What types of decisions need making in retail? Retail head office managers need to think about whether sales are in line with plans, whether market share is growing, what stock needs ordering, the performance of new products, any markdowns and rebuys required, customer satisfaction levels, and new product launches – and that’s just on Monday mornings!

We have seen advances in how we report and visualise data so that managers no longer spend Mondays gathering and crunching data. For many of our customers, SAS enables this. We were able to provide more time for making decisions. Now the question is how we can involve machines to help us make the information predictive? For example, can they start to make suggestions about what to do to hit particular goals? And if machines take this role, what will be the impact on resources?

What changes will this create?

Outwardly, this system could look very similar. But underneath, it is likely to improve the customer experience significantly. I certainly think there will be no adverse effects for customers. You may recall the initial fears that RFID technology would affect customers and their data, but its use is now standard, and greatly improves customer convenience. Our customers confirmed that it really does not matter to them whether the selection of goods, pricing or advertising are chosen by machine or a human buyer. Except that machines drawing on data may have a better idea of what customers want!

However, increased use of machines will minimise HQ functions and roles. This is already happening, driven by rising costs and squeezed margins. It will also mean faster decisions and changes in the ranges on offer.

The new retail

Again, this is already happening. In Manchester, the online fashion retailer In The Style has a two-week drop new range in fashion, with a 75% sell-through. The increasing number of convenience-sized units means that a higher proportion of ranges change more frequently. Aldi and Lidl have shown the way, with their weekly specials and smaller range (3,500 SKUs vs. 25,000 in an average supermarket), resulting in a smaller average basket but high footfall.

The use of machines should also mean more personalised offers and range. A good example of this is the Very group, an internet shop with personalised assortments and offers tailored to individual preferences, and delivered to mobile devices to improve conversion.

Finally, we should also see retail becoming more sustainable, with less waste resulting from mistakes and markdowns. Historically we have seen higher margins to enable businesses to afford the waste and markdowns. Will we now see a consumer tolerance measure, or even new legislation?

Drivers and risks from a SAS point of view

There are a number of pressures driving retail towards the use of machines and analytics. These include competition, customer demands and the resulting profit squeeze. It seems unlikely that changes in retail will be driven simply by availability and capability of technology. Generally, there needs to be a better reason. Adoption may, however, be driven by results, especially if the technology starts to perform better than its human equivalent, or – more likely in my opinion – in combination with people.

It seems unlikely that people will ever be removed completely from the equation. History tells us that a hybrid approach, combining people and machines, is usually stronger than either alone. However, we may see a shift in creativity from the retailer back up the supply chain to manufacturers and producers. We may also see the differences between offers and ranges shrink – but perhaps that is where the human input will become more important. After all, nobody wants to look exactly like everyone else.

There are also risks that the model will degrade over time. This will mean that customers gradually start to see less that they like. As always, machine learning and modelling will require maintenance and checks for bias. Over time, though, I think we may well see retailers becoming technology companies. Arguably Amazon has shown us the way, and this may be where many others are heading.

Adding AI and machine learning

I suggest that it is time to see what AI and machine learning could add to your retail organisation. To drive decisions, organisations must operationalise the results of analytics. This will help you to start creating business value from data. Retail organisations will certainly not be wiped out by AI. Instead, we can augment existing abilities, and increase efficiency, by drawing on some of the best brains in the industry.