Numbeo is the most used cost of living dataset; here is how we read it, what it captures, and the 4 structural gaps the atlas works around.
Numbeo is a Serbian based crowd sourced cost of living database started in 2009. As of May 2026 it carries 12 million data points from 6,300 cities, contributed by 360,000 active reporters. Each city carries 50 plus standard cost categories (1 bedroom rent in city center, 1 bedroom outside city center, milk, bread, restaurant meal, monthly transport pass, internet, and so on). Numbeo is free to query.
Strength 1: coverage. Numbeo profiles cities Mercer and EIU never visit (Asuncion, Lusaka, Mogadishu, Yangon). Strength 2: cadence. Submissions roll in continuously; small cities can get fresh prices monthly. Strength 3: granularity. The 50 plus categories let you build per item comparisons. Strength 4: transparency. Every number shows the underlying contributor count.
Gap 1: small sample bias. A city with 12 contributors versus a city with 600 is not comparable. The Atlas filters out city categories with under 30 contributors. Gap 2: expat skew. Numbeo contributors over represent expats in expensive central districts. Gap 3: aspirational reporting. People over report the rent they "should" pay versus what they actually pay. Gap 4: real estate lag. Rent moves faster than Numbeo refreshes. The Atlas cross checks rent against local platforms (Idealista, PropertyFinder, Daft, Rightmove) for every city profile.