Whether you use Claude, ChatGPT or another AI tool, the assistant chooses the right tool for your question, so you don't need to name them. This page is for anyone checking what data is available and how it is counted.
Tools
Tool | Permission | What it returns |
| Store read | Store name, domain, timezone, currency and plan |
| Analytics read | Revenue, orders and Cooee-attributed revenue and orders over a date range |
| Analytics read | Revenue and orders day by day |
| Analytics read | Sessions, visitors, funnel steps, orders, revenue and rates, broken down by landing page, exit page, date, UTM, country, device, traffic source, visitor type and more. Can be split by visitor intent (none, low, medium, high) |
| Analytics read | Visitors shown Cooee compared with visitors held out of it: conversion rate, revenue per visitor, lift, significance and incremental revenue |
| Campaign read | Campaign settings, status, widget type, placement and lifetime totals |
| Campaign read | Impressions, clicks, CTR, dismissals, holdouts, sessions, visitors, add-to-carts, orders and revenue for each campaign over a date range, optionally day by day |
| Campaign read | Experiment settings and, for each variant, visitors, orders, revenue, conversion rate, revenue per visitor, lift over control and significance |
| Product read | Products and their price range |
| Product read | Views, add-to-carts, view-to-cart rate, removals, purchases, units sold and revenue for each product or variant |
How to read the numbers
Rates are percentages: 1.92 means 1.92%. Money is in your store's currency. Dates are in your store's timezone.
Conversion rate in experiments and the impact report is orders per 100 visitors. In the landing page report it is per session.
Visitors are counted once per row, so visitor counts do not add up across rows or days. See Visitor & session counting logic.
Campaign revenue in
get_campaign_reportis revenue from orders where the campaign was the last touch within the dates you choose.get_campaignshows lifetime influenced revenue, which is higher.Product views count page views and variant switches, not unique visitors. A view-to-cart rate above 100% means the product is mostly added from collection pages or widgets.
Significance: a p-value below 0.05 means the difference in conversion is unlikely to be chance. Revenue per visitor has no significance test.
Incremental revenue is the extra revenue per visitor in the group shown Cooee, multiplied by the number of visitors in that group. The groups are different sizes, so compare them per visitor, never by totals.
Limits
Campaign, product and visitor-behaviour data starts on 17 June 2026.
Campaign, product and impact reports cover up to 366 days per question, or 92 days for a daily campaign breakdown. A very long landing page report on a large store may time out, so ask for a shorter range.
Visitor and session counts in the campaign report are accurate to about 1%.
UTM values are matched exactly as tagged, so "Meta" and "meta" are separate. Ask by traffic source or platform to group them.
Not available yet: Meta ad spend, customer segments, which products a campaign sold, and products bought together.
Numbers may differ slightly from Shopify. See Why session and visitor counts differ between Cooee and Shopify.