kikovic - stock.adobe.com
The tail end of 2018 saw online retailer Asos release Fit Assistant, one of several digital services the brand has introduced in the last 12 months.
Fit Assistant joined a number of recommendation services launched by Asos, focused on using machine learning and artificial intelligence (AI) to improve the customer experience, including personalisation features such as Your Edit, which shows products it thinks a customer might like; Style Match, which shows customers products similar to those they have previously bought or looked at; and its curated carousel “You might also like”.
But Asos’s digital product director, Andy Berks, points out these new features are not just digital for digital’s sake. “It’s part of a bigger thing,” he says. “It’s one feature that’s solving one customer problem, or trying to make one part of the customer experience better.”
Many retailers are focused on adopting third-party services, such as virtual fitting room startup Metail, in a bid to reduce the number of customer returns, a customer behaviour that can cost retailers a lot of money.
But Berks says the idea for Fit Assistant came out of customer needs rather than the business problem of reducing returns, highlighting the desire to implement new technologies to meet a consumer requirement, not just because you can.
Introducing a sizing service may well drive down returns of items that don’t fit, but, as Berks points out, returns are a result of “many things”, not just ill-fitting garments.
Instead, services introduced by the retailer are focused on “minimising potential disappointment” as well as making the customer journey more effortless.
“We’re a business, so obviously the things we do hopefully lead to us benefiting and growing, but we do that through ensuring the customer problems are solved,” he says.
“If you looked at it through the lens of returns, you might not end up making good decisions for the customer: you might end up doing things that help you. But if you always look at it through the lens of the customer, there may be benefits down the road. We always start from a customer perspective, it will lead us to a good place.”
A data-driven retailer
Much like other retailers focused on offering a personalised shopping experience, many of Asos’s recently launched services are underpinned by data and AI.
While Berks says AI is a “buzzword” used to describe a set of technologies that have been widely used for a long time, he also admits ASOS has been experimenting with AI as people currently understand it for the past few years.
“Over the past six to 12 months, we’ve really stepped up our efforts on that,” he says. “I think we’ve reached a point where we’ve started to see the value it can drive. We’ve started to release things like Fit Assistant.”
Now the digital retailer is investing more in AI and machine learning, “both from a scientist research side and an engineering technology side, to make AI something that can power the whole experience”, says Berks.
Both roles are equally important in delivering relevant services, as there is a possibility that a capability could be built by engineers without the ability to change the customer experience at all, due to a lack of data science knowledge. “That’s an engineering challenge as much as it is a data science challenge,” says Berks.
As an online retailer, Asos has always been considered by the industry as “digital first”, and with upwards of 18 million customers at the end of 2018, Berks points out the “huge technology ecosystem” needed to serve its market, as well as ensure features are scalable, responsive and resilient.
Asos has been building its teams to give them “engineering disciplines and processes”. This has enabled them to develop and scale services that solve customer problems, and ensure data is readily available to its scientists to either develop technologies that solve a customer problem or improve the
“[It also allows them to] start to influence a part of the business in such a way it previously might have struggled to do without the use of those technologies,” says Berks.
While many retailers are just discovering how they can use data for customer experience and personalisation, Asos has been using data from day one, and much of its business relies upon it.
“We’ve always used data across the business to empower our decisions,” says Berks. “That’s part of who we are as an online retailer. These technologies just allow you to do those at a greater scale.”
Making sure services fit customer needs
For many of its services, such as Fit Assistant, the retailer took an agile approach to development and launch, running a beta across its web platform to see whether there was an improvement to customer experience. Once a benefit was proven, the service was scaled, all feeding back into keeping customers happy.
Proving a service will be beneficial to customers can be a challenge in itself, especially when the focus is on delivering a good customer experience as opposed to measuring progress by profit margin or increased sales.
When launching Fit Assistant, Berks says it was obvious that helping customers choose the right size the first time would improve their experience of the brand.
“We stock between 800 and 900 brands on the site. We shouldn’t expect the customer to know what size they are in each of those brands, that’s not realistic, so we can help them with that through this sort of technology.”
Other innovations, such as using voice-driven commerce, take more experimenting to discover what works for customers and what doesn’t.
The brand initially launched its AI shopping guide, Enki, as a Facebook-based messenger bot, through which customers could use voice messages to chat directly with the retailer. This evolved into using Google Assistant to allow customers to browse Enki using voice commands.
Berks says these capabilities were tested and experimented with over an extended period of time, and are powered by AI and machine learning technology.
Testing new features
Part of testing whether or not a new feature is scaled sometimes comes down to a community of 4,000 customers called Front Row. These customers fit into a number of different market categories and are asked through hosted discussions whether certain features would be beneficial or not, and which customers would use them.
“Then we do things through testing. We’ll always try to get stuff out in front of a section of customers first so that we get real feedback. You can’t always know until you get something out into the app or on to the website how it’s going to land or how customers are going to use it,” says Berks.
Sometimes services are tested through a random, controlled beta experiment, where users are not told they are part of a beta to see how they react to, and use, potential new features.
Andy Berks, Asos
This is especially true of features such as early versions of Enki, where the customer problem the service is solving may not be as clear as it is for features such as Fit Assistant.
“With something like Fit Assistant, the customer problem is very clear,” he says. “We don’t necessarily know that what we put in will solve it at first, so we might have to iterate to make it perfect, but we’re pretty clear we’re solving a known problem.
“Whereas Enki, especially in the Google Assistant ecosystem, is a new space. It’s such an emerging domain that we don’t know exactly how we will be able to bring value to customers. It might not be what we originally think it is – customers might want something completely different from it.”
Curation and personalisation
Social media can be a huge channel for many retailers, allowing them to connect with customers and learn more about what the market wants.
Off the back of consumer chatter over social media, Asos ran with the idea of launching Boards, a service whereby customers can save and curate items they have seen on Asos into collections, such as products someone wants to buy for an event or a list dedicated to gifts.
“All of these different touchpoints have helped us validate that [the service] was going to be relevant,” says Berks.
This sometimes means some services are found not to fit the needs of the customer, which is why iteration and an agile approach to new products is important.
“Our product managers are spending a lot of time looking through those different sources of data and trying to find what the customer is telling us,” says Berks. “If you create something, get it out early and try to be fast about it, customers will tell you if it’s something useful or not. This is especially true in the voice ecosystem, where they’ll ask the questions they want you to answer.”
Sometimes it takes a few iterations of a technology to develop something that meets customer needs, underpinned by Asos’s focus on its customers to ensure they are tackling the right areas in the vast sector of retail technology.
“If you’re customer focused, you should be starting to solve those problems early, because you should be spotting where the opportunities are,” says Berks.
“You can be powered by technology, but you have to believe in the power of technology to be something brilliant and solve problems for the future. You have to have that spirit of experimentation.”
Read more about retail technology
- New West End Company hopes data on consumer behaviour will help retailers to increase spending in the area by one-third to £12bn over the next few years.
- Swedish furniture brand Ikea will be cutting jobs and focusing on smaller stores and its digital pipeline in a bid to usher in a new era of business transformation.