An online retailer reduced shipping costs by 4.8% in an experiment that took delivery expenses into account when ranking products for shoppers, according to research involving 2.8 million consumers.
The test shortened delivery distances by 34 miles per order and increased profit by 0.2%, the researchers report. Conversion and customer satisfaction were preserved during the trial, conducted over eight days on one US e-commerce platform.
The findings point to a decision that retailers may overlook when managing delivery costs: which products customers see first. They are based on a working paper and one company’s experiment, so the results should not be treated as a guaranteed saving for every retailer.
Shipping costs enter the search ranking
Product ranking is the order in which a shopping site displays items. The order can affect what shoppers notice and choose, particularly when a search returns many similar products.
Researchers Malika Korganbekova, Aliya Korganbekova and Alibek Korganbekov designed a ranking system that considered shipping costs alongside the likelihood that a shopper would buy an item. Their working paper describes a field experiment in which shoppers were assigned to the new system or the established ranking.
In an account of the research, Harvard Business School said the calculation considered a product’s size, the customer’s and warehouse’s locations, and the carrier’s rate. Products that were less expensive to ship received a smaller cost penalty. Strong customer preferences provided some protection against that penalty.
A ranking based only on delivery expense might put an unsuitable product ahead of the item a shopper actually wants. The researchers instead tried to balance shipping costs against the chance of a purchase.
The company in the trial had around five million products, 50,000 sellers and 150,000 seller warehouses. Those characteristics gave the researchers many possible product and shipping combinations to work with.
Why location and product choice matter
Delivery costs depend on more than the distance from a warehouse. Package size, carrier rates, service speed and the way a company fulfills an order can all affect the expense. A nearby item is not automatically the cheapest to deliver. As our earlier reporting on holiday marketing noted, delivery updates also form part of the customer’s experience.
When shoppers view similar products, the location from which each item would be sent can influence the cost of completing a sale. A retailer that knows those costs before presenting results may be able to incorporate them into its ranking alongside relevance, price and availability.
The study also estimated that delivery-related emissions fell 3.8% in the experiment. That estimate concerns the researchers’ delivery calculation, not every environmental effect associated with the products or their packaging.
The paper says the effect of the system varies with how readily consumers substitute one product for another and with warehouse coverage. A marketplace with several comparable items held in different locations may have choices that a specialist retailer with a narrow inventory does not.
Retailers need to watch more than delivery expense
The eight-day test offers evidence about a specific platform and period. It does not establish whether the same results would persist through a holiday season, across different categories or after sellers adjust their behavior.
Retailers considering a similar approach would need to monitor sales, returns, customer experience and the exposure of individual sellers as well as shipping costs. If an item is hard to substitute, giving delivery expense too much weight could make search results less useful.
Accurate cost data also has to reach the teams that design and operate search. Harvard’s account notes that shipping information and ranking decisions can sit in separate parts of a business. Bringing them together can change the trade-offs the ranking system is able to consider.
