How retail uses machine learning to increase revenue

Artificial intelligence could help retailers set better and more competitive prices

How retail uses machine learning to increase revenue

The retail market is becoming increasingly competitive. Customers are expecting more personalised offers and are also more aware of their choices. Operational costs are rising and the amounts of data retailers need to factor in when setting prices are accumulating nonstop. As a result, businesses are hunting for new strategies to increase revenue. 

What’s going on in retail pricing? 

Currently, the pricing process is in chaos. Even mature retail enterprises, which have in-house data collection algorithms and a robust pricing system, tend to copy pricing and promotional moves of their competitors. Managers are faced with the necessity to take into account thousands of factors for millions of products. Also, as there is no unified database with the results of their past pricing decisions, they’re unable to forecast the outcomes of the current and future ones. As a result, they usually base their decisions on pricing factors only when setting prices.

How algorithms raise revenue in consumer electronics 

There is much more to take into account when offering competitive prices. Price optimisation considers price elasticity, non-pricing factors like the weather, grace periods and many other variables. To analyse all this data, retail teams need to be empowered by technology. 

Machine learning, the core element of artificial intelligence, is steadily unfolding in the retail market and is poised to reshape its landscape as well as customer behaviour, says Deloitte’s recent report. It gives the ability to analyse massive amounts of data quickly and correctly and get valuable insights, which makes it far more than merely a compelling option. It’s becoming a must for retailers willing to stay ahead of the market.

Advanced retailers are already embracing the power of machine learning pricing algorithms. Foxtrot, an omnichannel consumer electronics retailer, used AI-driven price optimisation software to maximise revenue without a loss in profit margin. As a result, the revenue of the retailer and the number of transactions surged by 16% and 13.6% respectively. Meanwhile, the average check grew by 12.9%.

Retail managers can be skeptical about counterintuitive price recommendations made by algorithms. However, the final metrics – market share, profitability and turnover – are the only things that matters. Nikolay Savin, head of product at Competera, the pricing platform, shared the outcome of recent undertakings using algorithms: “In a pricing experiment, the algorithm and the manager received the same goal to increase revenue without losing margin. The manager decided cutting prices was the most logical choice. The profit margin was maintained at 47%. At the same time, the algorithms suggested raising the price. The profit margin was maintained at 98.5%.”  

What is the main challenge of implementing a price optimisation system?

Inconsistent or incomplete historical data is the main obstacle for retailers when using algorithms. This can happen for the following reasons: the format of the data has changed over time; the data has been collected for several different objectives; the retailer has just entered the market so the data is not full enough and/or the data covers flash sales and is heterogeneous. 

If for some reason the retailer lacks data to teach the algorithm, the company has to squeeze everything from the available data and simulate or buy the missing data. In addition, regardless of the source, the data has to be in a single format and span no less than a year. 


The market is intensive, while the consumer demands more relevant offers. In order to win the customer by crafting personalised prices, retail teams need a technological solution which can process increasingly growing amounts of data. To optimise prices, retailers should prepare data covering ideally one or two years. The right format of the data and its structure are crucial for final price calculations. 

Alexandr Galkin
Alexandr Galkin

Share via
Copy link