Skip to content
Switch Horizon

Methodology

How we score countries

Switch Horizon ranks 116 countries on 7 dimensions with a simple, deterministic multi-criteria method. Here's everything that goes into a score, including what we leave out.

Principles

  • Public, citable data only

    Every number comes from the World Bank's open datasets. You can look up any figure yourself using the indicator code we show.

  • No invented numbers

    If a source has no recent value for a country, we show the gap. We never fill it with an estimate. The score is then based on the remaining indicators, and we flag that.

  • Same answers, same result

    Your ranking is plain arithmetic: weighted averages of published indicators. There's no AI involved, so you can reproduce it yourself.

  • Scores are relative

    A score of 100 means best among the countries we cover, not perfect. A score of 0 means lowest in our set, not zero.

The seven dimensions

Each dimension averages one or more indicators. Indicators with long tails (such as prices, income and homicide rates) are log-scaled first, so a few extreme values don't flatten everyone else.

Affordability

How expensive everyday life is, based on the World Bank price level (GDP per capita at market rates divided by GDP per capita at purchasing-power parity). Lower prices score higher.

  • Price level (US = 100)· NY.GDP.PCAP.CD / NY.GDP.PCAP.PP.CD · lower is better · log-scaled

Safety

Intentional homicides per 100,000 people (World Bank WDI) combined with the Political Stability and Absence of Violence score (World Bank WGI).

  • Homicides per 100,000 people· VC.IHR.PSRC.P5 · lower is better · log-scaled
  • Political stability (0–100)· GOV_WGI_PV.SC · higher is better

Healthcare

Life expectancy at birth, physicians per 1,000 people and health spending per person (World Bank WDI).

  • Life expectancy (years)· SP.DYN.LE00.IN · higher is better
  • Doctors per 1,000 people· SH.MED.PHYS.ZS · higher is better · log-scaled
  • Health spending per person· SH.XPD.CHEX.PC.CD · higher is better · log-scaled

Governance

Rule of Law, Government Effectiveness and Control of Corruption scores from the World Bank Worldwide Governance Indicators.

  • Rule of law (0–100)· GOV_WGI_RL.SC · higher is better
  • Government effectiveness (0–100)· GOV_WGI_GE.SC · higher is better
  • Control of corruption (0–100)· GOV_WGI_CC.SC · higher is better

Civil freedoms

The Voice and Accountability score (World Bank WGI): freedom of expression, association, media and participation in selecting government.

  • Voice and accountability (0–100)· GOV_WGI_VA.SC · higher is better

Digital access

Share of the population using the internet (World Bank WDI). It measures access, not connection speed.

  • People using the internet· IT.NET.USER.ZS · higher is better

Economy

GDP per capita at purchasing-power parity, unemployment rate and how far inflation is from 2% (World Bank WDI).

  • GDP per person, adjusted for prices (PPP)· NY.GDP.PCAP.PP.CD · higher is better · log-scaled
  • Unemployment rate· SL.UEM.TOTL.ZS · lower is better
  • Inflation (consumer prices)· FP.CPI.TOTL.ZG · lower is better · log-scaled

How a score is calculated

  1. Each indicator is rescaled to 0–100 across the 116 countries we cover (min–max). For indicators where lower is better, the scale is flipped.
  2. A dimension score is the average of its rescaled indicators. Inflation is scored by its distance from 2%, so both high inflation and deflation lower the score.
  3. Your answers set an importance for each dimension, which becomes a weight: Not important = 0, A little = 1, Important = 3, Essential = 6. The steps grow on purpose so that one essential priority outweighs several minor ones. Your match score is the weighted average of the dimension scores.
  4. If a country has no data for a dimension, that dimension is skipped for that country and the remaining weights are rescaled. Results marked “partial data” rely on fewer indicators.
match = Σ (weightᵢ × dimension scoreᵢ) ÷ Σ weightᵢ
(only over dimensions where the country has data)

Worked example

Here's how Portugal scores for the example profile used in our sample report. These are the live values from our current dataset.

DimensionScoreWeightScore × weight
Affordability34634 × 6 = 204
Safety78678 × 6 = 468
Healthcare85385 × 3 = 255
Governance64364 × 3 = 192
Civil freedoms84184 × 1 = 84
Digital access83683 × 6 = 498
Economy84184 × 1 = 84
Weighted average (sum of contributions ÷ 26)69

Which countries are included

We start from an editorial list of countries that are realistic relocation destinations. We then exclude any country where the World Bank lacks enough recent data (at least 7 of 8 core indicators). That leaves 116 countries. These 11 places on our list are currently excluded:

  • Cuba · too little recent data
  • French Guiana · not covered by the World Bank
  • Guadeloupe · not covered by the World Bank
  • Liechtenstein · too little recent data
  • Monaco · too little recent data
  • Martinique · not covered by the World Bank
  • French Polynesia · too little recent data
  • Reunion · not covered by the World Bank
  • Taiwan, Province of China · not covered by the World Bank
  • Virgin Islands, British · too little recent data
  • Mayotte · not covered by the World Bank

We also leave out Russian Federation, Belarus, Islamic Republic of Iran, North Korea, and Ukraine because of international sanctions, armed conflict or severe practical barriers to relocating there.

What we don't measure

These matter a lot for a move, but we don't have reliable, openly licensed data for them across every country yet. So they're not part of the score:

  • Visa and residency rules, which depend on your citizenship and change often.
  • Personal income tax and special regimes for newcomers.
  • Climate and weather.
  • How widely English or other languages are spoken.
  • Rent and housing costs. The price level is a national average across all consumption.
  • Differences between cities and regions within a country.
  • Recency. Official statistics usually lag by one to two years, and we show the year for every figure.

Data sources

Our dataset was last rebuilt from these sources on October 6, 2026. It's regenerated by a script that downloads the latest values. No figure is entered by hand.

Try it with your own priorities