The pandemic has upped the ante for online merchandising

The challenge that brands and retailers face of shifting more of their sales online is huge. Supermarkets have had to invest in urban fulfilment centres in order to profitably deliver at scale. FMCG brands have done more through Amazon and new direct-to-consumer routes to market. Fashion retailers have had to quickly shift their focus to new categories and inspire customers through a screen.

These are admittedly trends that were developing anyway, but the pace of change in 2020 was well recorded and dramatic.

Success in ecommerce requires a strong brand, not just a smooth experience

In Autumn of 2020, I spoke to Kristen Miller, co-founder and CEO at Stylyze, a ‘merchandising-as-a-service’ platform in the fashion, beauty and home sectors. She commented that “a lot of the traditional omnichannel strategies and experiences that, pre-covid, were nice to have are now ‘must haves’”. This “acceleration in the adoption of new technologies and strategies” is necessary now, according to Miller, because “businesses need to figure out how to reach customers and be relevant”.

Where store closures have affected the fashion industry, this is particularly true. Retailers are having to provide a certain level of experience online, remain as profitable as possible, and react to fast-moving trends. This is the sweet spot where merchandising intelligence can theoretically make a big difference.

More agile merchandising

According to Miller, the typical refrain from their customers at the moment is “how do we get the right balance between having interesting products that customers want to see, but at the inventory levels that we can support. And how do we promote owned brands that we make more money from.”

Merchandising has become more agile over the past few years, with shorter product cycles and better ‘storytelling’, and retailers including M&S have sought to decrease the size of some ranges to help with saliency. But ecommerce poses unique challenges, especially for big retailers where large product catalogues may be advantageous for customer acquisition via search but difficult to wrangle and navigate when on-site. Machine learning is stepping into this breach and helping retailers with their inventory strategy.

In simple terms, Miller and her co-founder hypothesised that “what the retail vertical could really benefit from was some sort of digital stylist at scale”. So, where a sales associate can show you something similar when a product is not available in your size, or can curate products that fit a particular style, the tech can do this across a website (or email or ad).

Automating this is particularly useful in such a strange time, when event-driven sales cycles such as the ‘wedding season’ have been disrupted and even medium-term planning is difficult.

Retailers are having to provide a certain level of experience online, remain as profitable as possible, and react to fast-moving trends.

The profitability conundrum is one that recommendation algorithms have been tackling for a while – promoting products with higher margins. Miller gives an example of how this strategy works: “Maybe somebody is looking at Q-tips but we have the ability to tell a beauty story – that strategy has been very effective for mass retailers.”

But it’s the visual element of merchandising solutions that is newer and shows the potential of the technology. “We actually spin up looks or rooms or beauty assortments in real-time… as a catalogue is churning and products are going in and out of stock,” says Miller. She adds that “in a big catalogue we could have 500,000 to 1 million looks churning every single day and being used across the channel – website, Google ads, post-purchase, on mobile, or in store when working with associates.”

Style teams are at the start of this process, and they create a base set of looks across a catalogue that informs the machine learning platform. As the resulting looks get surfaced to an end customer, the engagement data such as clickthroughs and adds-to-bag can be used to optimise recommendations.

Replacing stylists

Though ‘complete the look’ type content has been used on homepages and category pages for years, Miller says that you “could spend huge investment, taking a tonne of time to get merchants or stylists to put these looks together, and by the time you get them on the website some of the products are unavailable – or it simply degrades over time”.

This ability to automatically understand the “aesthetic profile of a product”, as Miller puts it, is useful in situations where you may not know a lot about a shopper on your site. “There’s been this gap where [retailers] might know a customer is looking at sofas and then choose to recommend some relevant products, without understanding that she’s looking specifically at mid-century modern sofas in this colour range. That’s the opportunity.”

The appeal of this tech to customers is evident in the efficacy of the ‘view similar’ button that some ecommerce sites have on product listing or details pages. Miller earmarks this implementation as producing great results for their clients with impact on conversion, increases in average order value and even units per transaction.

Big ecommerce players such as Asos have taken this type of tech to its logical conclusion, with app users able to take a picture of an item of clothing and find similar items in the catalogue. It’s a matter for debate whether a mix of computer vision and machine learning can ever replace the role of the stylist – one would guess this is a long way off, and perhaps more of a question for my colleagues at Creative Review or Design Week.

More pressing, especially in luxury fashion, is the question of adapting to omnichannel retailing. In a sector where the sales process is high touch, as are the products, taking the store away “leaves a rudimentary website with a linear catalogue tree, and that’s really constraining and it doesn’t deliver that [experiential] promise” says Miller.

Using a merchandising tool to help with ‘clientelling’ from home is a neat solution, and Stylyze has been doing this with Neiman Marcus, rolling out to over 5,000 store associates last year. It’s a solution that can also work in-store.

But as Covid-19 continues to impact the high street and retailers get to grips with less of a physical presence in many shoppers’ lives, creating ecommerce experiences that both inspire and convert is a fascinating challenge. At Econsultancy we’ve seen near vertical uptake of our elearning, live-learning online workshops and skills mapping in this area, and expect this to only grow in 2021 as businesses from every sector, B2C and B2B alike, get used to life online.