Analytics & attribution
What Cartie measures, how revenue is attributed to widgets, and how to read the dashboard.
Cartie measures what happens in your cart and connects it to the orders that follow, so you can tell which widgets earn their place. The Dashboard is the summary; Analytics is the detail.
#What gets recorded
| Event | When it fires |
|---|---|
| Impression | A widget renders in front of a shopper. Deduplicated per page load, so a drawer opening five times is not five impressions. |
| Cart view | The cart drawer or cart page is opened. |
| Add to cart | An item is added — including from a Cartie upsell. |
| Checkout click | The shopper clicks through to checkout. Used for the funnel step. |
| Purchase | An order is actually placed. Taken from Shopify's order webhook rather than the browser, and deduplicated by order id — this, not checkout clicks, is what every conversion and revenue number on the dashboard is built from. |
#How revenue is attributed
When a shopper heads to checkout, Cartie stamps the cart with which widgets and experiment variants they were shown. When the order arrives from Shopify, that stamp is what ties the revenue back to the widgets involved. This is why the numbers survive ad blockers and browser tracking limits better than pixel-based attribution.
#The dashboard
- Revenue over time — order revenue attributed to your Cartie cart.
- Revenue generated by Cartie widgets — a breakdown by widget type.
- Cart performance funnel — cart views → add to cart → checkout → purchase, so you can see where shoppers fall out.
- Top performing widgets — ranked by assisted revenue.
- Active experiments — what is running and how it is doing.
- Key insights — plain-language observations about what changed.
#The analytics page
Impressions, conversions, revenue per visitor and average lift across running experiments, plus the per-widget revenue table and the full experiment list.
#Reading it honestly
- Give it time. A day of data on a store doing 20 orders a day tells you nothing. Compare weeks, not hours.
- Revenue per visitor is usually the metric that matters — it catches the case where conversion rate drops slightly but order value climbs more.
- Order events can lag by a few minutes behind the storefront events; a small gap between a purchase and it appearing here is normal.
- Widget revenue does not sum to total revenue. One order can be assisted by several widgets, so the columns overlap by design.
#Getting the data out
Experiment results are available over the REST API, and conversion events can be POSTed to your own endpoint with the custom webhook in Settings → Integrations.
Still stuck on this?Send us your store domain and what you’re seeing — we’ll take a look.