What is order count?
Order count is the number of distinct orders in the selected period that count as
sales — cancelled orders, and orders refunded in full, excluded, the same
is_net_sale rule net revenue uses. An order with a refund the shop recorded but
no matching credit slip for is left out too, until the connection doctor's
refund-gap warning clears. It is the volume companion to revenue: together they
give average order value, and their trends diverging is usually the first sign the
business is changing shape.
Formula
order count = COUNT(DISTINCT order_key) WHERE is_net_sale
Worked example
A month shows €110,000 of net revenue on 1,375 orders — an €80 average order. Next month revenue holds at €110,000 but orders fall to 1,100: the average order rose to €100. Flat revenue hid a real shift — fewer, larger baskets — which changes the right moves on shipping thresholds, bundles and ad targeting. Neither number alone would have shown it.
How Saldo Metrics computes it
The widget counts distinct order_key from v_orders over the selected period,
excluding cancelled orders and orders refunded in full. An explicit status filter
narrows that population further rather than replacing it, so it cannot bring
cancelled or fully refunded orders back in — filtering on "refunded" counts only
partially refunded orders that still count as sales. With a brand filter active it
counts through v_order_lines instead — an order qualifies if any of its lines is
that brand — and DISTINCT keeps a five-line order from counting five times.
Store, country and date filters apply the same way as on net revenue, so the two
KPIs always describe the same set of orders.
Why it matters
Order count is the metric least distorted by price and mix: revenue moves with pricing, discounts and FX, but an order is an order. It is the cleaner read on demand volume, the denominator behind average order value and per-order margin, and the number to check first when revenue moves — did we sell to fewer people, or for less each?
Common mistakes
- Counting lines instead of orders. A basket with five products is one order; brand-filtered counts go through lines but still count distinct orders.
- Comparing order counts across different filter states. A brand-filtered count includes mixed baskets that contain the brand, so per-brand counts can sum to more than the total.
- Treating order count as customer count. Repeat buyers place multiple orders; the customers metrics separate the two.
Where you see this in the app
The Orders KPI and trend on the dashboard.