What every Analytics number means

Sum of paid orders (Paid or Paid — Manual Review) in the selected range.

Conversations with any activity in the selected range — not just ones that started in it. A customer who first messaged weeks ago and comes back to order today still counts here, since their conversation is still active.

Count of orders in the selected range with status Paid or Paid — Manual Review.

Revenue ÷ Paid orders, for the selected range.

Paid orders ÷ Chats. Shows “—” instead of “0%” when there are zero active chats in range, since a true 0% would wrongly imply nothing converted.

Count of return requests filed in the selected range — not necessarily for the same orders as the Paid Orders count next to it. A return this week can be for an order placed weeks ago, so the “% of paid orders” figure can legitimately go over 100%.

Compares the second half of the selected range to the first half of the same range — a free signal computed from data already on the page, not a true previous-period comparison. Read it as “is this accelerating or slowing down within this window,” not “vs last month.”

Daily revenue for the range, paid orders only. Hover any point for the exact date and amount.

The single highest-revenue day in the selected range, marked on the chart.

Same Chats definition as the KPI card above, shown day by day.

Paid orders, day by day.

Total revenue for each weekday (Sun–Sat) across the whole selected range — which days actually convert.

Best sellers by revenue, paid orders only, top 8 in the selected range.

Catalog items with the fewest (or zero) units sold in the selected range. We don't track product views, so this is the honest stand-in for “lowest converting product” — it tells you what isn't selling, not why.

Conversations that reached a fully completed state (payment confirmed, or a return handled) without ever landing in manual review, in the selected range. This is a real, measured count.

A calculator, not a measurement — it multiplies the real Auto-resolved count above by two numbers you set yourself (minutes per manual reply, hourly staff cost). Change the assumptions to match your team; they're saved in your browser only.

Three stages: DMs received (same as Chats), Checkout started (an order record was created — the closest available proxy for “reached checkout,” since cart-add events aren't logged historically), and Paid.

Every order created in the selected range, grouped by its current status.

Average time between a customer's first message and an order being created for them.

Average time between a return being requested and you approving or rejecting it.

% of conversations in range with items sitting in the cart that never reached a finished state.

What hour of the day (0–23) chats most often start — useful for knowing when to be online.

A simple run-rate for the current calendar month: (month-to-date revenue ÷ days elapsed) × days in the month. It's today's pace held steady, not a predictive model — and unlike the rest of the page, it's always based on the current month, not whatever range you have selected.

Based on a customer's entire order history, not just the selected range — this is intentional, since it wouldn't make sense to call someone “Lost” using only a 7-day window. Refreshed once a day, not live.

3+ orders, and total spend above the average paying customer.

2+ orders, most recent one within the last 60 days.

2+ orders, but nothing in the last 60+ days.

Nothing ordered in 120+ days.

Exactly one order, placed in the last 30 days.

Everyone else — a single order, more than 30 days ago.

A customer gets exactly one badge. Order of priority: Lost → At Risk → VIP → Loyal → New → Regular. A big spender who's gone quiet shows as Lost or At Risk, not VIP — that's deliberate, since a valuable customer going quiet is the more actionable thing to see.