Best Time To Go
Discover the perfect moment to travel.
 Open app
✨ How it works

The science of going at the right time.

Best Time To Go reads 10 years of real weather data, layers it with local knowledge, and tells you the 2 to 3 weeks per year actually worth booking flights for.

The problem with averages

Travel guides say "April in Lisbon = 18°C." Technically true. Totally useless. The first week is rainy. The third is perfect. The fifth might be windy. An average flattens this into one number that hides the entire story.

We do the opposite. We look at every day of every April for the last 10 years and ask one question: which 7-day window has the highest density of "great" days?

The signals we use

🌡️
Median high temperature
Per calendar day, taken across 10 years. Median, not mean, so one heat wave or cold snap does not skew the score.
💧
Rainy day percentage
Share of years that recorded 2.5mm or more of rain on that exact calendar day. The 2.5mm threshold filters brief tropical showers ERA5 over-counts.
☀️
Sunshine hours and wind
Average daily sunshine plus peak wind feed into packing list and outdoor-comfort scoring.
👥
Crowd level heuristic
Peak months come from hand-curated local calendars plus Northern and Southern hemisphere summer rules. Low crowd is part of why we recommend shoulder windows.
🎉
Festivals and local events
For hero destinations we curate the major events that change a month: Nyepi in Bali, Cherry Blossom in Tokyo, Día de los Muertos in Mexico City. Locals tell you these matter. Algorithms miss them.

The algorithm in one paragraph

Score every day green (great), yellow (mixed), or red (bad). Slide a 7-day window across every day of the month. Score the window: green +2, yellow +1, red -2. Keep the top non-overlapping windows above a minimum threshold. The result is the 2 to 3 stretches of the month that consistently deliver the best travel weather, year after year.

Simple. Traveler-tested. No machine learning needed.

Where the data comes from

Weather: Open-Meteo's historical archive, using the ERA5 reanalysis dataset from the European Centre for Medium-Range Weather Forecasts. The same data climate scientists use. Covers every spot on Earth back to 1940 at 9km resolution.

Free version pulls the last 3 years for any city. Premium unlocks 20 years, which is when you can spot real climate trends like "Bali Mays have lost 2 dry days per decade."

What we do not do

Common questions

Is the data accurate for tropical destinations?

Mostly. ERA5 tends to over-count brief afternoon showers in the tropics. We compensate by counting only days with 2.5mm or more of rain as "rainy". This catches days that actually feel wet to a traveler and ignores 90-second mist showers.

How is this different from a regular weather app?

A weather app shows you what will happen tomorrow. We show you what happens every year on this day, so you can decide which week of which month to travel in 2026. Different question, different math, different value.

Do you support cities outside your six hero destinations?

Yes. Any city on Earth with a name Open-Meteo's geocoder recognizes (about 200,000 of them). The 10-year analysis runs the same. What is curated for hero cities is the seasonal narrative, festivals, and packing nuance. We are expanding this layer city by city.

Why do best-week windows sometimes shift between visits?

They do not, unless you change the month. Best windows are computed deterministically from the same 10-year dataset. If you see different windows, it is because the data added a new year or you switched between hero-curated and live data sources.

Try it on a city you know

Plug in a destination you have been to. See if our best-week pick matches your experience.

Open Best Time To Go →