Anyone with experience of working in a fashion retail store in the early 1990s will remember the hallowed role of the window dresser, or visual merchandiser.
Armed with an iron, a special board for ironing sleeves and lots of pins, they would roam from store to store and held demi-god status in the eyes of store staff, many of whom thought: “Maybe, one day.”
Back then there was a considerable degree of art involved in visual merchandising, depending who you asked the window had to grab attention and interest in between five and seven seconds.
The rise of data and the eCommerce challenge
As epos systems became more sophisticated and information was polled more frequently, data increasingly informed merchandising decisions. Lists were faxed to branches with instructions to front face bestsellers in store and place trending items in the window: right product in the right place at the right time.
In the eCommerce world the online merchandiser’s role is almost entirely data driven. Their tools are databases, spreadsheets and analytics software. What’s more, whilst their bricks and mortar colleagues have limited space to hold stock, the website has a virtually unlimited stock holding.
That’s not the only difference of course. For instance, a shopper can walk into a branch of Next or Marks & Spencer and pretty quickly scan the entire range across different product groups – their speed is dictated by the square footage of the store. But online they are confined to one page, be it the home page or other landing page, which severely reduces the opportunity to expose that massive inventory. Ensuring the right product is in the right place at the right time is more important than ever.
Furthermore when you add in the volume of data collected through interactions with a website that lead to micro conversions, the task of completely merchandising an eCommerce site for optimum ROI is at best overwhelming and at worst impossible.
Back in August when asked about the adoption of emerging technology at the Retail Hive Live: eCommerce meets Fulfilment, members stated that AI was their biggest priority above Robotics, IoT, Self-driving vehicles and Drones.
AI for online merchandising?
There are many eCommerce personalisation solutions that claim to employ AI, so it’s worth taking a look under the hood of what’s out there.
Many systems described as AI are, in fact, rule based. They require the merchandising team to set up triggers based on criteria defined by poring over spreadsheets and other data sources. But rules can conflict, they are not truly scalable, they do not react in real time and creating them requires a lot of manual effort – back to square one.
What’s more, many systems deal with just one part of a site – search or recommendations or ads or navigation. This means retailers can have up to four systems in effect cannibalising promotions from each other as they each run using independent logic and data sets, for example; a shopper searches for an item they know they want, they get the result and the recommendation shows them a lower margin alternative.
A sophisticated AI solution is self-learning, self-operating, consistent, accurate, scalable and efficient. It will run across all aspects of the site, learning and applying that knowledge to all facets. In fact as data accumulates it just gets better and better.
This kind of solution should enable retailers to hand over the heavy lifting of merchandising to computers, to put the right product in the right place at the right time, leaving them to set their overarching goals. The question is; can they let go and is it a leap of faith?
Man vs Machine – the results
One online store that did let go is leading Nordic beauty retailer Kicks. But it was no leap of faith. It A/B tested merchandisers versus AI. On the one hand, merchandisers used their experience and knowledge to manually produce a home page. This page was pitched against an automated page delivered by Apptus eSales using AI and machine learning.
The results were striking. The automated version received an eight times higher click-through rate than the manual version, eight times more add to baskets and where the manual version made one actual purchase, the automated version succeeded in completing 101 sales transactions.
According to Jenny Vesterlund at Kicks: “It was not long before the results unambiguously spoke for the Apptus eSales version, we saw a strong increase in all indicators; click through rate, percentage added to the shopping cart, number of purchases, conversion rate and average order value.”
After a few days Kicks decided to completely abandon the manual approach and hand sales generation over to Apptus eSales.
“We’re looking forward to doing even more testing and use of automation in the future, especially now that we’ve seen how well it can work,” concludes Jenny.
Opportunities abound - no leap of faith
As in other business areas, AI in online merchandising really can make a difference compared to a manual approach, and a quite amazing difference too. As Kicks discovered, there’s no need to take a leap of faith, just a step forward in the right direction: There’s still the opportunity for merchandisers to achieve demi-god status.