Written by: Michał Sobieraj, Operations Manager, Digital Colliers
Uber buying Delivery Hero at around $14.8B is the kind of deal that quietly rewrites your P&L six months later. If you sell through aggregators, marketplaces, or any platform that sits between you and the customer, consolidation moves pricing power away from you. The acquirer now owns the demand, the customer data, and the take-rate lever. You own the inventory and the risk.
This isn't a food delivery story. It's the same shape you've seen with Amazon 1P vs 3P terms, with Instacart on CPG, with Deliveroo on restaurants. Every few years a platform consolidates, and the merchants who don't already have their own margin model find out what their business actually earns when the commission goes up 200 basis points.
Why consolidation always renegotiates your unit economics
When two aggregators become one, three things happen in sequence. The merged entity re-tiers commissions. Promotional co-funding gets "rationalised." And the data you used to get about your own customers gets thinner, because the platform now has less competitive pressure to share it.
You can't negotiate against any of that if you don't know your true contribution margin per SKU per channel. And most brands don't. They know blended gross margin. They know channel revenue. The gap between those two numbers is where the platform makes its move.
It matters more now because the cushion is gone. UK eCommerce grew about 3% in 2024 versus 2023, so single-digit growth is the baseline you're planning against. DTC customer acquisition cost is up roughly 40% since 2023. Meta CPMs kept climbing through 2024 and 2025. There's no ad-funded growth to hide a bad take-rate renegotiation inside.
What a real merchant-side margin model looks like
The shape of the thing isn't exotic. It's just work that most brands haven't done because it wasn't urgent yet. Here's the minimum you want:
Order-level ingestion from every channel. Not daily rollups. Individual orders, with SKUs, quantities, gross price, discounts, and the channel identifier. Shopify, Amazon Seller Central, each marketplace, each aggregator, your own DTC site.
Commission and fee normalisation. Every platform reports fees differently. Fulfilment fees, referral fees, ad fees, storage, chargebacks. You need one schema so a £1 of Amazon fee and a £1 of TikTok Shop fee land in the same column.
Returns and refunds joined back to the original order. This is where most models break. Online return rates run around 19-20% of gross sales, and apparel in the UK runs 25-40% depending on category. If you can't reconcile the refund to the order line, your margin numbers are fiction.
Per-SKU contribution margin after fees, refunds, and variable fulfilment. Not gross margin. Contribution. The number that tells you whether this SKU on this channel actually pays for itself.
When brands build this properly, the pattern is uncomfortable. Roughly 30% of SKUs at a typical multi-channel brand lose money per order once you account for returns and ad spend. You're subsidising them with the winners, and you don't know which is which until the model exists.
The ops question nobody wants to answer
Building the model once is a project. Keeping it accurate is an operating discipline. So the real question is who owns it.
In the brands that get this right, it's usually a named person on the finance or ops side with SQL access and a standing weekly slot. Not a BI team ticket. Not a quarterly deck. A person whose job includes noticing that Channel X's effective take-rate ticked up 80 bps last week and flagging it before the quarter closes.
A few questions worth asking your own team this month:
If a marketplace raised commissions 150 bps tomorrow, could you tell me by Friday which SKUs go underwater?
How often does the margin model refresh, and who signs off that the refund data is joined correctly?
When was the last time we compared platform-reported fees against what actually hit the bank?
If those questions get vague answers, that's the gap. Consolidation deals like Uber and Delivery Hero don't create the problem. They just set the deadline for fixing it. The brands that already have the model treat these announcements as a Tuesday. The ones that don't spend the next two quarters building it under pressure, which is the worst possible time.
Sources
This article was originally published on the Digital Colliers Blog. Digital Colliers helps DACH and UK companies implement AI — see our AI consulting services or contact us.
Top comments (0)