River Island’s move to the cloud
While the pandemic drove many shoppers online, fashion retailer River Island has continued to invest in technology to boost its customers’ in-store experience.
Recent moves by the mid-market clothing store have included adding three high concept, technology-driven shopping spaces to its 300-strong roster of physical stores.
The first, River Studios, opened in the UK in Derby this August in a 11,000 sq. ft space, boasting smart changing rooms and an AI-style tool named Chloe.
Across all its stores, meanwhile, clunky single-use handheld devices for shop-based tasks (which staff had to share among them) have been replaced by Scandit-enabled smartphones for each of its 3,500 staff to answer shopper questions and function as a mobile point-of-sale.
Underpinning these innovations is the store’s technology division, River Island Tech – which looks after the networks supporting both in-store and online activities.
Speaking exclusively to TechInformed at the retail transformation fest IRX, the fashion brand’s head of cloud and infrastructure Tonino Greco says that not a clothing rack or a digital display is moved now without consulting IT.
“A lot of innovation from an in-store perspective usually now starts with the question ‘can IT support us?’ And usually that answer is ‘yes;’ Every now and again it is ‘yes, but…’ and very rarely it’s ‘no we can’t’.
“And it’s only ever really a ‘no’ when the product they’re trying to install requires an excessive amount of bandwidth – technologies such as AR screens, for instance, can be a challenge,” he admits.
Since Will Kernan joined the business as CEO in September 2019, the ideas have been coming thick and fast, but Greco adds that his new boss also requires every single one of these innovations – and the tech spend required – to be able to demonstrate value.
“All our spending needs to be justified for from a value perspective. For every pound we invest in a piece of tech, the business needs to work out what we need to sell to recover that amount,” he says.
“The business has done extensive work into what that needs to be, but in essence, whatever we spend needs to deliver the same type of value or better,” he adds.
This means that some profit and loss initiatives such as AI cameras and Radio Frequency Identification technologies which monitor stock inventory and the flow of people in and out of their stores, have been no-brainers from a cost perspective.
It’s also easy to justify the value in investing in SD-WAN to enhance the in-store experience by improving the network and supporting shopping apps and payment systems.
However, Greco points out that it is harder to allocate money to continuous improvements, like new switchers – or even, surprisingly, cyber security, because while necessary, he points out that these technologies and services are harder to quantify.
“We tend to slip ‘continual improvement stuff’ into these bigger networking projects,” he admits.
“I can’t say, you know, me putting in this new switch is going to lead to 10% more sales. So, we sometimes must deliver value on top of – or underneath – all the other projects that we deliver.”
“Sometimes that’s quite difficult, especially when it comes to security. It’s like how do you slip cyber into a project that the business is trying to deploy? It’s hard to explain sometimes why you need to have this observability platform to deploy a microservice that gives rewards to customers,” he explains.
Move to Cloud
One key area of investment in recent years has been River Island’s move to the cloud – which Greco estimates now saves the store £1.4m annually.
The retailer had already been an AWS customer for seven years when it started its move to cloud in March 2020, on the cusp of the pandemic.
The main challenge, according to Greco, was moving the store’s legacy systems into a cloud environment.
“Some were about five to ten years past their sell by date but still integral to the network – there were applications there that we didn’t even know the source code for,” he admits.
Rather than using APIs to bridge the old with the new, Greco says that they decided to ”do it the hard way.”
He notes: “We didn’t change the IP address of any of the servers we moved but we were able to extend our network into AWS’s using some smart technology.”
“In our case that required lots of stakeholders on the business side to tell us what functions servers were used for.
“Some of these servers had been used for years and had never failed – but it was about gathering the collective corporate knowledge to map out this information. In IT you hardly know these answers – it’s the users on the business side that do,” he explains.
Greco acknowledges that the job of finding this information from its store staff and other parts of the business, and then moving these servers, was made easier because of lockdown: “It would have taken twice as long if we had to move the servers out of hours,” he says.
He adds: “Then it was about creating a mind map of arrows: listing things that might fail and the things we needed to test etc. We got about 95% of that right. There were a couple of things that failed, but again, these failures happen all the time.”
Greco estimates that his team moved 300 servers in four months, and he claims that there was no downtime to any business applications “no one knew we’d moved them, they remained completely unaware”.
Not all the store’s assets are in the cloud yet and, according to Greco, the store is adopting a different approach to the ‘lift and shift’ process it deployed during the pandemic.
“Next up will be moving our non-production dev test systems and we’ll go through them all and ask stakeholders what actually needs replicating in cloud or what needs building better We estimate that this will take about a year with three months of planning four months of moving,” he says.
As we talk, River Island is approaching its busiest period as it gears up for Black Friday, Christmas and the January sales.
However, now that the store’s tech stack has been rewritten and the business has moved to the cloud, the focus has moved from scaling to operational maintenance.
“Now it’s all about autoscaling to handle the load – dynamically allocating resources to match performance requirements. This is less expensive to run and more elastic with all that good stuff that you get from cloud,” Greco explains.
“So, while our distribution centre goes into crazy mode and our stores start to hire 30% more staff, for us it’s just a heightened operations focus more than anything else.
“During this time, we’ll do a lot of techs drag – it’s about fixing all the stuff that’s been sitting on the backlog that no one’s really wanted to work on, because we’re not really changing anything on the production side,” he says.
Subscribe to our Editor's weekly newsletter