A decade ago, tracking a competitor's price meant a spreadsheet, a browser tab, and someone on the merchandising team refreshing it by hand once a week. Today, machine learning models scan competitor storefronts continuously, adjust a retailer's own prices within minutes of a rival's markdown, and feed that signal into automated repricing engines. This shift has genuinely moved margin for the companies that do it well, but it has also pulled in regulators who are drawing a sharper line between pricing based on market conditions and pricing based on a profile of the individual shopper. Understanding both sides of that line is now a core competency for anyone running commercial strategy.
Retail assortments have grown too large and change too fast for human price checks to keep pace. Grocery and general merchandise chains are now shifting to electronic shelf labels precisely because static, periodically updated pricing cannot respond to tariffs, supply chain disruptions, or a competitor's flash sale within the same business day, according to industry reporting on the retail sector's move toward AI-enabled pricing. Where a manual process might refresh competitor prices weekly, machine learning systems can ingest material costs, competitor prices, customer demand signals, and inventory levels continuously, and translate that into a pricing recommendation in near real time rather than in a batch update days later.
The financial case for AI-driven pricing is not theoretical. McKinsey's pricing practice reports that its Periscope pricing platform, which incorporates syndicated retail data and competitor prices into an automated recommendation engine, has driven sales lifts of 2 to 5 percent and margin uplifts of roughly 1.5 percent for retail clients. On the B2B side, McKinsey's more recent work on agentic AI in pricing describes one company that achieved a total margin uplift exceeding 250 basis points by combining analytical rigor with AI-orchestrated pricing execution. Separately, McKinsey's retail research found that 88 percent of retailers now use AI regularly in their operations, up from 78 percent the prior year, with pricing and promotion optimization consistently cited as one of the highest-value use cases because it touches revenue directly rather than only cutting cost.
Adoption data from inside the retail industry itself confirms pricing is now a mainstream AI use case rather than an experimental one. The National Retail Federation's research on AI in retail found that 40 percent of surveyed retailers are already using AI to dynamically tailor pricing and promotions, placing it among the most common generative and predictive AI applications alongside store analytics and inventory management. That figure matters because it shows real-time competitor price monitoring has moved well past the pilot stage for a meaningful share of the industry, even as broader AI budgets remain modest: NRF's separate survey of retail AI leaders found more than three-quarters of retailers still allocate 5 percent or less of their technology budget to AI overall, suggesting pricing is one of the areas getting disproportionate early investment relative to total AI spend.
Regulators distinguish sharply between two practices that can look identical at checkout. Traditional dynamic pricing adjusts a price based on neutral market conditions such as demand, weather, or inventory levels, and remains broadly lawful. Surveillance pricing adjusts the price shown to a specific individual based on personal data, such as location, browsing history, or an inferred willingness to pay. The Federal Trade Commission's own staff study, based on 6(b) orders issued to intermediary firms including Mastercard, Accenture, and McKinsey & Co., found that these pricing intermediaries worked with at least 250 retail and services clients, and that data points as granular as a person's precise location or browsing history were frequently used to generate individualized prices for the same product. Understanding which side of that line an AI monitoring and repricing system sits on is now a compliance question, not just a commercial one.
The commercial stakes of getting this wrong became visible in December 2025, when Reuters reported that the FTC had opened an investigation into Instacart's AI-based pricing tool, Eversight, after a study involving 437 shoppers across four cities found that prices for identical grocery items at the same stores varied by around 7 percent depending on the shopper. Instacart's stock dropped roughly 7 to 10 percent in after-hours and premarket trading following the report, a reminder that pricing-technology risk now shows up directly on a public company's share price, not just in a compliance memo. Instacart maintained that its price tests were randomized rather than based on individual shopper data, and that in most cases retailers, not the platform, set the prices shoppers ultimately see, illustrating how contested the line between acceptable experimentation and prohibited targeting has become.
The practical response for companies deploying AI price-monitoring tools is to keep the system anchored to market signals rather than individual shopper profiles, and to document that distinction clearly. Several states have already moved to require explicit disclosure when a price shown to a consumer was set algorithmically using personal data, and multiple state attorneys general have opened inquiries into specific retailers' pricing tools. A defensible AI pricing program tracks competitor prices, inventory, and demand signals at the market or segment level, keeps a clear audit trail of what data fed each pricing decision, and treats any use of individual browsing or purchase history as a distinct, higher-scrutiny category rather than folding it quietly into the same system that tracks a rival's storefront price.