The 360-Degree view of AI in Retail Pricing: Industry Analyst, Scientist and Retail Executives

The 360-Degree view of AI in Retail Pricing: Industry Analyst, Scientist and Retail Executives

The 360-Degree view of AI in Retail Pricing: Industry Analyst, Scientist and Retail Executives
August 9, 2018 Alison Raffalovich - Senior Corporate Communications Director

The topic of AI in retail pricing is hotly discussed these days, so it is nice to get a clear-eyed view by consolidating input from a visionary industry analyst, a data science guru and actual retailers who have benefited from the technology in action. Between sitting in on the recent webinar on “Cutting through the AI Hype” and attending the EMEA Retail Executive Summit in London, I got to hear from all those esteemed viewpoints.

R “Ray” Wang, Founder and Principal Analyst at Constellation Research, was the featured industry analyst on the webinar. Ray tells us that AI in retail has been around for some time and is here to stay. Leading retailers have adopted this science to innovate and dominate. The key to successfully applying AI in retail, he said, is access to a range of data in real time leveraging compute power, along with math talent (either in-house or third-party) and – equally important – industry expertise in your retail segment. The math has been around for a long time but having enough compute power to crunch the data is what’s enabled breakthroughs. Ray also talked about the importance of AI-driven smart services for pricing models, focused on value for shoppers, whether monetary or non-monetary. Ray walked us through his design rules for AI, which emphasized transparency in the science and the importance of having human oversight in place in addition to the automated capabilities – which we at Revionics definitely agree with.

Ray also cited a recent Constellation Research study that shows 70% of respondents are already using AI at some level, and 78% are increasing their investment. He notes that retailers can speed time to market with AI by adopting technology from an established vendor: “Digital Darwinism isn’t kind to those who wait.”

Revionics Chief Science Officer Jeff Moore followed with his viewpoints. Jeff said that he concurs with Ray Wang on the importance of human-led philosophy for AI in retail, where you need to strike a balance of risk aversion and beneficial outcomes. Jeff described how Revionics scientists apply an array of ML and AI tools to address different types of retail challenges, such as Gaussian Process Learning in pricing situations where there are relatively few data points (such as introducing a new item in your assortment).

Because all AI/ML predictions have some level of uncertainty, Jeff reiterated the importance of building in transparency and human-led design to factor in for that uncertainty. As he noted, AI-powered solutions are sensible, usable and cautious in the face of unknown or conflicting data.

Joachim Paulsen, Head of Pricing at Reitan Convenience, explained in a panel at the summit how Reitan has moved from focusing on competition in its zone approach to focusing on customer behaviors, leveraging Revionics price optimization science to make that shift. Reitan stores are owned by franchisees, and franchisees were supportive of the initiative because it was intended to increase profitability. Nonetheless Joachim noted that some franchisees were surprised to see that price recommendations included some reductions as well as price increases. To help drive adoption, Joachim did a store tour to explain the machine learning science. Today the proof of value is in the business results – an increase in market share across all the banners.

Joachim described another tangible benefit of science-based pricing. On January 1 this year, the Norwegian government recently taxes on sugar – and buying behavior is very elastic on certain products with lots of sugar. Using Revionics they were able to increase cost on background products while staying the course with prices where the sugar-based items were very elastic. As of the end of the first quarter, Reitan had increased market share on Confectionary, which got the highest tax increases.

When Reitan first implemented Revionics, they implemented a lot of business rules. Over time as they seek to unlock more value from machine learning, they have eliminated many business constraints in favor of more ML-based approaches. Herman explained that science in pricing is where you see the ultimate benefit of good sourcing, merchandising, etc. It enables Lenta to offer targeted pricing rather than taking a blunt one-size-fits-all approach.

Joachim sees agility in pricing key to responding to market factors like changes in weather, bank holidays, etc. Using Revionics, Reitan has implemented more localized pricing to be very targeted in deciding where to change prices. They use the science to follow customer behaviors and make more decisions driven by the customer’s preferences and behavior rather than vendor costs, and the confidence and accuracy of price recommendations continue to increase as the machine learning algorithms evolve.

The other featured retail panelist in London was Herman Tinga, Chief Commercial Officer at Lenta. At Lenta, which spans more than 80 cities across Russia, the teams are very regionalized for sourcing, category management and pricing, as well as lifecycles. Different formats also have different price strategies. Lenta conducts weekly competitive checks on more than 1000 items. They also do very granular customer segmentation and can make offers down to the personalized level. As Lenta continues to mature in its usage of Revionics’ science-based pricing recommendations, they may explore a more centralized approach to pricing.

Herman discussed Lenta’s disciplined, data-driven culture. The company focused initially on data-driven customer insights and then turned to category management, which in turn was fed by the customer data. This led logically to price optimization.

Herman’s team focuses on the balance between providing fair prices to customers but also avoiding the waste of being mispriced or of promotions that are ineffective. Science can help drive good price perception while being cognizant of competitors’ high/low promotion patterns. The Revionics system lets Lenta offer more relevant prices while maintaining healthy margins.

All told, the picture is clear: AI in retail pricing is here, it’s used by leading retailers, and it delivers powerful, measurable business impact.