You will be redirected back to your article in seconds
Skip to main content

How ASOS Manages Markdowns Across 85,000 Products

Growing as much as 40 percent annually, U.K.-based “online fashion destination” ASOS needed a better way to manage its markdown process and turned to Oracle’s Clearance Optimization (CO) tool to transform its approach.

Though the fast-fashion e-commerce pure-player originally planned to implement CO along with a major Oracle Retail deployment in 2019 as part of its “Truly Global Retail” initiative, retail subject matter expert Lucy Partridge and IT program manager Chris Metcalf said they found that the markdown tool would provide such a compelling “quick win” that would accelerate ROI that they decided to decouple the system from the larger tech-stack rollout and piloted it in 2016—even if that means extra work by having to tweak CO once Oracle Retail goes live next year.

With 85,000 products on site, 4,500 added weekly, 15.4 million global customers, 1.5 billion site visits and 11 percent of its businesses coming from the rapidly growing U.S. market, ASOS is accustomed to a hectic, fast-paced environment.

The company quickly recognized that it needed to streamline its markdowns to better manage profits and reduce the burden on merchandise planners, who simply didn’t have enough time to manually adjust pricing in a company where products are delivered from “sketch to store” in six weeks, Partridge said. Because fashion-oriented product has a limited shelf life in which to appeal to customers, it’s essential to “clear stock” before that lifetime runs out, she added.

Related Stories

“Twenty-somethings are mobile obsessed, checking their phones hundreds of times a day,” Partridge said. “We need to show them something new every time they visit.”

ASOS is undertaking a mindset shift, starting to drive volume in its assortment, whereas traditionally it has focused on having a broad selection of product but shallow depth. “It’s hugely frustrating for consumers to find their size is out of stock,” Partridge explained.

ASOS’s goal is to sell through 85 percent of products in 26 weeks.

Metcalf described CO as a forecasting tool that aids planners in optimizing the entire approach to markdowns. The markdown engine incorporates sales history and pricing elasticity—how sensitive a product’s sales are to a change in pricing?—coupled with a projected sales curve and established business rules.

ASOS started the CO project in January 2017 and piloted the software in late May in order to be ready to manage markdowns on the new system for a planned June sale. After another sale in late July, Metcalf’s team got the green light to fully roll out CO company-wide, and a September sale was the first event for which all markdowns came through the new Oracle tool. To date, ASOS has taken 35,000 markdowns across two sales using CO.

Prior to the implementation, ASOS planners had been working via Excel so getting them to switch wasn’t too difficult, Metcalf explained, though Partridge noted that some employees accustomed to the Microsoft spreadsheet’s “flexibility” initially weren’t keen on the new software. “We were quite nervous they’d be reluctant to try something that would take away their decision making,” Partridge said.

What’s more, planners have had to get used to take recommendations from a machine rather than driving decisions themselves. However, that shift to a data-driven process provides valuable results, automating a process that otherwise can dominate their time.

Metcalf said his team experienced some “tense moments” when CO was turned on and the first results started coming through but discovered that “our data integrity was better than we expected.”

Very quickly, ASOS discovered a challenge around returns. As an online-only company, ASOS doesn’t want to discourage customers from feeling free to return items, but returns that “vary wildly” across categories can create forecasting challenges, Metcalf said, and leads to unusual sales profiles. Typically, there’s an initial surge of sales, then stock levels drop to the point of running out, at which point sales naturally drop off. But then customer returns start rolling in, leading to a “secondary bounce of sales,” Metcalf added.

“An engine like CO needs to expect that,” he said. “You have to tune the algorithm for that.”

As a result, the biggest tweak ASOS made to CO was adding in the ability to forecast returns.

Partridge said some of the learnings from CO have been surprising—and have challenged long-standing assumptions. Athletic brands, seen as perennially and highly desirable, weren’t thought to be a prime candidate for markdowns, though ASOS discovered that to meet sales target, discounts would need to be applied sooner than expected. “The engine takes away subjectivity,” she explained.

Using CO, employees are now freed up to manage exceptions rather than spending hours on time-intensive data collection to execute markdowns.

To continue delivering useful recommendations, the algorithm that powers the CO engine is a “living, breathing thing” that must be updated continuously, Metcalf said. “It’s not a system you can just set and forget,” Partridge said.