What is a good average order value?

Average order value on its own is a vanity figure. It becomes a decision when you put two other numbers beside it.

The first is what it costs to fulfil an order regardless of size. Picking, packing, the box, the label, the payment fee, the customer service minute. That cost is roughly the same whether the order is 20 dollars or 90, so it eats a much larger share of a small order. Once you know it, you can see which orders are barely worth shipping.

The second is what it costs to acquire an order. If ads cost you a fixed amount per sale, the average order decides whether that spend leaves anything behind. Two stores with the same acquisition cost and different averages are in completely different businesses.

Put all three together and the question stops being what is a good average. It becomes how much room this store has between what an order is worth and what it costs to win and fulfil. That gap is the room you have to bid, to offer free delivery, to accept returns gracefully and to survive a bad month.

That is also why raising the average is often the fastest lever available. The visitors are already there and already buying. Adding value to an order you were going to receive anyway costs almost nothing, while every extra dollar of order value widens the gap that everything else depends on.

One trade order or a single large bundle can lift a monthly average without anything changing for normal customers. The average is sensitive to extremes, which makes it a poor description of a typical order.

Work out the median as well, the order in the middle when they are lined up by size. When the average sits well above the median, a small number of large orders is carrying it, and the typical customer experience is smaller than your reporting suggests.

Better still, look at the shape. Group orders into bands and count how many fall into each. Most catalogues produce a lump at the single item price and a second, smaller lump wherever a bundle or a shipping threshold sits.

That picture tells you where to act. A big lump just below the free shipping line means the line is doing something, but people are stopping short. A single lump at one item price with nothing above it means there is no natural second purchase, so bundles have to create one.

It also protects you from false progress. A campaign that brings a handful of wholesale style orders will lift the average and teach you nothing about the store, while a change that moves five hundred ordinary orders up by a few dollars will barely move it and be worth far more.

A blended average moves whenever your traffic mix moves, which makes it a poor early warning system.

By channel first. Email and returning traffic usually produce different orders than cold paid traffic, because the people know the range and trust the store. If you shift budget between channels, the blended average changes without a single customer behaving differently. Watching it per channel keeps the signal clean, and it also tells you which channel deserves the next dollar, because a channel with a lower acquisition cost and a lower order value can still be the weaker one.

By customer type second. First orders and repeat orders behave differently. First orders are cautious, often a single item, often the cheapest way into the range. Repeat orders are larger because trust is already paid for. If your repeat share is growing, the average will rise on its own, and you should know that is the reason.

By device third, where it often surprises people. Phone orders can run smaller not because phone buyers spend less but because comparing options is harder on a small screen, which is a design problem rather than a customer preference.

And by season. Categories with a gift period see a step change in both order size and item count for a few weeks. Compare like with like, so this year's November against last year's November rather than against October.

The average can be raised in ways that lose money, so pair it with margin per order every time you report it.

A sitewide discount that pushes people over a threshold raises the average and can cut the money you keep. A multi buy offer priced too generously does the same. So does free shipping set at a level so low that most orders would have paid for it anyway.

The test is simple. Take gross profit for the period and divide it by the number of orders. If the average went up and that figure went up too, the change was real. If the average went up and profit per order went down, you bought the improvement.

Watch returns as well. Bundles and multi packs can lift the average and then give it back through returns if the extra items were not really wanted. Measure the average after refunds, not before.

And watch order frequency. A subscription or replenishment change that raises the size of each order but stretches the gap between them may leave annual revenue per customer flat. The honest measure there is revenue per customer per year rather than per order.

None of this makes the average a bad metric. It makes it a metric that needs a companion. Report the two together and it becomes hard to fool yourself.