Customer Profitability Calculator

The maths behind it

Customer profitability is the contribution a customer generates less the costs attributable to serving them: returns, support contacts, discounts taken, and their share of acquisition.

Profit = (orders × contribution per order) − returns cost − support cost − acquisition cost.

Calculated per customer and ranked, it produces a distribution that is nearly always more extreme than expected.

The same thing with real figures

A customer placing 4 orders at $58 generates $127.60 of contribution. Returns at 15% on their orders cost $8.60 each on 0.6 returns, $5.16. Three support contacts at $4, $12. Acquisition $27.

Profit is $83.44. A second customer with the same four orders but a 40% return rate and eight support contacts: $127.60 − $13.76 − $32 − $27 = $54.84.

A third with two orders, a 60% return rate and five contacts: $63.80 − $10.32 − $20 − $27 = $6.48.

Same business, three customers, profitability ranging from $6.48 to $83.44, and all three look identical in a revenue-based report.

The catch

Attributing support and returns costs per customer requires data most businesses do not join up, and estimates introduce error that can flip individual rankings.

The distribution is also more informative than any individual figure, since the point is to identify groups rather than to price individuals precisely.

Applying it

Calculate it in bands rather than per customer, since the decisions it supports are about segments and the precision required is correspondingly lower.

Then look at the unprofitable tail and identify what they have in common. It is usually one behaviour: heavy returns, discount-only purchasing, or high support need, rather than a random spread.

What to do about unprofitable customers

The instinct is to remove them and it is rarely the right response, since the causes are usually addressable. Heavy returners frequently respond to better product information; heavy support users to better documentation.

Where the behaviour persists, the response is a policy rather than an exclusion: a restocking fee above a return threshold, or simply ceasing to remarket to the segment.

The most common mistake is spending acquisition budget attracting more of them, which happens by default when campaigns are optimised on orders rather than on contribution. Feeding profitability data back into acquisition targeting is what stops the problem compounding.

Feeding the results back into acquisition targeting is where the analysis pays for itself, since campaigns optimised on orders will otherwise keep buying more of whichever customers are cheapest to acquire.

Excluding lookalike audiences built from unprofitable segments is a small change with a compounding effect.

Reviewing the most profitable decile for what they have in common is as useful as studying the unprofitable tail, since it describes the customer worth acquiring more of.

Estimating support and returns costs at segment level rather than per customer keeps the analysis practical while still identifying the groups that matter.

Reviewing the analysis annually rather than once catches drift, since return rates and support costs move with product mix.

Allocating service cost per customer is what makes this calculation honest and what most implementations skip. Support contacts, returns handling and any account management vary enormously between customers, and in most catalogues a minority of customers consume a disproportionate share. A profitability figure using an average service cost will find every customer roughly equally profitable, which is never true.