what is published right now, linked to those text pages, never a file that goes stale on a timer.

Comparison tables

On category and collection pages, comparing the figures your products actually state.

Described photographs

Alt text written from the product’s own attributes, where an image has none. What a human wrote is never replaced.

Crawler reachability

Asks your own site the way each crawler would, and names who got what back. A correct robots.txt proves nothing on its own.

Nineteen crawlers counted

GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, PerplexityBot, DeepSeekBot, DuckAssistBot, Applebot, Amazonbot, Bytespider, CCBot, Google-CloudVertexBot, MistralAI-User, Meta-ExternalAgent and more, each request checked against the operator’s own network.

The same product page, before and after

Every value on the right is quoted from the description the shop already wrote. Nothing is generated, nothing is guessed, and the sentence each fact came from is shown beside it.

What the page publishes today

{ "@type": "Product",
  "name": "Espresso Machine E20",
  "image": "e20-front.jpg" }

No offer, no availability, no attributes. An assistant can name the product and nothing else.

What it publishes after

"offers": { "price": "449", "priceCurrency": "EUR", "availability": "InStock" }
Grinderbuilt in, conicalfrom “with a built-in conical burr grinder”
Pressure15 barfrom “15 bar pump for a proper crema”
Water tank1.8 lfrom “1.8 litre removable water tank”
Delivery2 working daysfrom “Ships within 2 working days”

Before any of that, can they reach you at all?

Hosting firewalls and password pages block AI crawlers silently. Nothing appears in your analytics, nothing breaks, your normal traffic is unaffected. This was a live result on one of our own domains, same page, same minute, only the name of the requester changed:

Crawler check
GPTBot200 OK
PerplexityBot406 Not Acceptable
ClaudeBotconnection closed

Perfect content on a page no assistant can fetch is worth nothing, and no amount of writing fixes it. So the check runs first, on both platforms, and tells you what to unblock before anything is published.

The free check

See what an assistant reads from your shop

Type your address and the report arrives by email in about a minute. Nothing is installed, nothing on your shop changes, and you do not need to buy anything to read it.

What the report tells you

  • Which AI crawlers can reach you, and exactly what each one got back
  • A readiness score, with every subtraction named so you can check the arithmetic
  • What is blocking, what needs a setting and what is one click away, each with its fix
  • How much of your catalogue an assistant can actually read, in numbers, and whether that is the whole catalogue or a sample
  • One of your own products, before and after, with the sentence each value came from beside it

The report is a page at a private link, not a PDF attachment. It is not indexed, and it is deleted after ninety days.

We store the shop address, the email address typed into the form and what the scan found, for 90 days, and then delete them. Nothing else is collected and nothing is shared. To have yours removed sooner, write to hello@mrdigital.ro.

Three things we chose not to build

All three sell well, and we could have shipped them. They do nothing for the shop, so the effort went elsewhere.

No prompt sampling

A share of voice percentage graphs beautifully, but the answer changes per person, per session and per day. A number that cannot be reproduced measures nothing. We count the real crawler requests your site receives.

No AI-written blog posts

They would fill a site fast and feel like progress. Generated filler is increasingly downranked, so it damages what it claims to help, and publishing words you never wrote is not honest.

No access to your orders

It would allow flattering revenue reports. We do not need it for what we do, so we never ask for it. Product content is read and written; orders, customers and payments are never touched.

Extraction runs on pattern matching, with no model involved. Nothing leaves your site, there is no cost per product, and the same description always produces the same result.

Running this for your clients

If you look after shops for other people, the work is the same on every one: install, run it once over the catalogue, and the client keeps whatever it publishes.

  • Nothing to maintain per client. It runs in the background and adds no code to the storefront.
  • Nothing you have to defend later. It publishes only facts already written on the client’s own pages.
  • No access to the client’s orders, customers or payments, so it raises no questions in a security review.
  • A reachability report you can send the client as it is, showing which crawlers can and cannot see their shop today.

On WordPress the Studio licence covers five sites for 99 EUR, one payment, and a licence moves between sites whenever you want. Above five, and for white label, invoicing through your own company or a partner arrangement, write to us. On Shopify each store installs the app itself, so there is nothing to move.

Send

This opens your mail client with the message ready. Nothing is stored on this page.