AI for dynamic pricing in retail and hospitality

Dynamic pricing is the practice of adjusting prices in real time based on changing market conditions — demand, competition, inventory, and customer context. Traditional dynamic pricing used rule-based systems (fixed discount schedules, manual markdown decisions). AI improves dynamic pricing by using machine learning to model the relationship between price, demand, and outcomes across thousands of variables simultaneously; learning from historical outcomes to improve future decisions; incorporating real-time signals (competitor prices, weather, local events) that humans cannot process at scale; and optimising for specific business objectives (revenue maximisation, inventory clearance, margin optimisation) with mathematical rigour rather than human intuition.

Price elasticity measures how much customer demand changes in response to a price change — typically expressed as the percentage change in quantity demanded for a 1% change in price. Inelastic products (essential goods, unique items) see little demand change when prices rise; elastic products (commodities, easily substituted items) see large demand drops with small price increases. Price elasticity is the foundational input to dynamic pricing optimisation — without knowing elasticity, a pricing model cannot calculate whether a price increase will improve or reduce total revenue. ML models estimate elasticity at SKU, category, and customer segment level from historical price and demand data, accounting for promotions, seasonality, and competitive context.

Hotels use AI revenue management systems (RMS) to optimise room rates across all customer segments and booking channels simultaneously. The RMS ingests data on historical demand patterns, current booking pace versus forecast, competitor rate positioning across OTAs and direct channels, local event calendars, group bookings, and macroeconomic indicators, then calculates optimal rate recommendations for each room type, segment, and length-of-stay combination. Modern systems like IDeaS G3 and Duetto go beyond rate setting to include group business evaluation (accepting or declining group bookings based on displacement cost), function space pricing, and channel mix optimisation to reduce OTA commission costs by shifting demand to direct channels.

Markdown optimisation uses AI models to generate optimal price reduction schedules for end-of-season or end-of-life inventory, maximising total revenue recovery while achieving target sell-through rates. The model predicts demand at each possible price point across the remaining selling period, accounting for remaining time, current inventory levels, historical clearance patterns, and channel capacity, then generates a markdown schedule that maximises the revenue outcome under the constraint of clearing target inventory by a deadline. AI markdown optimisation consistently outperforms manual buyer intuition and rule-based schedules, typically delivering 5–12% improvement in margin recovery on clearance merchandise.

Key ethical risks include: price discrimination — charging different customers different prices for the same product based on inferred willingness to pay, which can disadvantage lower-income customers; surge pricing on essential goods during supply disruptions or emergencies, which regulators and the public consider exploitative; algorithmic collusion — competing algorithms independently converging on higher prices without explicit coordination, which antitrust authorities are increasingly scrutinising; and personalised pricing that penalises loyal customers relative to price-sensitive first-time buyers, undermining long-term customer relationships. Mitigation requires price floors and ceilings, transparency in pricing policies, human oversight of algorithmic decisions, and regular fairness audits of pricing outcomes across customer demographics.

Industries with the highest benefit from AI dynamic pricing share common characteristics: perishable inventory (airlines, hotels, live events, fresh grocery) where unsold capacity has zero residual value; high price variability tolerance (travel, luxury retail, B2B software) where customers expect prices to vary; real-time demand signals (ecommerce, ride-hailing, online advertising) where AI can act on signals before they become stale; and commodity-like competition (fuel retail, consumer electronics) where small margin improvements compound at scale. The weakest fit is industries with regulatory price controls (utilities, healthcare), brand-sensitive categories where price instability damages perception (premium luxury), or low-margin categories where the margin improvement from optimisation is smaller than the implementation cost.

Measure AI pricing ROI through a combination of revenue and margin metrics: revenue per available unit (RevPAU for hotels, revenue per transaction for retail) compared to a control group or pre-implementation baseline; gross margin percentage on repriced categories versus control categories; sell-through rate improvement on markdown items; and competitive price positioning index (percentage of time priced within the target band versus competitors). Use holdout testing — apply AI pricing to a randomly selected subset of products, stores, or rooms and compare outcomes against a matched control group with existing pricing — to attribute performance improvement to the AI pricing system rather than external demand factors. Typical payback periods are 6–18 months.

Open pricing is a revenue management strategy where rates for each customer segment, room type, and booking channel are set independently based on demand conditions — rather than applying fixed differentials from a single Best Available Rate (BAR). Traditional BAR-based pricing applies uniform discounts (e.g., loyalty members always get 10% off BAR, corporate contracts at BAR minus 15%) regardless of demand conditions. Open pricing allows, for example, loyalty rates to be set above the OTA rate during peak demand when loyalty member segments show strong demand, or corporate rates to be released below BAR during low-demand periods to drive occupancy. Duetto pioneered the open pricing approach, which is now adopted by most leading hotel revenue management systems.