Written by: Michał Sobieraj, Operations Manager, Digital Colliers
Your category dashboard says footwear did £480k last month at a healthy blended margin. Everyone nods. The board is happy. Meanwhile one SKU inside that category is losing money on every order, and the rest of the range is quietly funding it. This is the default state of most eCommerce reporting, and it gets more expensive every quarter that ad costs climb and growth stays flat.
Why category views became the default
Category reporting exists because it matches how the business is organised. Buyers own categories. Merchandisers plan by category. Marketing runs campaigns against categories. The P&L rolls up by category. So the dashboard follows the org chart, and the org chart is not wrong, it's just too coarse for the decisions you now need to make.
The deeper reason is data plumbing. Getting to true per-SKU profitability means joining order lines to returns, to ad spend attributed at the ad-set or product level, to fulfilment cost, to payment fees, and to COGS that actually reflects landed cost. Most stacks don't wire that up out of the box. So teams settle for category averages and hope the shape inside is roughly uniform.
It isn't. Roughly 30% of SKUs at a typical multi-channel brand lose money per order once you subtract returns and ad spend. That loss is invisible at the category level because the winners cover for the losers in the average.
The math of a hidden loss leader
Here's the shape. Say a category does 10,000 units at an average £12 contribution after COGS, shipping, and fees. That's £120k of contribution. Looks fine.
Now break it apart:
- 7,000 units of your hero SKUs at £18 contribution = £126k
- 2,000 units of mid-range SKUs at £4 contribution = £8k
- 1,000 units of one "halo" SKU at negative £14 contribution = negative £14k
The category still reports £120k. But you have a SKU actively burning £14k, and you don't see it. Now layer in returns. Online returns run around 19 to 20% of gross sales, and UK apparel runs 25 to 40% depending on the sub-category. If that loss-making SKU also carries above-average returns, the real number is worse than the model above.
Then layer in acquisition cost. DTC CAC is up roughly 40% since 2023 and Meta CPMs kept climbing through 2024 and 2025. Every pound of paid traffic you spend pushing the loss-leader deepens the hole. And with UK eCommerce growth sitting around 3% year on year, you can't outrun this by scaling. Single-digit growth punishes mix problems.
Building the SKU view
The reporting shape you want is one row per SKU per period, with these columns wired in:
- Gross revenue and units
- Returns and refund value, at the SKU line, not just order level
- COGS at landed cost, not standard cost
- Fulfilment and payment fees allocated per unit
- Ad spend attributed at the product or ad-set level where possible
- Contribution margin per unit, and total contribution
The two hard joins are returns and ad spend. Returns often live in a separate system and get booked against orders, not SKUs. Ad spend rarely maps cleanly to individual products, so you'll need an allocation rule (by revenue share within campaign is a defensible default) and you'll need to document it so the number is trusted.
Sort the output by total contribution, ascending. The bottom of that list is your work list. Some SKUs get repriced. Some get pulled from paid traffic. Some get delisted. A few get repositioned as genuine loss-leaders with a job to do, which is a real strategy as long as you've chosen it deliberately.
When to zoom back out
SKU view isn't the answer to every question. It's noisy for trend spotting, useless for range planning, and it can push merchandisers into over-pruning the tail that supports discovery. Zoom back out to category when you're planning next season's buy, measuring brand-level campaign impact, or looking at cross-sell behaviour. Zoom in to SKU when you're deciding what to promote, what to cut, and where your paid budget goes next week.
The cost of not doing this compounds quietly. Every month you run category-only, you fund the loss-leader for another cycle, and your CAC keeps rising underneath. The operators who'll still have margin in 2026 are the ones who already know which SKU is the one bleeding.