Artificial intelligence and machine learning enable retailers to boost their bottom lines, in part by showing when it’s not necessary to offer a discount. At the same time these systems give shoppers prices they view as fair and non-arbitrary on the products they care most about.

Cheryl Sullivan, chief marketing and strategy officer, Revionics

Cheryl Sullivan, chief marketing and strategy officer, Revionics

There’s an old saw along the lines of “retailers are famous for being first at being second”—in other words, they have a reputation for somewhat being laggards in their pace of technology adoption and exploration into new channels.  But there are a couple of notable areas of exception.

First, the pure-play ecommerce retailers have driven some of the most innovative technological advances in areas such as recommendation engines, dynamic pricing, gamification and loyalty program models. Second, retail has been a surprisingly early adopter of AI and machine learning applied in the real world and driving measurable business impact and ROI.

The failure to leverage proven science capabilities can mean the difference between thriving and failure in today’s aggressive retail landscape.

Some of the applications are highly visible attention-grabbers such as in-store robots scanning shelves and interacting with shoppers or online chatbots enhancing customer service, offering related items to a shopper’s selections or speeding the path to purchase.  Many robots employed behind the scenes in distribution centers or retail manufacturing also leverage self-learning technology. Elsewhere, AI powers fraud detection and autonomous delivery technologies including drones and self-driving delivery robots.

Optimizing prices and promotions

Perhaps one of the most well-established and also one of the most high-impact areas of AI in retail is in price, promotion, and markdown optimization. Second-generation price and promotion technology appeared in the market more than a decade ago. Unlike the earlier first-generation solutions, these new approaches gave retailers transparency into how the science was weighing various parameters, overcoming the non-productized science whose model ran stale and produced a “black box” that had led retailers to be skeptical about adoption or accepting recommendations.


Perhaps more important, the newer generation embraced AI and ML, enabling the algorithms to continuously learn (and unlearn when necessary) and evolve at the pace of market, competitive and shopper behavior change, growing ever more capable and accurate as they chewed through real-world data and matured natively beyond the confines of a controlled R&D environment.

Today AI and ML continue to transform the retail price and promotion landscape, across all channels. Its ability to deliver a win-win—a positive, proven impact on the retailer’s bottom line while also giving shoppers prices they view as fair and non-arbitrary on the products they care most about—is why retailers are so willing to embrace AI in pricing and promotions.

In fact,a recent Revionics-commissioned global study by Forrester Consulting, “Retail Success Requires Personalized, AI-Driven Pricing Strategies,”found that 76% of retailers believe AI-driven pricing would have a positive impact on shoppers. Happily, shoppers agree, with a separate study finding that an overwhelming 78% of shoppers think it is fair to use data science to increase and decrease prices as long as they are presented with prices they’re willing to pay

How it works

So exactly how do AI and ML deliver such an impact in price and promotion optimization? In pricing, these capabilities can help retailers provide targeted, more personalized prices and offers that factor in shopper sensitivity and competitive elasticity, down to the store-item level. For context, machine learning is essentially a toolkit of different techniques and approaches to solving problems. Just as a specialized tool is only effective in the hands of a skilled artisan or technologist, highly experienced data scientists leverage knowledge of the tools in their ML toolkit to apply them successfully in the realm of real-world retail pricing and promotions.


Retail users can in turn “play with the knobs” to set the dials that achieve their pricing strategies—for example, applying science against different strategies on items that are effective as traffic drivers or transaction builders v. those that are successful at driving margins. The all-important transparency I mentioned above provides those weightings to the users along with the optimal price recommendations.  While considering basket affinities, the science factors in margin analysis, competitive price analysis, price elasticity analysis, brand sensitivity analysis, good-better-best relationships, price tiers, price families and more.

If you add the elements of automation, responding in real-time to other retailers’ price changes online, and the ability to provide new price recommendations at the speed at which a retailer wants in each of their channels, you’ve entered the realm of dynamic pricing.  Dynamic pricing is particularly valuable to online retailers, but we’ve seen adoption by retailers through in-store channels as well using Electronic Shelf Labels (ESLs), or just by prioritizing those price changes that will have the biggest business impact.

Promotions: Learning what not to do

On the promotion front, science can immediately review historical promotions and their associated tactics and help retailers stop harmful margin leakage immediately just by understanding what not to do.  This step alone has proven to save retailers over $60M.  Moving forward the science can recommend more optimal promotions that achieves category strategies, consider halo and cannibalize affects and meet financial objectives.

The October 2018 study that showed 52% of shoppers get weekly or monthly promotional offers from retailers for items for which shoppers would have happily have paid full price. Fortunately, science-based promotion performance analysis can pinpoint promotional waste. Putting a stop to ineffective promotions can save retailers millions—instantly.  On a more proactive front, using prescriptive AI-based analytics helps retailers account for consumer buying influences, forecast demand, promotional vehicle impact, cross-item effects and vendor fund influence to recommend promotions with the optimal channel, vehicle and offer to deliver carefully crafted promotions that meet the retailer’s strategic goals.


At the end of the day, retailers who leverage AI- and ML-based price and promotion capabilities have a clear advantage in delighting their customers with meaningful, carefully crafted prices and promotions. By contrast, retailers who fail to utilize these capabilities risk alienating shoppers, squandering scarce resources and undermining their brand.  The failure to leverage proven science capabilities can mean the difference between thriving and failure in today’s aggressive retail landscape.

Revionics provides price-optimization software for retailers.