Customer Profitability for DTC

Orders are the natural unit for a profit calculation and the wrong unit for an acquisition decision. You do not buy an order; you buy a customer, and what that customer does over the following year is what determines whether the purchase was sensible.

What goes into the figure

Customer profitability sums the contribution margin of every order that customer placed, then subtracts what it cost to get and keep them:

• Contribution margin per order — after product, shipping, fees and packaging, ideally return-adjusted.
• Acquisition cost, applied once, at the start.
• Retention spend allocated across active customers — email and SMS platforms, loyalty programme costs, retargeting aimed at existing buyers.
• Their own return behaviour, which varies enormously between customers and is usually ignored.

The last point deserves emphasis. Two customers who spend identically can differ by their entire margin if one of them returns a third of what they order.

Why this differs from lifetime value

Lifetime value, as usually calculated, is a revenue measure: total spend over some window.

It is a useful directional number and it cannot answer the question anyone is really asking, because revenue is not what you keep. A customer buying heavily discounted bulky items at 22% contribution margin and one buying full-price accessories at 55% can post the same LTV and be worth twice as different in profit.

Customer profitability replaces revenue with contribution margin at every step. Same structure, different input, and the ordering of your customer base changes.

The skew, and what it means

Customer value is not normally distributed. In most stores a minority of customers account for a large share of total contribution, and a majority never buy again at all.

That has a direct strategic consequence. If a mean customer contributes $98 but the median contributes $7, then “what is a customer worth” is the wrong question. The right one is “can we acquire more of the customers who repeat” — a targeting, product and onboarding question rather than a bidding one.

Paying up to the mean across all channels systematically overpays for the segments that never return and underpays for the ones that do.

The check: for one cohort, plot cumulative contribution by customer decile. If the top decile carries more than half the total, you are running a skewed book, and average-based bidding is leaving money on the table in both directions.

Segments worth separating

Acquisition channel. Customers acquired through discount-led paid campaigns repeat less and return more than those from organic search. That difference belongs in the channel comparison, and it usually widens the gap already visible in first-order margin.

First product purchased. The entry product predicts repeat behaviour more strongly than almost anything else. Some products acquire customers who come back; some acquire customers who wanted that one thing.

First-order discount depth. Customers acquired at a deep discount frequently return only for the next discount, which is a materially less valuable customer than the headline conversion rate suggests.

Geography, where delivery cost and failure rates differ enough to change margin per order.

Each of these is actionable. The channel and product cuts in particular tend to change acquisition strategy rather than just reporting.

The maturity trap

Any customer profitability figure for a recent cohort is incomplete, because those customers have had less time to buy again.

Comparing “customers acquired this quarter” against “customers acquired two years ago” will always favour the older group, and reading that as declining customer quality is one of the more common analytical errors in ecommerce.

Two defences. Compare like-aged cohorts — the first six months of this year’s customers against the first six months of last year’s. Or use a fixed window for every cohort, accepting that the most recent ones cannot be judged yet.

Unprofitable customers exist

Not every customer is worth having, which is uncomfortable and worth stating.

A customer who buys once at a deep discount, returns half the order, contacts support three times and never orders again has consumed acquisition cost, two shipping labels, packaging, handling and staff time in exchange for a fraction of one order’s margin. In categories with high return rates a meaningful share of customers fall into this pattern.

The response is almost never to refuse them. It is to stop actively buying more of them.

That means looking at which acquisition sources produce this profile — typically deep-discount campaigns and marketplaces with lenient return norms — and adjusting spend rather than adding customer-level restrictions. The one exception worth considering is serial returners, where a small number of accounts can account for a surprising share of return volume, and where most stores eventually adopt some policy.

What makes this decidable rather than anecdotal is having the per-customer figure. Without it, the pattern is invisible: those customers look like everyone else in a revenue report, and they are frequently among the highest spenders.

What to do with it

Three uses, roughly in order of value.

Set the acquisition ceiling properly. If customers reliably contribute beyond the first order, the ceiling is higher than first-order margin — but only by the measured amount, and only if the cash to fund the payback period exists.

Reallocate acquisition toward channels producing repeat buyers. A channel with higher CAC and much better retention can be the better investment, and only a customer-level view shows it.

Invest in the entry product, once you know which first purchase predicts repeat behaviour. Improving what happens after that first order is usually cheaper than buying more first orders.