Analyze average order value

I need an AOV calculation for the last 90 days segmented by acquisition channel and device. Produce a table that shows orders, total revenue, and average order value for Organic, Paid Search, Email, and Social, split by Desktop and Mobile. Include confidence notes if sample sizes are below 50 orders.

Before I present average order value to the growth team, make the metric actionable: put AOV at the top, show the median order value and the distribution deciles, flag channels where AOV diverges from median by more than 25%, and list the top three product categories that lift AOV for each channel. Note where high AOV is coming from one-off bulk buyers.

Paid social AOV fell by about 30% since the last period. The paid media manager says targeting hasn’t changed; the merch lead says no new promos. I don’t know if the problem is low-value customers, a tracking issue, or misattributed refunds. What is the most likely explanation and what two checks will quickly tell me whether it’s a real change or a reporting/attribution problem?

Every month we argue about which orders to include in AOV — discounted samples, refunded partials, and multi-shipment orders— and that debate delays reporting and decisions. I need one change to the definition and a short rule set to stop the debate while keeping the metric useful for marketing. What simple, defensible AOV definition should we adopt and what guardrail ensures it stays consistent?