02
Leave a group alone
A control group stays untouched. While the experiment runs it is paused from publishing, and its product pages in Clione say why.
Features · Measure
AI attribution that tells you when it's only noise.
Connect Google Analytics and Clione matches your orders to the visits that produced them, then shows which came from AI assistants. Every comparison carries the range that says when a difference may be noise, and what can't be measured is named, never shown as zero.
The orders
AI-referred orders, beside the rest.
Clione matches each order to the visit that produced it and says which came from an AI assistant. The Attribution page sets those orders against every other order in the store.
- Orders referred by AI assistants against the rest
- The range on each figure that shows when a difference may be noise
- The AI sources that can't be measured, named as such instead of counted as zero
Enriched products
Did the enriched products sell more?
The same page compares sales of the products you enriched with semantic enrichment against the products you didn't, with the same range on every figure.
- Enriched against not enriched, side by side
- A difference inside the range is shown as what it may be: noise
In your reports
The same figures, under your logo.
Put the attribution section into a branded report over one store or many, in your organization's logo and colors.
- One store or a whole book of clients, over the date range you choose
- Shared by a private link, or embedded
Experiments
Then test it on your own products.
A controlled experiment on your own products, run with our team.
01
Change a group
Groups of products get a changed product page, a changed feed, or both.
How the feeds work →03
Read the result
A report per group says whether AI assistants mentioned and cited the changed products more than the control.
The data
Orders, not shoppers.
Clione keeps a read-only copy of the last 30 days of orders: what was bought, when and for how much. No name, email, address or customer id, so no shopper data enters the system, and it is stored in the EU.
Before you connect
What people ask about AI attribution.
What is AI attribution?
Working out which orders started with a visit from an AI assistant. Clione does it by matching your orders to the Google Analytics visits that produced them. The glossary has the long version of LLM attribution, and of AI referral traffic.
Do I need Google Analytics?
Yes. Attribution reads Google Analytics to match orders to visits, so it starts once Analytics is connected to the store.
Why are some AI sources marked as unmeasured?
Because some can't be measured from the data Clione reads, and a zero would say they sent you nothing. Clione names them as unmeasured instead, so a gap in the data never reads as a gap in sales.
Will it tell me how much Clione earned me?
No. It shows which orders came from AI assistants and how enriched products sold against the rest, each with its noise range. Turning that into a return on investment would claim a cause the figures can't prove.
Can I run an experiment myself?
Not on your own: experiments are run with our team. The groups are set up with you, and the report comes back per group.
What does it cost?
Nothing, during the beta. Clione is taking selected clients into this stage, in limited places, free of charge, and prices are published when the beta ends in Q1 2027. Attribution is not on every plan.
Keep reading
Where it goes from here.
Measure
AI visibility
Whether ChatGPT and Gemini recommend your products, and who takes your place when they don't.
Send
Product feeds
The enriched catalog to ChatGPT shopping every day, and a supplemental feed to Google Merchant Center.
Check
Verify
What the live page actually serves, read back after every push.
Start
See which orders AI sent you.
On a short call: your store, and how Clione sets its AI-referred orders against the rest.
Clione is in beta: selected clients, limited places, free until the beta ends in Q1 2027.