How does GEO get your products named in AI answers?
Autobasket Discover is generative engine optimization for online shops. We prepare the pages and product data that ChatGPT, Perplexity, Claude, Gemini and Google AI Overviews read, then measure every month how often each one names you against your competitors. ChatGPT ranks sellers on availability, price, quality and who makes the product, so those fields come first.
How does an assistant pick the pages it names in an answer?
Before an assistant names a brand, it searches, fetches pages, cuts them into passages, scores each passage against the question and keeps only the best few to read. A page that drops out at any step is never cited, however good the product is, so Discover checks and fixes each step in order.
Search
The assistant decides the question needs fresh results and runs its own queries.
Fetch
Its crawler requests your page, if
robots.txtlets it in.Split
The page is cut into passages of a few hundred words.
Rank
Each passage is turned into a vector, a list of numbers, and scored for how well it matches the question.
Place
The top passages enter the context window, the limited text the model reads.
Cite
The answer names some sources, and their order shapes the click.
| What was measured | Source, date and scope |
|---|---|
| ChatGPT reads structured metadata such as price and product description, and ranks merchants selling the same item on availability, price, quality and whether they are the maker or primary seller. | OpenAI Help Center, Shopping with ChatGPT Search, fetched 16 October 2026. ChatGPT shopping results |
| On Shopify, sessions referred by AI chats convert at nearly 50% higher rates than organic search, and AI-referred orders grew nearly 13 times year on year in the first quarter of 2026. | Shopify, 11 May 2026. Shopify merchants, product page sessions |
| 76.1% of the pages cited in Google AI Overviews also rank in the top 10 organic results for the query. | Ahrefs, study of 1.9 million citations from 1 million AI Overviews, 21 July 2025, updated 2 July 2026. Google AI Overviews |
| A chocolatier near Geneva was named in 58% of Google AI Overviews and 33% of Perplexity answers we tested, yet none of its 26 rendered product pages declared a price an agent could read. | Autobasket audit, July 2026. Switzerland, one shop, 24 prompts per engine |
How does Discover give assistants a clean copy of your pages?
Discover gives assistants a plain-text and data version of your site, built from your existing pages and kept in step with
them, so what they read about your products matches what customers see. It has three parts: an llms.txt index,
a Markdown copy of each page, and JSON-LD @graph data linking your company, products, prices and policies.
- What an agent reads matches what a person reads, because every file is rebuilt from the rendered page at each deployment, including the full-text
llms-full.txtand the HTTP headers that point agents to each Markdown copy - Wrong prices are caught before an agent repeats them, because the product data is validated after every catalogue change
- The eight search and shopping agents we track are allowed into your pages, because we review what your crawler rules let in
- Google is served by the same structured data, since it states that AI Overviews need no AI text files, so the
llms.txtand Markdown layer is for assistants that fetch pages directly
This site runs the same setup. Open our llms.txt or this page as Markdown, and read what each layer does in llms.txt, Markdown and schema for online shops.
# Autobasket
> Autobasket prepares online shops to be found,
> understood and bought by AI shopping assistants.
## Solutions
- [Autobasket Discover](/products/autobasket-discover.md)
- [Autobasket Convert](/products/autobasket-convert.md)
## Learn
- [How do you get your products recommended
by ChatGPT?](/learn/how-ai-assistants-...)How often do AI assistants name your brand, and how is it measured?
We ask ChatGPT, Perplexity, Claude and Gemini your customers' questions many times over and record who is named, where and from which sources. The result is your share of AI answers, often called share of model: the share of answers that name your brand, with a margin of error, because one run of an assistant proves very little.
- Presence
- Share of answers that name you, with the range it most likely sits in (95% interval)
- Position
- Where you appear when you are named
- Competitors
- Who is named instead of you, and how often
- Sources
- The pages the assistant cited, so you know which pages to improve first
- Accuracy
- Whether the price and facts stated about you are correct
What does each Discover retainer include?
Discover runs as a monthly retainer in three tiers, set by the customer questions tracked, the assistants monitored and how often content is updated. Every tier keeps the assistant-readable copy of your site current, since the way assistants pick sources keeps changing. Prices follow a scoping call, as they depend on catalogue size and languages.
| What is included | Growth | Enterprise | Global Scale |
|---|---|---|---|
| Customer questions tracked | Up to 250 a month | Up to 1,200 a month | Custom prompt sets, no fixed cap |
| Assistants monitored | ChatGPT, Perplexity | ChatGPT, Perplexity, Claude, Gemini | All four, plus models you name |
| Assistant-readable copy of your site | llms.txt | llms.txt and Markdown mirrors | Custom endpoints served from the edge |
| Content and product data updates | Quarterly | Monthly | As the catalogue changes |
| Report on how often you are named | Monthly export | Dashboard, updated every two weeks | API and webhook feed |
What do teams ask before starting Discover?
These are the questions marketing and SEO leads ask before a Discover retainer, answered in full. They cover what the work changes, what nobody can promise about AI answers, which assistants we measure and how long a change takes to show, so you can judge the retainer before the first call.
What is generative engine optimization?
Generative Engine Optimization, or GEO, is the work that gets your pages found, quoted and named by AI assistants such as ChatGPT, Perplexity, Claude and Google AI Overviews. It starts with letting their crawlers in and publishing structured data, and continues with writing each section so the passage an assistant picks up already carries the answer. Discover is our GEO service for shops.
Why does ChatGPT recommend a competitor and not my shop?
Usually because the competitor's offer is easier to read and confirm. OpenAI states that ChatGPT uses structured metadata such as price and description, and ranks merchants selling the same item on availability, price, quality and whether they are the maker or primary seller. A shop whose price sits only in the page text, as on 117 of the 211 Romandie shops we audited in July 2026, gives it less to confirm.
How do I get my products into ChatGPT shopping results?
According to OpenAI, Shopify merchants are included through Shopify Catalog without extra work, and other merchants can apply to send a product feed directly. Inclusion is only the start: product results are not ads and are ranked on relevance, availability and price, so Discover makes sure the feed, the structured data and the visible page say the same thing and answer the questions shoppers ask.
Is GEO different from SEO?
They share a foundation. Google's AI Overviews cite pages that are indexed, and Ahrefs found that 76.1% of cited pages also rank in the organic top 10. GEO adds the passage level, because an assistant quotes a few hundred words rather than ranking a whole page, so each section has to carry its own answer, figure and source. Our guide on GEO vs SEO for online shops sets out the differences.
Can you guarantee a ranking in AI answers?
No. Answers vary from one run to the next and no vendor controls them. We remove the reasons a page is dropped and measure the share of answers that name you, with confidence intervals, so you can see whether it moves over several months.
Which assistants does Discover measure?
ChatGPT and Perplexity on the Growth tier, and ChatGPT, Perplexity, Claude and Gemini on Enterprise, as listed in the retainer table above. Global Scale adds the models you name.
How long before assistants name you more often?
It depends on how often assistants revisit your pages and how much other sites write about your category. We measure a starting point in the visibility audit and report against it every month, which is how a real change is told apart from the normal variation between runs.
Start with an audit of how AI answers treat your category
The audit measures how often assistants name you on the questions your customers ask today and names the pages that hold you back, so the retainer starts from a baseline you can check against every month.