True Profit in DTC: How to Read the Real Margin
- Treat a published market margin as uncomparable: definitions differ more than performance does.
- Attribute returns, fulfilment variance, payment fees and discount leakage at order level.
- Expect channel rankings to invert once contribution replaces revenue in the comparison.
- Compare your own figure over time and by cohort, which is the only benchmark that holds.
- Read the campaign with the best return on ad spend as a candidate for the worst contribution.
- Build the figure once in the warehouse rather than repeatedly in spreadsheets.
Every few months somebody asks me what True Profit in DTC averages out at across the market. It is a reasonable question and the honest answer is unsatisfying: no published figure is comparable, because the number depends almost entirely on which costs each business decides to attribute to an order. That is not a data gap waiting to be filled. It is a property of the metric. Last updated: November 2026.
I have spent 13 years in eCommerce, and the CROBenchmark dataset of 7,000+ websites in 15+ industries assessed against 248+ audit criteria shows the same thing the conversations do. Two brands will report margins eight points apart while running near-identical economics, and two more will report the same margin while one is compounding and the other is quietly shrinking.
So this piece does something more useful than quoting an average. It explains why the market number cannot be compared, names the four cost layers that go missing in most order-level margin calculations, and sets out the benchmark that actually holds. The mechanics of the calculation itself are in how to measure True Profit, not ROAS.
Why a market figure for True Profit in DTC cannot be compared
Consider two brands with the same products, the same prices and the same advertising efficiency. The first treats returns processing as a warehouse overhead. The second attributes the handling cost and the written-down unit to the order that generated it. Both are being honest. Their reported margins will differ by several points, and the difference describes an accounting choice rather than a business.
Now multiply that by fulfilment, payment fees and discount treatment. The spread between two reasonable interpretations becomes larger than the spread between a good operator and a mediocre one. At that point a market average is a number assembled from incompatible inputs, which is worse than a weak signal because it reads as a strong one.
This is also why comparing yourself against a published figure tends to produce the wrong decision rather than no decision. A brand that attributes carefully will look worse than its peers and conclude it has a margin problem, when what it has is a measurement discipline its peers lack. The broader version of this argument, that the numbers each function reports are rarely the numbers the business runs on, is in CRO, creative & AI visibility tie into one growth system.
The four cost layers that go missing
Contribution leakage is defined as variable cost that belongs to a specific order but is recorded above it, as an overhead or an average, so the order appears more profitable than it was. Every one of the four layers below is a form of it, and they compound in the same direction.
- Returns, in full. Not the refund. The refund plus inbound shipping, inspection, repackaging and the units that cannot be resold at full price. In categories with high return rates this layer alone can decide whether a product line contributes anything.
- Fulfilment variance. An average cost per order hides the fact that a two-item basket and a bulky single item cost very different amounts to ship. Attributing by basket shape rather than by average moves the ranking of product mixes, not just the total.
- Payment and platform fees. Processing, instalment provider fees, marketplace commission and currency spread. Individually small, collectively material, and almost always recorded as a monthly line rather than against the orders that incurred them.
- Discount leakage. Stacked codes, loyalty credit redeemed later, free shipping thresholds that are met with a low-margin add-on. The order's recorded revenue is correct and its effective revenue is lower.
The groundwork for all four is a profit and loss statement structured so that variable and fixed costs are genuinely separable, which is covered in the eCommerce profit and loss statement. The distinction those four layers sit inside is set out in gross margin versus net margin.
Where contribution leaks, and how each gap presents
| Cost layer | Where it usually sits | How the gap presents |
|---|---|---|
| Returns in full | Warehouse overhead | High-return products look profitable |
| Fulfilment variance | An average per order | Bulky items under-costed |
| Payment and platform fees | A monthly finance line | Channel mix looks cost-neutral |
| Discount leakage | Recorded revenue stands | Promotions appear margin-safe |
| Servicing cost | Support headcount | Complex products look cheap to sell |
| Market average margin | Somebody else's definition | Adopted as a target |
Unexpected cost and unclear information have been among the leading reasons shoppers abandon purchases for as long as Baymard Institute has measured it, at around 70% average cart abandonment [Baymard Institute, 2025]. There is a symmetry worth noticing: the costs a shopper discovers late are frequently the same costs a brand attributes late, which is how a free-shipping threshold becomes a conversion tactic and a margin problem at the same time.
What inverts once you measure contribution rather than revenue
This is the finding that changes behaviour, and it lands harder than any argument about methodology. A team sees its best-performing campaign drop to the bottom of the table and its quiet, unglamorous campaign rise to the top, and the conversation about attribution ends in about ten minutes.
It happens because return on ad spend is a revenue ratio and revenue is the number least affected by the four layers above. A campaign that sells a heavily discounted bestseller with a high return rate maximises exactly the metric that ignores discounting and returns.
