UK Ecommerce & DTC

Ecommerce is the furthest sector in this series from the clinical model the Axial factory was built around. Every other pack has a physical bottleneck — a chair, a van, a seat, a fee earner’s hour. This one has none, and that absence is the whole analytical problem rather than an advantage.

What the sector publishes

• UK internet sales share, April 202627.3%Observed
• GB average conversion and median site3.4% / 2.35%Observed
• Mobile against desktop conversion1.8% / 3.9%Observed
• Average order value, March 2026, down 3.6%£122.02Observed
• UK clothing return rate23.6%Observed
• Modelled contribution after returns, goods, fulfilment and acquisition14%Modelled
• UK acquisition cost by vertical—Not established

The Office for National Statistics publishes UK internet sales as a share of total retail monthly — 27.3% in April 2026, against the United States at 16.9%, Australia at 12.7% and Canada at 5.7%. That is better demand measurement than any other sector in this series enjoys. Against it, the sector’s largest controllable cost is unobtainable: no UK source publishes customer acquisition cost by vertical, and the closest proxies are global, in dollars, and drawn from a single analytics platform’s merchant base. A brand cannot know whether its acquisition cost is good or bad by comparison.

The subscription archetype carries a modelled central EBITDA of 17% against 9% for both the multi-SKU own-brand and the marketplace reseller, on the lowest net revenue of the three published cards, and both mechanisms behind that are structural. At a modelled 78% repeat share, acquisition is amortised across many orders and falls from 24 points of gross order value to 17. And because subscription revenue is known in advance, stock is bought against a committed order rather than a forecast, so inventory days fall and the cash cycle with them — from 61 days to 22.

UK ecommerce has the best-measured demand in this series. The Office for National Statistics publishes online retail share monthly, the UK is the most online-penetrated of the large comparable markets, and the category spread matters more than the headline.

UK internet sales sat at 27.3% of total retail in April 2026 on the ONS J4MC series, or 28.1% on the bulletin headline for the same month, and neither figure is much use to an individual operator. The reason is the category spread: clothing runs at 29.3% online while food runs at 10.1%, a nineteen-point gap inside the same national average. A brand forecasting its own channel share from the national figure will be wrong by more than that figure’s own year-to-year movement. The operationally relevant number is category penetration, and it is published.

Revenue begins with sessions the brand has usually paid for and is reduced at four successive points, of which the largest two are returns and acquisition. Neither appears in a gross margin calculation.

Conversion, average order value and return benchmarks come from analytics publications aggregating merchant data, not from surveys, and their panels self-select toward merchants using analytics tooling. Acquisition cost is modelled throughout and marked Not established as an observed UK figure.

Ecommerce brands are routinely managed on gross margin, and it is the wrong measure: a brand with a 58% gross margin looks healthy and may be losing money on every first order. Gross margin is struck after cost of goods and before returns, fulfilment and acquisition — on these figures it would sit at 47 points while actual contribution is 14, and the thirty-three point gap contains the two variables that decide the business. A brand that cannot state contribution after acquisition by cohort has nothing external to check itself against either.

All corridors on this page are stated after returns, cost of goods, fulfilment and customer acquisition. Gross margin is deliberately not shown, because in this sector it conceals the two deductions that decide whether the business works.

Base central model: 3.4% conversion · £122 average order value · 19% return rate · 24 points of gross order value to acquisition

Contribution and EBITDA break-even are conventional; the third — cash-cycle break-even — is specific to this sector and it is the one that ends otherwise healthy businesses. At a modelled 61-day cycle a brand pays for stock and acquisition roughly two months before the corresponding revenue is in the bank, and every additional order widens that gap in absolute terms. Nothing in the profit and loss account warns of it, because on an accruals basis the brand genuinely is profitable. The subscription archetype’s 22 days against the multi-SKU’s 61 is therefore a bigger difference than the margin gap suggests.

