How the Safe Trip Score is calculated
Every country gets a single number between 0 and 100, recomputed daily. This article explains how that number is built, without code or jargon. The formal version lives on the methodology page; this is the friendlier read.
1. We start with raw events
Each day, automated jobs pull in events from 15+ public data sources: the World Health Organization, the European Centre for Disease Prevention and Control, the US Centers for Disease Control, ProMED-mail, ACLED for armed conflict, USGS for earthquakes, NOAA for hurricanes, the Smithsonian Global Volcanism Program, GDACS for floods and wildfires, OpenAQ for air quality, OSAC for crime baselines, and the major government travel advisories (UK FCDO, US State Department, Smartraveller, travel.gc.ca, and others).
Each event is tagged with three things: a category (disease, conflict, disaster, crime, civil unrest, infrastructure), a confidence tier (official, confirmed news, unverified), and a severity. The full list and refresh cadence is on the sources page.
2. Events fade over time
Recent events count more than old ones. We apply a decay window per source, an unverified ProMED tip drops out after 14 days; an ACLED-reported armed clash decays over 90 days; a USGS earthquake matters for weeks, not months. This is why the same country’s score moves day to day even when nothing new happens.
3. Each event becomes a sub-score
We group decayed events into six categories and produce a 0 to 100 score for each:
- Disease, outbreaks, vector-borne disease seasonality, vaccination-preventable disease
- Conflict, armed conflict, terrorism, regional military activity
- Disaster, earthquakes, storms, volcanoes, floods, wildfires
- Crime, tourist-relevant crime, violent crime baseline
- Unrest, protests, civil disturbance, election-period instability
- Infrastructure, transport strikes, air quality, power-grid disruption, visa changes
4. The six sub-scores blend into an overall
We weight the six sub-scores into a single number. Conflict carries the highest weight (22%) because its consequences are the most acute; disease and disaster sit at 18% and 16% respectively; crime, unrest, and infrastructure round out the remainder. The exact percentages live in the version manifest on the methodology page.
5. Government advisories nudge the result
We then pull in advisories from the major foreign ministries and average their levels. The overall score is nudged toward that average by 20%. A country whose data looks calm but every government has elevated to caution will end up a few points lower; a country whose data looks rough but no government has raised an advisory will end up a few points higher.
6. An editor can apply a calibrated override
Some signals require human judgement. A wildfire in the central interior does not affect the capital’s airport; an outbreak in one province does not collapse a country score. A Safe Trip editor can apply a time-bound, signed override of up to ±25 points. Every override is published with the editor’s name and an expiry date; we cannot quietly nudge a score and pretend an algorithm did it.
7. The result is auditable
Every score links to the events that produced it. If you ever wonder “why is Country X at 64 today?”, the country page’s Active drivers section has the answer, with sources, confidence tiers, and how each event affected the number.
Why versioning matters
Algorithm changes are released as numbered versions (we’re currently on v2.3). Material changes carry 14 days’ advance notice on the changelog, and corrections are published within 2 hours of identification. Pro and B2B Teams customers can pin a specific version when citing the score in a report, useful for compliance work where reproducibility matters.
What this is not
The Safe Trip Score is not a forecast and not advice. It is a synthesis of observed data across sources we trust. For a trip you’re seriously planning, read the country page in full, click through to the government advisories, and consult a qualified professional if your case is unusual (medical, journalistic, diplomatic, etc.).
Where to go from here
- Methodology, the formal version with weights and pseudocode
- How to read the Safe Trip Score
- FCDO travel advisory levels, explained
- Changelog, every algorithm change, ever