Retailers have tested the waters when it comes to using artificial intelligence for pricing, but many have yet to ride the wave.
A high proportion of retailers reported they have an affinity for using AI for pricing, but that interest may not represent the depth—or lack thereof, thus far—of the use cases they have implemented.
New data from Coresight Research and Competera shows that 92 percent of decision makers in retail organizations said their company leverages an AI-based pricing solution. The data also shows that 96 percent of leaders that indicated their company uses an AI solution have “moderately” or “highly” benefited from the technology.
Related Stories
Manik Bhatia, head of co-branded research at Coresight, said he was “not at all” surprised to see that such a high proportion of leaders stated that AI had become part of price strategy in retail.
But because pricing strategy encompasses such a wide array of considerations—from season, to markdowns, to consumer preferences, to competitors and more—the data doesn’t address how deeply companies have really gone on implementing AI into their procedures.
Alex Galkin, Competera’s co-founder and CEO, said the comprehensive pricing platform has not seen fully automated AI adoption at most retail organizations.
Galkin said it seems unlikely, based on the time that his organization spent trying to convince retailers to use machine learning for pricing in 2023, that retailers are far along in the process of integrating AI systems into their pricing plans. Though many report using AI, he said between 10 and 20 percent of retailers—at the most—have sophisticated, real-time automation in place for pricing.
“[Retailers] are probably using machine learning to generate and optimize their elasticity curves,” he said. “But they’re not using AI that can automatically determine which factors impact your sales for a particular store on this day, [during] this particular time of the week, time of the day, weather conditions, store location. That’s a different level of using AI for pricing.”
He said more advanced models require sound data foundations that many companies don’t have.
That could be why the data showed that, despite nearly all of respondents stating that they used AI for pricing, over half of respondents’ organizations still misprice more than 10 percent of their products.
Bhatia said many retailers aren’t getting the strongest results from their AI trials because it’s still early days.
“The pricing maturity curve shows that retailers are just above halfway on the maturity curve,” he told Sourcing Journal. “They are using AI, but they have not been able to leverage it to the fullest potential. Improving company-wide integration of business functions and using real-time pricing can help bridge this gap.”
The maturity curve Bhatia referenced states that retailers have achieved a “moderately automated” pricing strategy that, at large, does not address real-time pricing problems. To level up, the maturity model suggests, retailers would need to commit to fully automated pricing with real-time repricing that relies on internal and external factors.
Galkin said he expects that will take an additional two to three years. Coresight’s data shows 97 percent of retailers that have already implemented some form of AI into their pricing strategy plan to increase investment in the technology in that area.
Even still, other data shows that retailers aren’t nearly so consumed with adopting AI for pricing. Accenture data shows that 73 percent of leaders said AI would be critical to future success around pricing models.
Coresight’s data shows that some of the primary challenges retailers face when implementing AI-based pricing solutions are lack of integration with normal business functions and lack of trust on AI models.
Galkin said the latter is something most AI providers today face, since many leaders outside of technology organizations don’t fully understand the training behind models and feel some allegiance to humans’ ability to complete tasks AI can now help streamline.
“[Executives] know AI is happening—you [get on] the train, or you pass. They’re open to do A/B testing to see how it works, but still, the majority say, ‘Oh this is not safe,’ or, ‘This is black box, so we will not do it,’” he explained. “There is a psychological barrier on the executive level.”
