Machine Learning for Retail Pricing: Is It Worth Investing In?

Press enter to search
Close search
Open Menu

Machine Learning for Retail Pricing: Is It Worth Investing In?

By Vladimir Kuchkanov - 10/25/2019

An additional $15.7 trillion — that’s how much AI and machine learning are expected to bring to the global economy by 2030. North America, including the US, is projected to be among the two regions which are likely to experience the biggest gain — a potential 14% boost in GDP, or $3.7 trillion. 

Retail is named among eight sectors which are poised to get the most out of AI adoption, thanks to the potential of artificial intelligence “to increase productivity, product quality and consumption.” It is good news for early adopters that are already using the technology or are quickly gearing up for it. It is a wake-up call for those that are still hesitating. The thing is if they don’t move fast and capitalize on the opportunities which AI offers, they risk losing it all to more adaptive competitors. 

“We think AI should be like electricity, it powers everything so we can serve the customers in a way they’d like to be served and to take the friction away,” stated 7Eleven’s VP of digital customer and store experience, Tarang Sethia, at Shoptalk 2019. Let’s dive deeper into what benefits artificial intelligence and its core element, machine learning, can bring to the retail table; as well as what bottlenecks are likely to emerge, and how to deal with them. 

Enhancing your pricing process

Among the most valuable improvements which machine learning ensures are the automation of routine tasks, enormous computational power which translates into precise and real-time insights, and better decision making for humans. When it comes to pricing, AI-powered price optimization software can boost sales by up to 24% and revenue by up to 16%.  

It is no secret that setting optimal prices today is a significant challenge even for advanced retailers. Retail teams need to factor in hundreds of parameters like cross-price elasticity and sales cannibalization to come up with the best prices (from a standpoint of retailers and customers) in near real time. This calls for the ability to process vast amounts of data, which goes way beyond human capabilities. 

Among the other challenges connected to pricing are human errors, low speed of repricing, too many ineffective promos, targeting wrong competitors, incorrectly identified KVIs and invisible price thresholds. All of them need to be tackled by pricing managers when making pricing decisions. 

Elasticity-based self-learning algorithms help to overcome all of the above challenges. They can process any amounts of data, take into account as many parameters as necessary, and reveal all the hidden relationships between products in the product portfolio to suggest individual prices which will altogether maximize revenue and sales of the entire portfolio as often as necessary for as long as needed. And all of this happens based on the goals retailers are pursuing at any particular moment. 

Making your team more productive

AI-powered improved productivity will account for the biggest economic uplift, or over 55% growth of GDP. Enhancing retail pricing teams is likely to “augment their capabilities and free them up to focus more stimulating and higher value-adding work” like pricing strategies and more beneficial negotiations with vendors. Machine learning speeds up repricing by five times, helping pricing managers to use the free time to boost their careers and even earn more as they can take up more products and categories to manage. 

Meanwhile, to make the most out of AI, retailers need to prepare their teams to work with machine learning in pricing in the right way. To begin with, managers need to take a series of workshops covering two aspects: technical interaction with the software and the ability to work with algorithm-generated recommendations. 

Retail managers often hesitate to apply AI-powered suggestions. This happens since the logic behind some recommendations can be unclear to retail teams. You may hear something like that: “I’ve been managing pricing for eight years and never have I ever made such decisions. What can you teach me?” But the thing is that managers need to learn to trust the algorithms (if they doubt machine-powered decisions, they can always test them via a “sandbox” or simply put relevant constraints) when it comes to individual prices. The performance of the whole category and whether KPIs have been hit are the metrics which the efficiency of algorithms should be judged by. What else managers need to understand is that humans still call the shots and have a final say in pricing. 

To sum up, artificial intelligence is here to stay and bring value to many sectors, including retail. Its biggest potential lies in its ability to augment human capabilities and make them significantly more productive and precise. As an increasing number of market players is already adopting the technology, you need to speed up, learn to seize every opportunity machine learning and AI give to you and your team, and grow in the highly competitive and quickly changing retail market. It’s a now or never moment: the adoption needs to start as soon as possible as there is a big risk of disappearing from the market if you don’t act quickly. 

-Vladimir Kuchkanov is a pricing solutions architect at Competera. Vladimir is a Data Scientist, top rated domain expert in business analytics, pricing and media management with a successful track record in world-class FMCG companies.