High street retail hasn’t changed much in the last few decades.
Yep, there’s click and collect and online returns but, as in years gone by, product buyers decide what will sell by using a mix of nous and trends analysis.
Fashion, for example, may be getting faster (quicker production time and fulfilment) but the knack is still in predicting the season’s trends and riding the wave. In-store merchandising, too, is a matter of long-honed instincts as to what should go where.
Blending art and science
What I’m saying is there’s a lot of art in the high-street retail business (particularly fashion), and it attracts suitably artistic people. Yes, sales and seasonal analysis comes into it, but it’s a ways behind some of the technology emerging in online shopping, for example:
- Dynamic creative optimisation in retargeted advertising
- Automated merchandising optimisation (menus, sorts, categories)
- Visual product discovery, such as on Sunglass Hut’s website which recommends glasses that share visual affinity with pairs you have previously selected
- Conversational commerce, such as the much covered North Face online shop which asked customers where they were going and recommended suitable jackets
- A similar personal shopper style experience on 1-800-Flowers’ website which asks questions and recommends gifts
Whilst some of this ecommerce tech is still in its early days, automated merchandising optimisation is of particular interest. Ecommerce companies with big product catalogues (far bigger than stores can hold) are able to optimise sales by presenting products that each user is most likely to buy.
This is effectively the same job that a product buyer or retail analyst has, but machine learning may use data points that vary from the visual appearance of products to customer demographics or browsing history, from weather to time, from price to product descriptions.
The question is, why can’t this machine learning approach be applied to the high street store? Self-learning algorithms creating geographical segments and looking at lots of latent variables in order to choose what products are placed in store?
Obviously, the personalised aspect of ecommerce cannot wholly be replicated at scale in store, but what about the data-backed merchandising?
Well, I’m being a bit disingenuous, because there are companies that are already starting to look at high street product inventory and prices in this way.
Predicting trends, online to offline
What if a computer could ingest fashion magazines and influencer Instagram feeds, along with a fashion retailer’s first party data (who is buying what) and help that particular brand pick the styles for the upcoming season?
This is not quite happening right now, but an analytics company, Edited, is doing something similar, using natural language processing and computer vision to create a searchable database of millions of products from many brands. This database can be used to inform buying strategy, with brands able to investigate their competitor’s pricing and product assortments.
Stylumia is another company that offers something similar, analysing unstructured data and images to form trends analysis.
This surely hints at a future where ecommerce and social media is a sort of data playground, allowing brands to test certain products, and formulate the right plan for their stores, where (let’s not forget) the great majority of sales are made.
Once a business’ own consumer data is factored in, the technology may become even more powerful.
Illustration of the sort of data Edited compiles
An auto-merchandised high street store?
In a recent roundtable discussion at Econsultancy and Marketing Week’s Digital Therapy Live event, I spoke to some retailers who were intrigued about machine learning and its ability to drive commercial decision making.
What if real-time weather data, footfall and sales were used to merchandise a store each day. Could positions in the store be formalised in the data set, too? Could store tracking be used to analyse where people are browsing, and then add this into the algorithmic mix, too?
There is an obvious counter to many of these questions – would it really be that much more efficient than the brain of an expert human, and wouldn’t it be far too expensive?
At the moment, maybe these questions only make sense online, where data is more manageable. In an offline world, without an all-seeing computer eye understanding everything going on in a store, the number of variables involved may be prohibitive.
What’s much more likely, in the long run, is the concept that IBM Watson Marketing calls ‘augmented intelligence’. Rather than letting a computer optimise merchandising in stores, technology such as that provided by Edited will get more and more sophisticated and be used as an aid to human buyers and merchandising, cutting down on product gambles and costly mistakes, and making sure product assortments are statistically likely to sell.
It’s exciting times in retail.
For more on in-store tech, see: How Coca-Cola is using smartphone data to personalise in-store ads