Find out in 60 seconds. It is free and there is nothing to install. Then keep score: we ask the buying questions every week, and count the times an AI actually opened your pages.
Every tool in this category runs sample questions and reports a percentage. That is an estimate, and it should be called one. Because Goppa also runs inside your shop, there is a second number that is an actual count.
We ask real buying questions on every engine we run, several times, and report how often your shop comes up, with the number of questions next to it, always.
When someone asks ChatGPT a question and it opens one of your pages to answer, your server sees it. When that person clicks through, your server sees that too. Those are events, not estimates. We count them.
We do not report a count of times an AI mentioned you to other people. No API publishes that, and any number claiming to be it is an estimate wearing a count's clothes.
A person browses menus and images. An agent looks for price, stock and structured data. If that information is not clear, the product can be left out of the answer.
Aurora Fit is a controlled demonstration: the same store, measured with the same technical index before and after the fixes.
Result of a controlled demonstration. The score measures technical readiness and is not a promise of revenue or traffic.
After the first analysis, Goppa helps you prepare your content and track how you show up in AI answers.
What stops AI from reading your site?
Are your products clear and searchable?
Which questions does your brand show up in?
Is your presence in AI answers improving?
They have different owners and different rules, and only one of them is the one people actually mean. Knowing which is which is the difference between buying attention and earning a recommendation.
The sentence that names a shop when someone asks what to buy. It is assembled from pages the model opens while answering. Nobody sells a place in it — not us, not anyone. What moves it is being on the pages it reads, which is the list above.
Product cards shown inside the chat. It is an invitation-only programme and, for now, United States only. A clean, structured catalogue is what makes a store eligible when it opens — but it is not a switch anyone can flip for you today.
Self-serve advertising inside ChatGPT, now open in Europe. It is a separate system from the answer above it: buying an ad does not make the assistant recommend you. Useful for reach, useless for the sentence people were asking about.
What is available depends on your technology. Here is what works today.
Everything, plus the free plugin. It submits changed product URLs through IndexNow so participating search engines can discover updates. Submission does not guarantee indexing or an AI recommendation. It also publishes the agent protocols on your own domain and reports what AI agents did on your site.
Get the WordPress plugin →Diagnosis, measurement and catalogue fixes. Shopify already serves the agent protocols, but every Shopify store shares the same generic llms.txt, so your product data is the only place you can stand out.
Diagnosis, measurement and catalogue fixes. Nothing is served natively here, so there is usually the most to gain.
The report shows what was checked, how each item affects the score and what needs fixing.
No. Goppa only reads public information and changes nothing.
Not for the diagnosis. On WooCommerce, the plugin helps with the setup; on other platforms, the report tells you what to change.
The free diagnosis captures your site as it is today. Paid plans track the changes and keep the history.
Everything works, except the plugin, and you do not need it: Shopify already serves the agent protocols. What Shopify does not do is tell you whether agents recommend you, or fix your product data. Every Shopify store publishes the same generic llms.txt, so your product data is the only place you can stand out.
Enter the URL to get the score, the issues found and the fixes that matter most.
The other half of the answer
This says AI can read you.You see whether you are named, and which pages the AI read instead of yours.
The diagnosis is free and always will be. What you pay for is the question being asked again next week, on every AI, with the answer written down.
Once is not a measurement. Assistants answer the same question with almost the same words and a different set of sources each time — so a single check is a snapshot of something that keeps moving.
