Our data promise
Useful estimates,
without false certainty.
City data changes. We show where a number came from, when it was reviewed and how much confidence it deserves.
How to read a number
“Official” does not mean universally complete, and a popular marketplace does not represent an entire rental market. We keep source authority, freshness and sample bias as separate ideas.
Cost estimates
Every complete guide has a cohort of at least five dated public asking-price samples across at least three public root domains. To keep the comparison coherent, every unit is a self-contained furnished studio or one-bedroom. We store the original amount, currency, monthly or 30-day period, structured minimum and maximum stay, utility status and availability status. For each 30-, 60- or 90-day request, the Planner keeps only samples whose published terms support that exact duration and shows the compatible sample and domain count.
Compatible endpoints are converted with a dated official ECB, Banxico, Banco de la República or Bank of Thailand rate; USD observations remain USD. If a listing does not confirm utilities, we add the separately reviewed city utility allowance and keep the range at low confidence. The lowest and highest adjusted compatible endpoints form the envelope. This is not a live quote, average, median, availability promise or citywide rent index, and a duration with fewer than five compatible samples is labeled accordingly. Food and daily life, transport, mobile service and optional coworking remain separately labeled directional ranges. Flights, deposits, visa fees and insurance sit outside every total.
Freshness and review status
Housing, mobile and coworking evidence is flagged for review after 90 days. Transport, safety, healthcare and community evidence is flagged after 180 days. Climate follows the version of the documented monthly dataset. Each planning value carries a status, confidence, check date, source and method. Housing FX snapshots are stored with their official source date and conversion direction. A stale value stays visible with its status; it is never silently presented as current.
Visa and stay limits
We show a general stay limit and link to a government source. We do not calculate previous Schengen stays, remote-work permission, tax residence, criminal history, passport validity exceptions or individual immigration decisions.
Stay-fit signals
With remote work off, the matcher weights budget 35%, seasonal climate 25%, safety 15%, healthcare 15% and community 10%. With remote work on, it weights budget 30%, seasonal climate 15%, safety 10%, healthcare 10%, community 5%, connectivity 15% and team-hour fit 15%. A selected priority receives an extra ten percentage points and the remaining active dimensions are normalized back to 100%. Missing non-critical evidence lowers confidence and reweights the remaining dimensions; missing cost data excludes a city from ranking. Safety, healthcare and community use four visible editorial grades anchored at 60 (extra planning), 70 (workable), 80 (strong) and 90 (very strong). Country guidance and documented local option coverage support the rationale; the anchor only makes the grade rankable and is not an official citywide index.
Connection conditions
The remote-work layer grades documented mobile and coworking option coverage. It does not publish a citywide Mbps estimate or predict the Wi-Fi in a specific apartment. Travelers should ask for a recent speed test and confirm a backup workspace before relying on a connection for work.
Planner coverage
Only the 8 planner-ready complete guides can enter Stay Matcher results. Their cost, seasonal and fit records must pass structural validation. Ranking requires the strict planner-ready state: at least five structurally valid self-contained studio or one-bedroom asking-price samples, at least three root domains, official FX coverage, structured stay terms and a valid duration- specific normalized interval for 30, 60 and 90 days. Housing can therefore be a dated planning envelope while the overall result remains directional when active food, safety, healthcare, community or connectivity evidence has low confidence. The other 142 city records remain useful for discovery and comparison, but their regional proxies are not used to recommend a stay.
Search indexing and coverage
Browseable does not automatically mean editorially complete. Catalog-only cities remain available for discovery but are excluded from search indexing until they meet the modeled-guide standard. Service records are indexed only when they contain substantive decision text or an enriched property record with confirmed contact, location and property details.
Corrections
Every city record carries a review date. If a linked source has changed, use the contact and corrections page with the source URL and the field that needs correction.