Why Arcadia turned to AI to simplify the personalisation process

Artificial intelligence may be the biggest tech buzz phrase around at the moment, but it’s already making an impact at Arcadia group.

During a presentation at GITEX 2018 in Dubai, the multichannel director of Miss Selfridge, Gareth Rees-John, explained how in his previous role as digital director of sister brand Topman, the company underwent a journey which started with the need for big data analytics and has now led to the introduction of AI.

Rees-John told delegates that about five years ago at Topman, the company decided it would try to improve the way it personalised the website and content for its customers.

It started with simple things such as sending different kinds of communication material to men and women – and this led to click-throughs doubling and open rates increasing. However, he said it would soon get a lot more complex due to the abundance of data at its disposal.

“We then did a survey of 3,000 customers with 50 questions such as the clothing they buy, their influences, their thoughts on technology and more, and then we separated them into five different categories which ranged from ‘not bothered’ to ‘extreme fashionista’,” he said.

Then, taking advice from Steve Krug’s book Rocket Surgery Made Easy, Rees-John said he wanted to ensure that the categories were accurate, so Topman brought in people to the office who fitted the mould of these five categories.

After feeling as if they had the categories right, the company sent out another quiz to customers.

“It was a competition and we engaged with thousands of people to fill this in and again built segments in – and now we knew for sure that these segments existed as we had them in the office, had the data before and the survey afterwards,” Rees-John explained.

Programmatic advertising

The company then opted to send programmatic ads to the different segments of customers, so when a consumer was shopping on a sports website they would get an ad for clothes that fit into the ‘not bothered’ Topman range, while those searching for clothes at a more fashion-conscious website would get an ad for the ‘fashionista’ Topman range.

When they went to click through to this, the landing page would specifically fit their category too; so they would see the same model as the advert, and the same category of clothing.

While this revolutionised the way Topman personalised its advertising and shopping for consumers, Rees-John said it was very difficult to maintain.

“That’s what I find with personalisation, you get excited about it and then the financial director says you can’t do it anymore and it means going back to the drawing board"

“You have to continually maintain these experiences for customers, so it works but it was very hard, there wasn’t a way to do it without leveraging further cash,” he stated.

“That’s what I find with personalisation, you get excited about it and then the financial director says you can’t do it anymore and it means going back to the drawing board,” he added.

Other initiatives

Topman then dropped the idea of five categories, and did some AB testing. Things that helped to increase revenue included reminding people that they were VIP guests, incorporating student prices on all items when it recognised the user was a student, and personalising the experience depending on geography – this had a lot to do with the data the company held comparing what big ticket items people were buying in the north of England versus the south.

The company also used Granify, an eCommerce conversion and revenue optimisation technology that uses machine learning to encourage the retailer to speak to the customer at the most crucial time to tell them what else they could add to their basket or other items they may be interested in. Rees-John said different customers would be more likely to engage at different times based on their previous history.

AI implementation

Three years on from starting the ‘five categories’ approach, Topman now uses Exponea, an automated marketing platform that uses AI.

“This means instead of me having to build five different journeys to constantly update my data set, Exponea are working with us to look at our data and deciding when to automate the process,” he said.

Previously, the company had to manually update the landing pages with relevant products for each category.

“That’s what’s exciting – it’s about personalising the experience,” he stated.

While AI has been a success so far for Arcadia, Rees-John said that fully integrating AI is very difficult but advised companies to start by incorporating the tech into customer relationship management (CRM) systems first.

The biggest challenge isn’t necessarily technology-based.

“A lot of these AI companies are new and you have to persuade your board to take a chance on start-ups who aren’t big like ourselves; the board are worried about working with these guys but as long as the data is safe the risk is worth taking,” he said.