Average order value is sensitive to large orders, while the median shows the middle of an ordered set. For a small report, it is useful to see both values and the number of purchases. Neither of them on its own shows profit or the result of promotion.
Conditional set
Five confirmed purchases amounted to 400, 500, 600, 700 and 7 800 rubles. The sum is 10 000, the average is 2 000, the median is 600. All values are for illustration. The average is calculated correctly, although four of the five purchases are less than it.
If the usual purchase size is explained with a single number 2 000, it is easy to create a false expectation. But removing a large order without reason is also not acceptable: it really does affect the total amount. Show both values and separately note the concentration of the amount in one purchase.
The same sampling rules
Before calculating, define the period, currency, purchase status and handling of refunds. Do not mix amounts in rubles and another currency as if they were the same numbers. Do not compare the average across all created orders with the median of only paid ones.
In the NIST reference, the average is defined through the sum and the count, and the median through the middle of a sorted set; for an even number of observations, the average of the two central ones is taken. Keep the anonymized amounts themselves so that the calculation can be repeated.
What to check when it changes
If the average has grown while the median has barely changed, check whether one unusually large order has appeared. If both have grown, check the composition of the products, the prices and the conditions of the period. These are directions for checking, not proof of a cause. A conclusion about profitability additionally requires costs; for evaluating a campaign, correct attribution is needed. Do not carry the average order value of a small sample over to future buyers as a promise.
Diagram: 1 — five comparable amounts; 2 — average 2 000 and median 600; 3 — a large order explains the difference. The numbers are for illustration.
Related task: Sales grew after the promotion: how to check whether more money is left.
Sources
NIST: average and median. Verified on 21 September 2026. Practical examples and diagrams are editorial methodology; the illustrative numbers are not client data.
For a separate ready-made public post, you can check view terms. This service does not perform the work described and does not guarantee organic interest, inquiries, or sales.