Skip to content
Advisory Compass

Methodology

Advisory Compass collects travel advice published by foreign ministries, maps it onto one scale, and shows where governments agree and where they don't. The consensus and its strength are computed by deterministic rules described below, never by an AI model.

1. Collecting the advice

We read each government's official open-data feed where one exists (for example the UK, US, Canada, Germany, the Netherlands, Japan and Italy), and its public web pages otherwise. Where a government states its level only in prose, we read fixed phrasings (such as "advises against all travel to …"). When a page advises against travel but we can't tell how strongly or where, we show that government's level as not determined and leave it out of the consensus. We never assume there is no advice. Our crawler identifies itself, respects robots.txt and crawl delays, and re-reads a page when the government signals a change, or every 12 to 24 hours where it gives no signal. Every distinct version is stored, so changes over time can be shown.

2. One common scale

Not every government uses four levels. Norway, Sweden, Germany, Italy, Belgium, UK, New Zealand, and Singapore only advise against travel, or sometimes issue no level at all. When they publish no such advice, the destination may still carry elevated risk. We therefore map their "no advice" to a range, normal or increased caution, so a coarse scale counts as uncertainty rather than disagreement. Wording that leaves the extent open, such as Germany's "Von Reisen wird abgeraten" (non-essential or all travel), maps to a range in the same way. Japan's and Korea's first level already means "exercise caution", so it maps to increased caution. See every mapping on the Governments page.

3. Whole country and most severe area

Most advice has a baseline for the destination as a whole plus stricter advice for specific regions, such as border areas. We keep both: the whole country view uses each government's baseline (its advice for the country as a whole, or for "the rest of" it when it singles out regions), and the most severe area view uses the strictest level it gives any part of the destination.

4. Consensus and its strength

We work up the scale one step at a time. Is the risk at least increased caution? At least avoid non-essential travel? At least do not travel? At each step we count only the governments whose advice answers that question. The UK, for example, can't tell normal precautions from increased caution, so it sits out that step but counts on the others. The consensus is the highest step a majority supports, and an even split counts as yes, the more cautious answer. Because every step is a yes-or-no question, governments that split between avoid non-essential travel and do not travel still count together against those that advise neither. When no government can separate two levels, we show both.

Its strength reflects the closest call: the deciding step where the majority was narrowest.

Every government counts equally, however long ago it last changed its advice: advice that is still published is that government's current position. The date it was last updated is shown next to it.

Agreement counts the governments whose advice fits the consensus, including coarse scales whose range covers it. Disagreement measures how far apart governments are: if no single level fits all of them, they disagree by one or more levels.

5. Why governments differ

Common reasons, and how we detect them:

Limitations

We may lag behind a government's latest update, and automated parsing can make mistakes. A government that publishes no advisory isn't saying a destination is safe: some only publish advice against travel. Every page links to the original advice, and you should always read your own government's advice before travelling.