Shop Direct, the owner of Littlewoods and Very, has improved site availability from an "abysmal" 57 percent to 99.99 percent after re-architecting its front-end ecommerce platform on Amazon Web Service’s public cloud.

Shop Direct, the owner of Littlewoods and Very, has improved site availability from an "abysmal" 57 percent to 99.99 percent after re-architecting its front-end ecommerce platform on Amazon Web Service’s public cloud.

In recent years, the retailer has undergone a transformation of its business, successfully shifting from catalogue sales to becoming a fully fledged digital retailer. This has placed more importance on its customer-facing systems and underlying infrastructure as it competes in a multi-billion pound retail market.

According to Shop Direct CIO Andy Wolfe, the migration of its main Oracle ATG ecommerce platform to Amazon Web Service's public cloud played a key role in improving customer experience across its retail websites, which had not always met expectations in the past.

“When approaching our sales peak in December 2012 we achieved about 57 percent [site] availability, which is absolutely abysmal for a digital retailer,” Wolfe said at a recent event. "Last year I am pleased to say we achieved 99.99 percent, right across every single one of our sites. In part that was down to a lot of the work we have done with AWS in re-architecting our front-end platform.”

Legacy architecture problems

The Oracle ATG ecommerce platform is a core part of Shop Direct’s retail operation, supporting around £1.7 billion transactions across the business each year.


The company had previously relied on an on-premise deployment of ATG, version 7, which was integrated into legacy back-end systems, creating problems with application performance.  

When deciding to upgrade to the latest version of the software, Shop Direct chose to implement it in the cloud and is now running ATG version 10, using between 400 to 600 EC2 instances depending on peaks in demand.

“One of the biggest reasons for us deploying ATG in AWS was service reliability. We had a number of problems: we have a fairly eclectic architecture, as you can imagine for a retailer that spans 80 years. There are some interesting applications that I have to deal with on a daily basis,” said Wolfe.

“Our previous version of ATG was heavily entwined with a lot of that legacy architecture. As we approached peak trading periods, we found that problems in the back end were resulting in outages in the front-end. We had to do something about that - you can't be a world class retailer if your shop is shut.”

The migration had immediate benefits for the company, such as enabling a substantial increase in the amount of business it was able to transact.

“On Black Friday last year we did 10,000 orders per hour across one of our brands, Very. We have never previously done that order rate across all of our brands together, let alone one. That availability and increased promotional activity helped transform our business, and manifested itself in our bottom line, in terms of online sales growing by nearly 14 percent.”

Configuring application for the cloud

Migrating a major production application to the public cloud wasn't without its challenges, however.

Wolfe explained: “AWS is a fantastic platform, but when you couple that with an application that wasn't engineered for cloud you start to have some interesting challenges. It was really hard work. It took a lot of support from Oracle and from AWS to be able to configure ATG to scale for our business in the cloud.”

The company has a number of other systems in the cloud already. This ranges from Amazon S3 storage to Microsoft Office 365 and Sharepoint tools, as well as onsite search and navigation platform Endeca, and WebLogic middleware, which sits in the cloud to support back-end integration with the ecommerce platform.

However, the public cloud is not seen as a panacea, said Wolfe, and there are no plans for a wholesale move of its core applications in the near future. Decisions will be made on an application by application basis where there is a clear benefit, with customer-facing systems making the most sense for the business, he said.

“It is not a ‘cloud for clouds sake’ decision, everything else lends itself well to fixed infrastructure,” he told ComputerworldUK.

“We have over 80 applications which support our business, and some of those applications won’t suit being moved to the cloud.”  

Next section: Improving user experience

Improving user experience

AWS is also being used for dev and test purposes, helping to drive user experience differentiation across the various retail sites.

Previously the company had “lots of constraints” with its dev and test environments, which had slowed its ability to innovate and react to customer preferences.

“If you are going to differentiate as a retailer you have to be able to develop very quickly,” he said.

“We run about 35 experiments every month on the look and feel of the site, navigation, how we present content to customers, and we are looking to extend that. We couldn't do that if we didn't have the infrastructure to allow us to scale.”

Data analytics and social media analysis

Shop Direct now expects to expand its use of public cloud resources to enable additional functionality for its data analytics capabilities. This will involve utilising the scalability of the cloud and tools such as AWS’ Hadoop service Amazon Elastic MapReduce to gain more insight into customer behaviour.

“Fundamentally the cloud gives the ability to take in more data, and with in-memory computing it allows us to process large volumes of information and work out new insights that we can apply our algorithms to, helping us provide a more relevant personal experience to our customers,” he added.

For example, Shop Direct intends to make greater use of unstructured data, such as information gathered from Facebook, to drive decision-making around sales and services.

“The plan is to create more personalised experiences for our customers, and to leverage the huge amount of data we have today which sits in a Teradata data warehouse,” he said.

“We will also leverage more unstructured data - more Facebook, more Open Graph, and apply analytics to that so we can start to do some decisioning with this. We are looking at a big application and infrastructure deployment in that space over the next 12 to 18 months.”

Part of this data analytics strategy will involve leveraging the scalability of the cloud through tools such as AWS’ Hadoop service Elastic MapReduce (EMR) to gain more insight into customer behaviour.

“Our data coupled with Facebook data can give you unique insights and you can start to see trends that you wouldn't necessarily normally see, for example in terms of what colours are trending the most today and how that may impact what we buy in future terms of our portfolio.”

“It is that type of data that we think will be valuable to us in terms of the decisions that we make when presenting products or proposition to customers.”

Location data

The company also aims to be able to provide a more personal experience by understanding its customers better. This could involve drawing on site browsing data, whether Shop Direct sites or others, to offer specific vouchers or discount codes.

Enabling location-aware services could also allow the company to target customers with offers when they enter a rival’s store.

“Hopefully at some point we will be able to look at where a customer is in relation to their shopping behaviour. So if they have opened up our website, and we know from location that they have opted in that they are in John Lewis, then maybe we will present them with an option to come back to Very with a discount code. It is those types of things we are looking to leverage.”