How it works

Methodology

What we measure, how, and where the numbers can mislead.

For each city we took the first ~36 listings from vacation-rental search results for the area (2 result pages) on 2026-10-13.
For each listing we read its public availability calendar for the current and next two months. "Calendar fill" = nights unavailable ÷ nights counted. Unavailable nights include booked nights and nights the owner blocked, so it can overstate true bookings.
For the current month only the remaining nights (from the 13th) are counted. Listings whose booking window does not reach a month are excluded for that month.
Price: a separate search with check-in 17 Nov 2026 and check-out 20 Nov 2026 (3 nights, a weekday-start stay), 2 adults, the same dates for every city, run on 13 Oct 2026. We take the per-night rate from each result's price breakdown (before fees, discounts and taxes) and report the median and middle half. A dated search only returns listings free for that stay, so we use it for price only, never for calendar fill. Earlier versions used an undated search, which applies each listing's own default ~5-night dates and surfaced pricier homes (Bucharest: $168 undated vs $70 for these dates), so those figures were withdrawn. All prices USD.
Small sample (36 listings per city, shown as n on every city page), and search results favour popular listings. Treat as indicative.

What "calendar fill" means: the share of nights a listing's public calendar shows as unavailable. Public calendars cannot tell a booked night from a night the owner blocked, so calendar fill is an upper bound on true occupancy, not a booking rate.

Generated with the STR Market Occupancy Estimator.