The chat on two stores: how we counted, and how to set it up
For the analyst who wants to know how chat users were compared with everyone else on two stores, and what the chat was set up to do.
- Every definition behind the numbers on the case study
- The setup in the store, step by step
- What to check in your own data
Open the method and setup notes
Enter your work email and the notes open on this page. We also send you the link, so you can come back to it.
Thanks. The notes are open below.
We use your email to know who reads our research. Privacy policy
What was measured
Period and population
Two European stores: a kids' fashion store over 35 days and a DIY and home store over 30 days, in mid-2026. On each store, one window for both groups: visitors who used the chat and visitors who did not.
What counts as a result
Buyer rate, average order and revenue per visitor for chat users against non-users. Kids' fashion: 6.48% against 0.68% (9.5×), average order 197 against 100 (2×), revenue per visitor 21×. DIY: 5.08% against 0.66% (7.7×), 275 against 100 (2.7×), 49×. On the DIY store, orders placed after an agent recommendation averaged 3.9 times the site's average order.
How an order is credited
A visitor who opened the chat in the window is a chat user, whatever they asked. Their orders are compared with everyone else's. For attribution, revenue was credited to the chat only when the shopper bought after it; on the DIY store that was 7.5% of all the revenue Releva was credited with across channels in 30 days.
Controlled or observed
Not a controlled test. Shoppers who open a chat are often closer to buying, so part of the gap is intent. Each store is compared within the same window, so promotions affect both groups.
What the numbers do not say
They do not say what share of all visitors used the chat, and they do not measure lift. The kids' store funnel is the one place the page counts sessions: 1,654 sessions, 3,267 questions, 2,177 answers with products or advice, 921 product clicks, 5.2% of sessions ending in an order, 29% of advised shoppers adding to cart, 1 in 15 buying.
The terms on the case study
| Term | How it is defined here |
|---|---|
| Chat user | A visitor who opened the chat at least once in the window. |
| Buyer rate | Buyers divided by visitors in the group: 6.48% against 0.68% on kids' fashion, 5.08% against 0.66% on DIY. |
| Average order, indexed | The average order in the group, with non-users set to 100: 197 and 275. |
| Advised shopper | A chat user who received an answer with products or advice. Every one viewed a product; 29% added to cart; 1 in 15 bought. |
| Order after a recommendation | A DIY store order placed after the agent recommended a product in the chat: 3.9 times the site's average order. |
| Conservative attribution | Revenue credited to the chat only when the shopper bought after it: 7.5% of Releva-credited revenue on the DIY store in 30 days. |
What was switched on, in order
The live catalog connected
Products, stock and sizes as they are now, so an answer only proposes what the store can sell.
Understands the need
Answers product and how-to questions in natural language, in the shopper's own words.
Recommends what goes together
Proposes specific products and builds the basket, the way a good in-store assistant would.
The same memory
Runs on the same profile as search, recommendations and email, so each conversation sharpens the rest.
The read-out
One window per store, chat users against everyone else, and the session funnel from question to order.
To check it in your own store
Split one month of visitors into those who opened the chat and those who did not, and compare buyer rate and average order. Then follow the chat sessions: questions, answers with products, product clicks, carts, orders.
Credit the chat only with orders placed after a conversation, and compare that with the revenue your other channels claim. If the chat's share is small and the advised buyers' orders are large, the chat is doing what an in-store assistant does.
See which of last week's visitors you missed
Thirty minutes, with your store and ad accounts open. Then 30 days free. A holdout group decides: if the agent doesn't add orders in 30 days, you don't pay.
Book 30 minutes
You'll talk to one of us.