What's driving your delivery mix? How checkout choices shape the cost of every order

When Superdry made InPost lockers its only free delivery option in the UK, the share of shoppers choosing a locker at checkout rose from 5-15% at launch to as high as 49%, according to a case study InPost published in September 2026.

A delivery split is easy to read as a record of what customers prefer. Superdry's figures show it also follows what a retailer offers at checkout, and at what price. That puts the split, and the different cost attached to each method in it, within the retailer's control.

The mix worth aiming for is the one with the best combined result for conversion, cost per successful delivery and customer experience. Each market has its own, and a retailer finds it by testing.

Delivery mix is the share of orders a retailer sends to each delivery method, such as home delivery, parcel lockers and staffed pickup points. Checkout pricing, free-delivery rules, option order and presentation all influence it. The right mix is the one with the best combined result for conversion, cost per successful delivery and customer experience, and it differs by market.

What delivery mix means for a multi-market retailer

Delivery mix is the split of your orders across delivery methods, measured per market and per period. It is a different number from carrier mix, which describes how volume is spread across carriers. One carrier can offer home delivery, lockers and pickup points, and one delivery method can be served by several carriers.

Track two versions of the number. Checkout share is what shoppers choose under a given setup; delivered share is what your network carries over a period. Change the checkout and the first number moves straight away. The second follows as orders placed under the new setup make up more of the period. Superdry's public figures include both: locker share at checkout as high as 49% under one setup, and InPost lockers at around 20% of Superdry's UK deliveries as of 30 April 2026, according to CEP Research.

Each method has its own cost and service profile, so the mix sets the blended delivery cost you pay across all orders. For the formats themselves, our guide to PUDO delivery covers lockers, parcel shops and how collection works.

How far a checkout change can shift the mix

Superdry is the clearest public example of a checkout decision moving the mix. It made InPost lockers its sole free delivery option and adjusted how the option was positioned and signposted at checkout. Locker share at checkout then reached as high as 49%.

Making one option the only free one is a strong price signal, so the result shows how far the mix can travel when price does the work. Smaller changes, such as a narrower price gap, a badge or a new option order, need their own tests in your own market.

Two other findings point the same way:

• Choosing the point can change the sale. Flying Tiger Copenhagen used to have a pickup point assigned to each order automatically. After it let shoppers choose their own point in nShift Checkout, it reported a 20% rise in conversions at checkout in 2025.
• Framing changes stated choice. In three scenario experiments with US consumers, published in the Journal of Business Logistics in 2025, the way sustainability information was framed changed how likely people were to choose a locker over home delivery. The effect differed between urban and rural shoppers.

Shoppers respond to price, to choosing their own pickup point and to how options are framed. Your own checkout tests show which of these moves your mix, and by how much.

Measure each method by cost per successful delivery

The carrier rate is the most visible cost of a delivery method, and only part of it. The costs that separate one method from another arrive after checkout: failed first attempts, redelivery, parcels returned to sender, loss and damage claims, and customer contacts.

InPost reports that its locker flow cut Superdry's total cost of delivery by up to 19%, through higher delivery success, less customer contact, fewer loss and damage claims and no redelivery charges. Behind that figure:

• Locker parcels were returned to sender nine times less often than home deliveries.
• Loss and damage claims ran 76% lower than for home delivery.
• Locker customers were four times less likely to contact customer care than customers using other delivery methods.
• WISMO contacts per order fell 20% across Superdry's deliveries as locker volumes grew.

Each of those costs is booked in a different place: carrier invoices, the customer service budget, the claims ledger. A comparison built on rate cards alone misses most of the difference between methods.

The most useful unit is cost per successful delivery, built separately for each method in each market. A low carrier rate with frequent failed attempts can cost more per delivered order than a higher rate that arrives first time.

Our article on cost to serve explains how to build that model and allocate failure costs to the orders that caused them.

Which method comes out cheapest depends on the network. A cost study published in Sustainable Futures in 2024 found that time spent on delivery tasks, and the salary cost of that time, drives delivery cost, and that deliveries to pickup points can be more efficient than deliveries to lockers, depending on the number of parcels delivered. Parcel density and network format vary by market, so the ranking can change from one country to the next.

Steering has a cost of its own

When you make one method free or cheaper to pull orders toward it, the incentive applies to every order that chooses that method, including the orders that would have gone there anyway. The saving comes only from the orders that switch.

A steering change pays when the gain outweighs the cost:

• The gain is the difference in cost per successful delivery, multiplied by the orders that move to the cheaper method.
• The cost is the delivery revenue you give up across every order on that method, plus any orders lost because a shopper's preferred option became more expensive or harder to find.
• Conversion can move in either direction, so it belongs in the same calculation.