The specific way finance and growth teams get this wrong together is by agreeing to build the contribution figure and then letting each team keep its own version of it. Growth calculates contribution with returns at a category average, because order-level return data arrives too late for a weekly decision. Finance calculates it with actual returns, a month in arrears. Both numbers are defensible, they differ by enough to change a budget, and the monthly meeting becomes a reconciliation argument rather than a decision. Agree one definition, accept that the fast version is an estimate, and label it as one.
Retention arithmetic sits on the same foundation. A five percent lift in retention can raise profits by between 25 and 95 percent, a finding credited to Bain's Frederick Reichheld [Harvard Business Review, 2014], and none of that compounding is visible in a revenue-ratio view of a campaign. The payback version of the same question is in the CAC payback period, and the per-customer version in customer profitability analysis.
The only benchmark that holds is your own
The discipline is less exciting than a benchmark table and considerably more useful. Write the definition. Specify which costs are attributed to an order and which are not. Then stop changing it, because a definition revised mid-year destroys the only comparison you had.
Cohorts matter as much as periods. Contribution by acquisition cohort tells you whether the customers you are buying now are worth more than the ones you bought a year ago, a question most growth plans assume rather than check.
AliveCor used Omniconvert to run a structured A/B testing programme and achieved +21% conversion rate, +5% revenue per visitor, and 94% statistical relevance across their experiments. [Omniconvert, AliveCor case study] That is a testing result rather than a margin result, and the transferable discipline is the same one this whole argument rests on: fix the definition, change one thing, and read the effect on the number you agreed to measure.
Where to start this quarter
- Write the definition down. One page. Which costs are attributed to an order, which are not, and who signs off a change to the list.
- Start with returns. The largest missing layer in most catalogues, and the one most likely to reverse a decision about a product line.
- Build it once, where the data lives. A contribution column in the warehouse beats four spreadsheets that disagree politely.
- Re-rank one month of campaigns. By contribution rather than by revenue. The reordering is the argument.
Doing this properly is also what makes automated prioritisation worth anything. Omniconvert Commander (formerly Nexus) is an AI for eCommerce growth engine that unifies commerce data, prioritises experiments by True Profit, and generates campaigns and creative you approve before they go live, and the quality of that prioritisation is bounded by the quality of the contribution figure underneath it. A system ranking experiments on a revenue proxy will rank them confidently and wrongly.
The longer argument about what happens when this coordination is paid for in people's time instead of solved in the data is in the Human Middleware Problem.
True Profit in DTC FAQs
What is the average True Profit margin in DTC?
No published figure is comparable, and that is the useful answer rather than an evasion. Every business decides which costs it attributes to an order, so two brands reporting the same margin can be running opposite situations, and two brands with genuinely identical economics can report figures far apart. Measure your own on a stated definition and compare it against itself over time.
Why does the definition vary so much between brands?
Because the hard part is attribution rather than arithmetic. Returns processing, fulfilment variance by basket shape, payment and platform fees and discount leakage can each be treated as an order cost or as an overhead, and both treatments are defensible. Four defensible choices produce a wide range of honest answers to the same question.
Which cost layers are usually missing?
Four, in rough order of size: the full cost of a return including handling and the lost unit, fulfilment cost variance by basket rather than an average per order, payment and platform fees including currency spread, and discount leakage from stacked codes and loyalty credit. Each one is usually averaged somewhere above the order rather than attributed to it.
What changes once you measure contribution instead of revenue?
Channel and campaign rankings often invert. The campaign with the best return on ad spend frequently sells the discount-heavy, high-return products that generate revenue and very little contribution, so it looks like the best performer in one view and the worst in the other. Teams that make this switch usually reallocate budget within a quarter.
What should a brand benchmark against?
Itself, by cohort and over time, on a definition written down and left alone. A contribution figure for the same product set this quarter against last quarter is a real comparison. The same figure against a market average built on somebody else's attribution choices is a number that feels like a comparison and is not one.
Where should a team start?
Write the definition down, then build the figure once where the data lives rather than repeatedly in spreadsheets. Start with returns, because it is usually the largest missing layer and the one most likely to change a decision. A single order-level contribution column is worth more than a dashboard built on revenue.
What is the bottom line on True Profit in DTC?
Stop looking for the market number. It does not exist in a form you can act on, and chasing it has cost more brands a correct decision than it has ever saved one. Write your own definition, attribute returns first, build the figure once where the data lives, and then re-rank a month of campaigns by contribution rather than by revenue.
The reordering will be uncomfortable and it will be the most valuable half-day of analysis your team does this quarter. After that, the only benchmark worth holding yourself to is the same figure, on the same definition, three months from now.