Post-sale revenue

How to calculate customer purchase frequency

Purchase frequency starts with the gaps between valid orders. The aim is not to force every account into one alert rule, but to establish a defensible reference for that customer.

Map My Purchase Frequency Back to the Pain

Frequency read

Orders

Valid orders only

Gaps

Orders minus one

Average

Reference per account

Recency

Days since the last order

01Method

Calculate intervals, not order counts

Sort one customer’s valid orders from oldest to newest. With at least two orders, the average interval is the time from the first order to the last, divided by the number of gaps between them.

Average interval = (last order date − first order date) / (number of orders − 1)

The subtraction by one matters: four orders create three intervals. With only one order there is no interval to measure, so do not present a frequency as if it were observed behavior.

02Worked example

Four orders, three intervals

Consider orders placed on 10/01, 09/02, 10/03, and 12/04 of the same non-leap year. The dates are written as day/month.

From	To	Interval
10/01	09/02	30 days
09/02	10/03	29 days
10/03	12/04	33 days

The interval from the first to the last order is 92 days. Therefore: 92 / (4 − 1) = 30.7 days. Summing the individual intervals gives the same check: (30 + 29 + 33) / 3 = 30.7 days.

03Data discipline

Keep the calculation traceable

At minimum, keep these columns in an export or table:

Customer ID or customer name, to group orders correctly.
Order ID, to preserve one record per order and allow checking.
Order date, used to sort the sequence and calculate the gaps.

A cancellation or status field helps exclude orders that should not represent demand. Product line can also be useful when a customer buys distinct categories on different cycles. Compare like with like before combining them.

Measure recency separately

Recency is today’s date minus the date of the last valid order. It answers how long the account has been quiet; the average interval describes its historical rhythm. They are related, but they are not the same measure.

04Interpretation

Treat variation as information

An average is useful when intervals are reasonably stable. When they swing, inspect the median and the spread of the intervals as well. The median describes a typical gap with less influence from one unusual order; the range or another variability measure shows how much the cadence actually moves.

Separate seasonality before judging a delay. A customer that routinely pauses in a planned shutdown or orders around an annual budget cycle needs a seasonal reference, not a single all-year average. A conversation or alert should use that account’s history, commercial context, and the reliability of the data. There is no universal number of days that proves risk.

05FAQ

Frequently asked questions

How do I calculate a customer's purchase frequency? +

Sort the account's orders by date, calculate the interval in days between each order and the next, and look at the typical interval. That interval is the reference for when the next order should happen.

Should I use the mean or the median of the intervals? +

The median is usually safer when there are unusual orders, such as an emergency purchase or a long pause. The mean works when intervals are stable. Show both if in doubt.

How many orders are needed to calculate frequency? +

You need at least two orders to have one interval, and the reading becomes more reliable as the history grows. With few orders, treat the result as a hypothesis rather than a pattern.

Do orders on the same day count as separate purchases? +

Usually not. Orders issued on the same day or in very close sequence are often the same buying decision split for tax or logistics reasons. Group them before calculating intervals.

How should I handle seasonal buyers? +

Compare the current period with the same period in previous cycles instead of using a single interval. A seasonal account can go months without an order and still be within its pattern.

Should frequency be calculated per account or per product? +

Start per account, to know whether the relationship is active. Then calculate per product line, because a customer may keep buying while having stopped precisely the highest-value line.

Why does the calculation need to be traceable? +

Because whoever acts on it needs to check where the figure came from. Keep the extract, the cleaning rules and the calculation sheet, so the list of overdue accounts can be reviewed and challenged.

What if the intervals vary a lot? +

Treat the variation as information about the account. Irregular intervals may indicate project-based buying, a partial supplier change or customer inventory, and call for a conversation rather than an automatic alert.

Want frequency calculated account by account?

The diagnosis shows where the portfolio lost pace and which sales routine brings revenue back.

Get a Frequency DiagnosisFree diagnosis, no commitment