Your own checkout and delivery data can supply all three, which makes the size of the incentive a variable to test. A narrower price gap, a different free-delivery threshold or a change in presentation each costs less to run than a free option, so test them against it before giving delivery away.

The right mix differs by market and by order

Markets that look alike on a map can behave differently at checkout. PostNord's spring 2026 research found that in Norway, home delivery is still the most preferred option and service points the most used, even though locker use has more than doubled since 2024. In Finland, PostNord found the opposite pressure: preference for parcel lockers slightly exceeds actual use, which it reads as a sign that infrastructure may not yet meet demand.

The mix keeps moving even in a locker-dense market. On its Q2 2026 results call, InPost said its Polish locker volumes grew about 1% year on year while to-door volumes grew 46%, largely driven by international marketplaces, as its Polish machine count grew about 12%.

Distance to the pickup point affects satisfaction too. In a 2023 survey of cross-border shoppers by the International Post Corporation, about half of those who traveled up to 100 meters to collect a parcel were extremely satisfied, against 18% of those who traveled more than a kilometer. A locker share pushed up with price, among shoppers whose nearest point is far away, can cost satisfaction that the delivery cost line will not show.

Some orders also belong at the door: bulky items, heavy baskets and gifts going to another address. Our article on click and collect covers when pushing pickup is the wrong call, and our analysis of parcel locker growth in Europe gives the country-by-country picture.

So the target is a mix for each market, and often for each type of order, found through your own tests.

Re-test as locker and pickup networks change

The cost and coverage of each method also shift as the networks behind them expand and change hands.

In September 2026, Quadient said it intends to sell its lockers business to focus on its Digital strategy, keeping a smaller European private network. The first deal sells its UK open network of about 3,000 lockers to IDS Holdco for an enterprise value of €65 million. In FY2025 the lockers business was profitable and growing, with revenue up 22.4% to about €114 million at a 5.0% EBITDA margin. Quadient expects the exit to remove about €120 million of planned locker capital expenditure over five years, a measure of what network density costs the companies that build it.

Others are still building. GLS says its out-of-home network nearly doubled, from 70,000 to 130,000 points, in 24 months. DPD and GLS plan 20,000 shared out-of-home points in Germany, including up to 6,000 lockers, by the end of 2027.

Our article on how out-of-home networks are being reshaped covers the ownership changes and what they mean for carrier agreements.

A cost per successful delivery you measure this year may not hold next year, in either direction.

How to test your way to a better delivery mix

Testing delivery mix works like any checkout experiment, with two differences: the outcome includes where orders go, and some results only arrive after the parcels do.

Start with a baseline. For each market, record checkout share and delivered share by method, the conversion rate, and cost per successful delivery by method.

Write the hypothesis in mix terms. For example: a lower locker price for baskets below the free-delivery threshold will raise locker share without lowering conversion, and will reduce cost per successful delivery across the market.

Change one checkout variable at a time, in one market, against a control.

Read conversion and mix from the same test. A change that lifts locker share but loses orders may cost more than it saves.

Follow the orders downstream. Failed attempts, returns to sender, claims and contacts trail the checkout, so keep reading the test cohort after the experiment window closes.

Repeat by market. Treat a result from one market as a hypothesis to test in the next.

Of the checkout variables you control, price has the strongest public evidence behind it:

The rows with no published test are where your own results will tell you the most.

For setting up the experiments themselves, see our guide to A/B testing checkout delivery options.

Run delivery-mix tests across carriers and markets

Managing delivery mix takes control over the options shoppers see in every market, without a development project for each change. In nShift Checkout, delivery options, prices, free-delivery thresholds, delivery times, PUDO locations, badges and option order are all configuration, with rules that can depend on market, cart value, product and postcode. Teams can run A/B tests on delivery options, pricing, order and thresholds without developer involvement, and send a share of traffic, or the traffic from one market, to the variant.

In the results file for each experiment, you can see which delivery option every checkout session selected, alongside the carrier, the price and the shopper's country and postcode. The same test that compares conversion rates therefore shows where the orders went.

For the downstream half, you can filter shipment reports in nShift Track by carrier, service and status category, across outbound and return shipments, and follow delivered, failed and returned parcels method by method.

You can run all of it across one network of 1,000+ carriers and 70+ PUDO networks. Where you have the carrier agreements in place, a test in one market can compare a locker network with a staffed pickup network, or one carrier's home delivery with another's.

Deploy the first test in your highest-volume market, the one where you know least about cost per method: set the baseline, change one checkout variable, and read conversion, mix and downstream cost together.