This page is deliberately different. Every other sector in this series has a physical bottleneck — a chair, a van, a seat, a fee earner’s hour. Ecommerce has none, so the capacity equation is restated as the working capital cycle, which is what actually limits how fast the business can grow.

The cash cycle has three components and the brand controls exactly one of them without changing its operating model. Inventory days are a function of category, range depth and demand predictability; settlement days are set by the payment processor. Supplier payment terms are negotiable, and every additional day of supplier credit is a day removed from the cycle directly, with no operational change at all. That is why the terms rung sits second on the staircase and is the one most often left untouched — usually negotiated when the brand had no volume, and never revisited.

Ecommerce does not usually have a traffic problem. It has a conversion problem concentrated on one device and a retention problem that makes every order carry a full acquisition cost.

Ask a prospect for these twelve figures. Most can produce sessions, conversion, revenue and return rate in aggregate. Conversion by device, contribution by cohort after acquisition and return rate by SKU are the ones that are missing, and they bracket every loss on this page.

Mobile converts at a UK average of 1.8% against desktop’s 3.9% and carries the larger and growing share of sessions, which produces an effect most brands experience as a mystery: blended conversion falls with no change to the site, the product, the pricing or the market. Nothing has gone wrong and every aggregate metric says something has. The diagnosis requires only that conversion is reported by device rather than blended — which almost every analytics platform does by default and almost no brand reviews. That gap is the single largest conversion finding available in this sector.

Ecommerce has almost no barrier to entry and an unusually high failure rate among brands that look successful from outside. Revenue growth is visible and cash consumption is not, which is why the ladder below is ordered by funding self-sufficiency rather than by size.

Ecommerce is unusually good at hiding its own failure mode. Revenue is visible daily, orders are countable in real time, and growth produces immediate and satisfying feedback; cash consumption produces none of that. Stock ordered two months ago against demand that has not yet materialised does not appear as a problem until the supplier invoice falls due, and by then the brand has usually ordered again. The ladder is therefore ordered by funding self-sufficiency rather than by revenue or sophistication: a £3m brand at rung one is in more danger than an £800k brand at rung five, and both would describe themselves as growing.

Modelled effect on normalised EBITDA of a one-standard-step move in each driver, multi-SKU own-brand archetype, widest first. The widest driver is the one with no UK benchmark.

Customer acquisition cost ranks first on this chart and is the only driver anywhere in the series that cannot be checked against a published UK figure. The available proxies are global: a February 2026 benchmark across more than 30,000 DTC brands worldwide reports a paid-ads median cost per acquisition of $32.74 and a Meta median of $38.19, with separate data putting the cost of a single DTC purchase at $68 to $84. Those figures span every category and country and come from one analytics platform’s merchant base. They frame the question and cannot answer it.

Technology creates value in ecommerce only when it moves one of the drivers on the previous page. This is the most heavily tooled sector in the series and the one where tooling most often adds cost without adding contribution. Each intervention below names the driver and the test.

Ecommerce brands are sold traffic more insistently than any other sector in this series is sold anything, and on the modelled economics it is the wrong first purchase. If contribution after acquisition is 14 points and the brand does not know whether its last cohort covered its acquisition cost, more traffic scales whatever the current answer is — including a negative one. That is the mechanism behind a pattern the sector knows well: a brand grows revenue rapidly, raises capital against that growth, scales acquisition further, and fails while reporting increasing revenue throughout.

The sector model becomes commercially useful at the moment a named brand is scored against it. The eight dimensions below convert everything in this pack into a diagnostic that can be completed in a single working session from the brand’s existing analytics and platform data.

Ecommerce is the most data-rich sector in this series, so the exercise is rarely about collection and almost always about assembly. Cohort discipline scores one most often — not because brands lack data, but because contribution by cohort requires joining advertising spend to order data to return data, which no single dashboard does by default. The output is a profile rather than a total, and sequencing follows the sensitivity ordering with one exception: where cash-cycle governance scores one and the brand is growing, it is addressed first, because the failure it produces is terminal and arrives without warning.