Shop Direct believes that fostering a culture of experimentation and failing fast has helped transform the former catalogue business into a leading digital retailer.
The retailer, which owns the brands Very.co.uk and Littlewoods, started to move from being a catalogue to an online retailer in 2007. At the time, just 20 percent of sales were from online, and the company sent out 25 million catalogues a year. The company was also "business-led", the company's head of ecommerce, Paul Hornby, told an Oracle Retail conference in London.
Today, Shop Direct reports annual sales of £1.7 billion, with 84 percent of sales from online. Some 44 percent of sales are made via mobile, it counts 950,000 visits to its website each day and sends out two million catalogues a year. The business is also now "customer-centric" and reported its first profit in a decade in October last year.
"The biggest thing we needed to do was move away from the business-led and to this customer-led strategy," Hornby said.
"One of the big things of the culture change was recording the impact of what we do. We try to validate the impact of what we do and we decided to run experimentation to do that."
He added: "When it was business-driven, failure was not an option. Now, we genuinely encourage this culture of fast-fail. We develop as a far smaller piece of work, understand impact on customer and measure that."
This way of working is a significant change to the traditional waterfall approach that Shop Direct used to have, which involved developing large, 18-month projects for the business' internal stakeholders, and spending just 15 days retrofitting the user experience to what customers might want.
Shop Direct runs 55 experiments, which it refers to as 'challengers', a month on its website as part of its "desire to remove friction for the customer". It plans to increase this to 140 challengers a month by July 2016, which is more than a consumer would see on a competitor's website in a year, Hornby said.
It does this by coming up with hypotheses in advance, and splitting the web traffic, for example, via A/B testing, to try out a change to the website.
Examples of this include replacing pagination with scrolling (expected to be positive, but in testing failed) and showing a pop-up window displaying a discount or offer if a customer appeared to be abandoning the website. This experiment failed because customers were being interrupted when they may have just been opening a new tab to copy and paste the link, rather than to leave the site. Hornby described this experiment, which was live for just a day or two, as an example of a "fast fail".
Sign up for MIS Asia eNewsletters.