Plane tickets usually cost more close to departure: cheap seat inventory sells first, and advance-purchase rules at 21, 14, and 7 days lock out discounts even when seats remain. STR operators do the opposite — inside 21 days many begin dropping rates programmatically until the night expires. This practice should stop.
Airlines vs STRs
The markets differ substantially. A Phoenix–Denver nonstop has maybe five airline choices; Denver has 1,000+ short-term rentals. Airlines also file prices publicly in the Global Distribution System, so competitors observe and match instantly — effectively coordinated pricing. STR operators instead compete ruthlessly, frantically cutting rates to steal last-minute demand.
Why airlines don’t discount last-minute
Two strategies prevent it. Segmentation distinguishes business from leisure travelers, who have different price elasticity — last-minute pricing exploits the inelasticity of non-advance bookers. And forecasting and optimization models let airlines predict the demand still to come, giving them confidence to hold rate. That sophistication wasn’t available in STR until recently.
Why STRs discount last-minute
Discounting fills otherwise-vacant nights through an occupancy-focused strategy that prizes bookings over rate. It’s also fast and cheap: the alternatives — better photography, higher ratings, property upgrades — take time and money, while a price cut executes instantly, making it perpetually tempting.
The downward spiral
Programmatic discounting deteriorates the whole market. When big operators auto-cut unsold inventory two weeks out, average market rates fall, which drags base-price models down, which forces week-before discounters to cut further; scrapers detect the reductions and push pricing lower still, in a destructive spiral.
The solution
Last-minute discounting is a symptom of poor forecasting — in STR specifically, of no forecasting. An optimization model, which requires forecasting, makes last-minute discounting unnecessary and will only reduce total revenue if layered on top. Choose a real forecast-and-optimization model and let pricing science set the rate instead.