Dynamic Pricing

## Differences from Fixed Pricing Traditional fixed pricing is determined by "cost + margin" and, once set, remains unchanged for an extended period. Dynamic pricing overturns this assumption by raising prices during high-demand periods or seasons and lowering them during slow periods, thereby maximizing revenue. While this approach has decades of history in the airline industry, the barriers to implementation were high. This is because demand forecasting, competitor monitoring, and price simulation all need to run in real time. ## Changes Brought About by AI Advances in machine learning models have significantly reduced the cost of implementing dynamic pricing. An approach in which historical sales data, weather, event information, and competitor prices are fed into a model to predict optimal pricing has become commonplace. In Thailand's hotel industry, the adoption of demand forecasting models that combine OTA (Online Travel Agency) pricing data with in-house reservation status is progressing. Balancing room occupancy rate and Average Daily Rate (ADR) is a key management challenge, and AI-driven price optimization is attracting attention as a means of alleviating this trade-off. ## Pitfalls When Implementing If prices fluctuate too frequently, there is a risk of eroding customer trust. When the price of the same product changes within a matter of hours, a psychological response of "I don't know when to buy" can set in, dampening the desire to purchase. Furthermore, when competitors use the same algorithm, the risk of "algorithmic collusion"—where each party references the other's prices and triggers a race to the bottom—has also been pointed out. In practice, designing systems that combine price floor settings with rule-based constraints becomes critical.