In an ever-changing consumer centered retail industry, it’s an obvious trend that brick-and-mortar stores are suffering. Despite all the buzz about the collapse of the physical, 80% of purchases are still made in stores, and perhaps customers are not changing as much as initially thought. Previously, a customer might try on three shirts at a store and purchase one; now they’ll order all three, and return two. The process is the same, but the latter results in a staggering cost to retailers. Returns are hitting the clothing industry hard, and companies have to get creative about decentivising them. A German brand began showing customers the CO2 emissions that would result from their returns, and saw the volume of returns decrease significantly. But who is actually driving change: the consumers or the companies? No consumer would have expected 2-day delivery before Amazon started offering it, and it has now become the standard.
Consumers are continuing to bamboozle retailers with their inconsistent loyalty. Sometimes people are dedicated patrons of a particular grocery store, but show no preference beyond speed when it comes to getting lunch, or coffee before work. The idea that the end goal dictates how an individual shop challenges the “segment of 1” theory by suggesting that a personalised shopping experience may look different depending on their “shopping mission,” rather than one optimised experience for a specific individual. All of the above are relevant concerns for companies who are trying to improve their sales processes.
Where could AI and machine learning help?
The millions of small, repetitive decisions that a retailer has to make every day is a ripe place for machine learning to take over. It would be hard to find someone who would argue that a question like “should we expand into the Chinese market?” should be decided by an algorithm than by a human, but machine learning can provide valuable data to consult in answer to that question. AI software can be set up as to not be overzealous; it’ll manage the day-to-day tasks, but still flag decisions that need human approval before proceeding.
Machine learning would streamline stock management where the costs vary depending on the industry. For example, overstock at a grocery store means a lot of money wasted because the perishable food has to be thrown out, whereas having insufficient stock could be more costly than overstock for a clothing retailer. It’s also able to track vast amounts of variables that manpower cannot, like the numerous factors that go into a single purchasing decision. AI would be able to calculate the optimal amount of stock for products to hold at any one given time and therefore save the company from waste and extra cost.
It’s important to note that having a dirty data set means machine learning is unlikely to produce useful conclusions. How to go about getting clean data? First, never throw out old data, but more importantly start collecting data with a purpose: for example, by how often a consumer clicks on related products. From this, machine learning can produce predictive analytics, and ultimately become a self-learning supply chain.
What could the future of data harnessing look like?
Although AI provides an optimal way of creating an intelligent supply chain, the reality of such extensive data sharing seems unlikely. Retailers and suppliers would not want to share data because generally price negotiations take place, and open data access would give one party leverage over the other. Similarly, retailers would not want their competitors to be able to access their customer and sales data out of fear that others would exploit their weaknesses. The predicted outcome is that ecosystems would emerge that are more likely to be made up of companies from different industries who can share data to assist each other without the fear that competitors will take advantage of them.
Internally, the adoption of AI within a company would benefit greatly from the presence of a data scientist who can speak both the business language as well as the technical understanding about how the machine learning system works and the best ways to utilise it. Connecting the data to the story within it is a crucial bridge for companies to better understand their problems and where machine learning could help provide a solution.
The Smart Retail Works Breakfast was held in partnership with JDA, a provider of end-to-end, integrated retail and supply chain planning and execution solutions.