Why you need to treat your fulfilment strategy like a marketing campaign

Ask any eCommerce director how they improve their conversion rate, and they'll talk about A/B testing. Different headlines, button colours, email subject lines, checkout flows, retailers run experiments constantly, feeding live data into decisions until the best-performing variant wins. It's disciplined, iterative and deeply embedded in how modern retail teams operate.

Now ask the same director how they improve their fulfilment strategy. The answer is usually some combination of gut feel, historical reporting, and periodic reviews with logistics partners. Someone pulls together a spreadsheet. Maybe a simulation is run. And then a decision worth millions of pounds gets made based on a best guess. The contrast with the marketing team's approach is striking and it needs to change.

Fulfilment is a performance channel so treat it like one

Every order your business ships is the product of a series of decisions: which warehouse or store to pick from, which carrier to use, whether to split a multi-item order or hold it for consolidation. These decisions have a direct, measurable impact on your margins, your delivery promise, and your customer satisfaction scores.

Fulfilment is a performance channel. It has inputs, outputs, and outcomes that can be tested, measured, and optimised just like paid search or email marketing. The numbers make the case: fulfilment costs already consume 12–20% of eCommerce revenues, and McKinsey projects that rising to 25% as wage pressure and last-mile costs increase. Last-mile delivery alone now accounts for 53% of total shipping costs, up from 41% five years ago. If a quarter of your revenue is at stake, testing your sourcing rules isn't a nice-to-have it's a margin strategy. And it cuts both ways: 48% of cart abandonments are driven by unexpected shipping costs, meaning your fulfilment decisions affect the orders you never even get.

The problem is that most retailers don't have the infrastructure to treat it that way. Fulfilment logic tends to live in a single, monolithic ruleset applied uniformly to every order routing by cheapest carrier or nearest location, never both, and never with a controlled comparison running alongside.

Why simulation isn't enough

Many retailers attempt some form of modelling before changing their sourcing logic. But simulation has a fundamental flaw: you can never perfectly recreate your inventory picture at any given moment. Stock is being sold in stores. Returns are being processed back into available-to-promise pools. A flash sale is draining stock at three distribution centres faster than your replenishment cycle can respond.

No simulation captures all of that in real time. Live A/B testing routes actual orders through actual conditions and that difference is everything. If your test logic routes orders to a node that turns out to be oversold on a key SKU, you see exactly what happens. A simulation would have assumed perfect inventory accuracy and missed it entirely.

What A/B testing looks like in fulfilment

The basic principle is identical to marketing: define two strategies, split your live order volume between them, and measure the outcomes against defined metrics.

Strategy A might be your current approach always fulfilling from the distribution centre closest to the customer. Strategy B might be a new hypothesis: route orders to stores when stock is available and the store is within a defined radius, reducing outbound distance and using inventory that might otherwise require markdown. You run both simultaneously against real production orders, then let the data decide.

One of the most under-appreciated aspects of a well-built A/B testing framework is the ability to control what percentage of order volume runs through the experimental logic. With the right tooling, this becomes a continuous loop: measure strategy performance with live data, surface what the data means, and act through controlled tests where only a fraction of orders are ever at risk.

The metrics that matter

The discipline is knowing what to track. Retailers often fixate on a single number,  usually delivery cost and miss the fuller picture. A strategy that saves on shipping by splitting orders might be generating twice as many customer service contacts.

The metrics worth tracking across any fulfilment experiment: gross margin per order, delivery cost, on-time in full rate (your guardrail if this moves negative, the test isn't ready to scale), split shipment rate, order-to-door time, and average delivery distance. These tell a complete story together.

Start small, learn fast

You don't need to overhaul your entire fulfilment network to begin. Pick one sourcing logic hypothesis your team has been debating and run a clean experiment at a small percentage of order volume. Define your success metrics before you start. Treat it with the same rigour your marketing team applies to a landing page.

Your fulfilment strategy is already making thousands of decisions a day, which node, which carrier, which trade-off. The retailers who win the next decade won't just have the best products or the smartest advertising. They'll be the ones who have turned their supply chain into a continuous learning and optimisation engine. The only question is whether you're ready to start learning.