Online fashion house Asos is revolutionising the way its merchandising team set markdown prices during sales periods via a new system launched across the business earlier this month.

The clearance optimisation engine aims to help the online retailer set prices that will boost sell-through, ensure products are being sold at the most profitable level and free-up one of the busiest merchandising teams in the industry.

On a basic level, it is a calculation engine which ingests wide amounts of historical sales, price and seasonality data – combined with specific business rules set by Asos – to produce a table of markdowns and forecasting curves to meet expected demand.

Lucy Partridge, retail subject matter expert at Asos, said: “The tool removes the subjectivity [involved in traditional merchandising] – it’s much more analogical and analytical.

“We’ve been very impressed with the speed, accuracy and consistency of the tool. We can come in on a Monday morning, open up the tool and there we have the recommendations.”

She added: “It stops of merchandising teams having to spend so long pulling it all together in Excel and it allows them to focus on exceptions rather than having to analyse every single option.”

Asos, which currently offers 85,000 different products on its website, has set itself a product lifecycle target of 26 weeks and a sell-through rate of 95% within that period. It hopes the new tool will help meets its ambitions.

Oracle’s data science team and consulting division worked alongside the internal Asos IT team to create and implement the solution this year. The story of its development was detailed at the Oracle Retail Industry Forum 2017, in Barcelona, last week.

Chris Metcalf, merchandise planning tech programme manager at Asos, commented: “A selection of merchandisers got together and alongside Oracle we set a sensible set of business rules – you can specify things like minimum/maximum price, first markdowns, exit dates and sell-through targets.

“You then get an optimised markdown recommendation, so you maximise your margin by selling through the product by the target exit date.”

From concept to reality with speed

The idea for the tool was formed at the turn of the year, and has been launched across the Asos business in time for its mid-season sales period. A section of the business have tested and tweaked its functionality during two pilot programmes, and the merchandising team spent much of the summer being trained on the new system.

“Technically it’s actually quite a straightforward project because we’ve got no legacy optimisation solution because the business has been running on Excel,” explained Metcalf. “We haven’t had to rip out a system and replace it.”

“We decided to put this into the cloud. We’ve never had to deploy Oracle Retail systems in the cloud before and we’re not actually aware of another retailer who’s got clearance optimisation in the cloud but we felt it was the right thing to do and it’s in line with our tech strategy which is to be very cloud based.”

Complications lay in Asos’s unique trading patterns, which often comprise rapid sell-through of new lines before relatively high returns due to the nature of online fashion. A second surge in demand occurs once returns have been processed and re-merchandised.

Metcalf said this was the single most challenging thing the Oracle data scientists encountered when creating suitable algorithms for the new solution.

“We also saw an opportunity with this because of our major transformation planning with the full Oracle Retail stack next year, and this gave us real operational experience of running Oracle in the cloud,” Metcalf added.

Business-led, not tech-led

Partridge said the initial results of using the new solution have been encouraging, and she attributed much of this to the way the technology has been received and embraced internally.

A shift from Microsoft Excel-based merchandising decisions to machine-led markdowns would clearly be a major shift for any organisation, but it was a technology change made with the wider business in mind.

“We had great senior management support from the business and we had a strong sponsor in the head of merchandising,” noted Partridge.

“Because of her confidence in the tool it gave the rest of the business confidence. And it was also great for when the teams were challenging [the rules]. We stuck to the vision which helped us to deliver it successfully.”

Users representing different parts of Asos helped shape the solution from the beginning, and the pilot team contributed to training the wider organisation.

“It was a business-led programme rather than an IT-led programme. [The initial] users went on to become part of the pilot team, which was very important for ironing out the bugs and tweaking the rules and the user interface and getting the right look and feel.”

Visual search implications

Partridge said Asos sees itself as a technology company as much as it considers itself a fashion retailer – and that mind-set makes for innovative thinking across the business.

Visual search is the next big customer-facing technology launch on the horizon for Asos. The system, already available on iOS, allows shoppers to post a photograph of an item of clothing before listing the closest matches in the Asos directory – it will soon be rolled out on Android platforms too.

“This is going to change how shoppers interact and how they behave but we don’t actually know how – we’re launching it with a ‘let’s see what happens’ approach,” explained Metcalf.

“We should see new trends and patterns in the data that will start to manifest itself. It’ll have all sorts of unexpected knock-ons, and one of those things will be pricing.”

He added: “There may be opportunity for further optimisation of our pricing. We need to keep a close eye on what’s happening across our business and stay in tune with it so we can further improve our optimisation.”