How to Measure Revenue From a Store Chat: Attribution Guide

To measure revenue linked to a store chat, log each chat session with an ID, and match it to orders in a fixed time window, such as the same session or a set number of days. Report two numbers: direct revenue (the order came right after the chat) and assisted revenue (the chat was one step before). Treat both as linked revenue, not as proof of cause. Only a controlled test can show cause.

Why chat revenue is hard to measure

A customer who opens a chat is often already interested in buying. They have a question before they pay. So chat users may convert at a higher rate even if the chat did nothing to help.

This is called selection bias. If you compare chat users with all visitors, the chat looks very strong. Part of that gap is simply that chat users were more keen from the start.

Keep this in mind for every number below. Linked revenue is not the same as caused revenue.

Step 1: Decide what counts as a chat touch

Write down a clear rule before you look at any data. Options include:

• A visitor opened the chat widget.
• A visitor sent at least one message.
• A visitor got at least one answer from the AI.

The second or third rule is stronger. Opening the widget by accident is not a real touch. Pick one rule and keep it fixed, so that your numbers can be compared over time.

Step 2: Log the data you need

You need three things in your records:

1. A chat session ID, with a start time and the number of messages.
2. A link between the chat session and the browser session or customer, using a first-party method that respects consent.
3. The order ID, order time and order value, from your shop.

Store the minimum. Do not store more personal data than you need. If you use cookies or similar tools to link a chat to a later order, check the consent rules for your country with a qualified adviser.

Step 3: Choose an attribution window

The attribution window is the time between the chat and the order that you still count. A short window is strict. A long window counts more, but with less certainty.

Common choices:

• Same session: the order happens in the same visit.
• Short window: for example 24 hours or 7 days after the chat.
• Longer window: for example 30 days, for products with a long decision time.

Match the window to your buying cycle. A low-price accessory needs a shorter window than a mattress or a bike. Analytics tools such as Google Analytics 4 have their own lookback window settings for key events, so check the settings there and make sure your chat reports use the same logic.

Do not change the window after you see the results.

Step 4: Split direct and assisted revenue

Direct revenue

Direct revenue is the value of orders that came in the same session as the chat, or right after it. The link is strong, but it is still not proof of cause.

Assisted revenue

Assisted revenue is the value of orders where the chat was one touch among several. For example, a customer chats on Monday, comes back from an email on Thursday and buys. The email might get the last-click credit, and the chat is an assist.

Report these two numbers apart. Do not add them to one big figure without saying what each one means.

Step 5: Use net revenue

Raw order value overstates the result. Adjust the numbers:

• Subtract refunds and returns in the same period.
• Remove cancelled orders.
• Be clear about whether you count prices with or without tax and shipping.
• Count each order once, even if the customer chatted several times.

Step 6: Test for cause

If you want to know what the chat really adds, run a controlled test.

1. Randomly show the chat to part of your visitors and hide it from the rest.
2. Run the test long enough to cover your normal weekly cycle.
3. Compare conversion rate and revenue per visitor between the two groups.
4. Check that the groups are similar in traffic source and device.

This is the cleanest method. It needs enough traffic to show a real difference, so small stores may not get a clear answer. In that case, say so and treat your numbers as a rough guide.

Common pitfalls

• Counting chat sessions that began after the customer had already put items in the cart and was about to pay.
• Double counting the same order in the chat report and in your ad reports.
• Ignoring returns, which can remove a large part of the linked revenue.
• Comparing chat users with all visitors and calling the gap a gain.
• Changing the window or the rule after you see the data.
• Judging the chat only by revenue. Also look at support load, answer quality and customer feedback.

Frequently asked questions

What is a good attribution window for a store chat?

It depends on how long your customers take to decide. Same-session and 7-day windows suit low-price items. Longer windows suit expensive items. Pick one in advance and keep it.

Is assisted revenue real revenue?

The orders are real, but the chat's share in them is uncertain. Report assisted revenue as a separate line and describe it as linked, not as caused.

Can I trust the revenue number that my chat tool shows?

Check how it works. Ask what counts as a touch, which window it uses, and whether it removes refunds. If the tool does not say, treat the number as a rough guide.

Do I need consent to measure this?

It depends on what data you collect and how you link it. Rules for cookies and tracking vary. Ask a qualified adviser and read the official text for your country.