Level 3 · AOV and units per order

Revenue is orders times AOV

Average order value (AOV) is net sales divided by the number of orders. Units per order (UPO) is units sold divided by orders. With orders they form a small identity that every e-commerce report rests on:

net sales = orders × AOV
AOV       = units per order × average price per unit

When revenue moves, this tells you where to look: more orders, bigger baskets, or pricier items.

Counting orders, not lines

The data has one row per order line, about 2.5 lines per order. Any count over lines counts lines, so orders have to be distinct order ids:

lines.groupby("year")["order_id"].nunique()   # 2,972 · 3,772 · 4,409
lines.groupby("year").size()                  # line counts, two and a half times bigger

Units are a plain sum of quantity. Both AOV and UPO use the same order count as their denominator, so both are computed after the groupby, from the totals.

What the numbers say

Harbor & Pine’s yearly picture:

year  orders  net_sales     aov     units_per_order
2023  2,972   1,649,804.36  555.12  5.20
2024  3,772   2,046,520.24  542.56  5.03
2025  4,409   2,362,232.24  535.78  4.98

Net sales grew 43% over two years while AOV fell 3.5%. Every bit of growth came from order volume; the baskets got a little smaller. That is not automatically bad, since more customers buying is good news. But a forecast that assumes AOV stays at 2023’s $555 overshoots.

AOV is a mean

AOV inherits every weakness of the mean (see the Statistics track): a handful of business orders worth thousands pull it up, and the median order is about half of it. Keep AOV for planning (it multiplies back into revenue) and use the median for “what does a typical customer spend”.

Common mistakes

• size() instead of nunique(). AOV comes out per line, not per order.
• Averaging line values. lines["net"].mean() is the average line, not the average order.
• Different order sets. Net sales on booked orders divided by all placed orders mixes two definitions and understates AOV.

Interview note

“Revenue is up 15% but AOV is down. What happened?” Answer with the identity: orders must have grown faster than 15%. Then ask which customers drove the extra orders.