Written by: Michał Sobieraj, Operations Manager, Digital Colliers
If you run a DTC brand, roughly 30% of your SKUs are losing money on every order once you factor in returns and paid acquisition. That's not a doom stat pulled from nowhere. It's what Profitero sees across multi-channel brands when someone finally builds the unit economics view properly. The frustrating part is that the P&L looks fine. Blended margin is healthy. Cash is moving. And yet a third of the catalogue is quietly funded by the other two-thirds.
That's the shape of the problem. Now the mechanics.
Blended margin is a liar by design
When you sum revenue and subtract COGS at the account level, the winners subsidise the losers and you never see the split. A hoodie doing 55% contribution margin covers a t-shirt doing negative 12%, and the average lands at a respectable 28%. Everyone nods at the monthly review. Nobody kills the t-shirt.
The blend gets worse as costs climb. Customer acquisition cost across DTC has risen about 40% since 2023, and Meta CPMs have kept drifting up through 2024 and 2025. Every point of CAC increase moves more SKUs from marginal-profitable to marginal-loss, but the blend keeps hiding it because your hero products absorb the pain.
Meanwhile the top line isn't bailing anyone out. UK eCommerce grew about 3% in 2024. Single-digit growth is the baseline now. You can't outrun a leaky catalogue on volume.
The four-system tax
Here's why the loss-makers stay invisible. The data you need lives in four separate places, and none of them talk to each other cleanly:
- The storefront (Shopify, or similar) knows what sold and at what price
- The fulfilment layer (3PL, WMS) knows the pick, pack, ship and returns cost per order
- The ad platforms (Meta, Google, TikTok) know acquisition cost, but only at the campaign level
- COGS lives in a spreadsheet someone updates quarterly, if you're lucky
To answer "does SKU-1472 make money," someone has to stitch all four together, per order, net of returns, with the right slice of ad spend attributed. Most finance teams don't have time. Most analytics teams don't have the mandate. So the question doesn't get answered, and the catalogue drifts.
Returns make the maths brutal. Online return rates sit around 19-20% of gross sales overall, and UK apparel runs 25-40% depending on category. A product with a 22% contribution margin and a 35% return rate is almost certainly underwater once you count the return shipping, the restocking labour, and the units that come back unsellable.
The last-click ROAS trap
Even brands that do build a unit economics view often anchor on the wrong ad number. Platform-reported ROAS is last-click and self-attributed. It flatters. When you swap in a blended CAC calculated from actual new-customer count divided by actual total ad spend, the picture darkens fast, especially on SKUs that only sell through paid.
The pattern worth noticing: a SKU can have a strong platform ROAS and still lose money per order, because platform ROAS doesn't know about your returns, your 3PL invoice, or the fact that half those "conversions" would've happened organically.
What the sharper operators do
The brands I see pulling ahead have done something unglamorous. They've built a per-SKU, per-order contribution margin view that updates weekly, joins all four systems, and nets out returns on a rolling window. It usually takes an engineer a few weeks. It's not clever, it's just plumbing.
Once they have it, three things tend to happen:
- They kill or reprice the bottom decile of SKUs. Often 5-10% of the catalogue accounts for the majority of the loss.
- They stop bidding paid traffic against loss-making SKUs. CAC comes down because the ad budget consolidates onto products that can actually pay for it.
- They renegotiate 3PL rates on the SKUs where fulfilment cost is the killer, or move them to a different pick path.
The compounding effect is the point. Cutting the losers doesn't just remove the loss. It frees ad budget, warehouse slots, working capital, and merchandising attention for the SKUs that actually earn. A brand that trims 8% of its catalogue often sees contribution margin move by 3-5 points inside a quarter.
The cost of not knowing
Every month you run the current setup, the 30% keeps bleeding. That's not a tooling problem or a talent problem. It's a decision to keep looking at blended numbers because the per-SKU view is annoying to build. In a market where CAC is up 40% and top-line growth is 3%, that decision gets more expensive every quarter.