Beta — free, no account

Pick where to go on holiday
by describing it.

One phrase in, 158 European destinations ranked — and every score explained line by line. Beacon is the destination finder from Arabel: you write the trip you want, it shows you the arithmetic behind the answer.

Month: October Nights: 5 Culture: weight 9 Off the beaten track: weight 7
158 destinations
35 countries
9 interest axes
0 accounts required
// how it works

From a sentence
to a ranking you can audit.

The hard part isn't finding a destination: it's knowing why one came out above another. Beacon answers that in three steps, and never hides the middle one.

01

Write the phrase

Written rules turn it into criteria — month, nights, budget, interests, vetoes — and show them to you before applying them, each beside the word it came from. They run in your browser, with no network call and nothing to configure.

02

Read the ranking

Destinations are ordered by a weighted average of the axes you asked for. Every card carries the total and each axis's contribution, so when a result surprises you, you can see which axis produced it.

03

Correct what doesn't convince you

The scores are estimates, not measurements. Change them and the ranking becomes yours — kept in a separate layer, exportable, and never written back over the catalogue.

// the ranking

No mystery ordering.

Travel engines rank by popularity or by commission, and don't say which. Here the contribution of every axis is on the card, and the detail view spells out the full arithmetic, row by row, down to the last decimal.

A total that goes up without an axis explaining it would be the black box this tool exists to avoid — so the theme bonus is added outside the average and shown as its own line.

#1
Prague
🇨🇿 Czechia · City
100.0
92.0 + 8 gothic theme

Leading axis Culture 34.0 pts · 5 of 8 axes

Estimated cost €420–610
14 °C air no sea
// what gets weighed

Nine axes.
You control eight.

Nature Culture Sea Food Nightlife Outdoor Family Off the beaten track Value — computed

The ninth is the only one nobody fills in. Value is computed from the costs, so the most expensive destination in the catalogue scores 0 and the cheapest 100. It can't contradict the price shown on the same card, because it is that price — which is also why the way to change it is to correct the cost, not the score.

// drive costs

And, in a section of its own,
what it costs to drive there.

Two points — your location or an address, and any destination — give you distance, driving hours, fuel, tolls and vignettes, divided by the number of people if you travel together.

Florence → Barcelona
return trip, 2 people · 2,207 km · 23 h 36 of driving
Fuel€251
Tolls€157–215
France 1,054 km · Italy 802 km · Spain 138 km
Total€409–466
each€204–233
  • A range, not a price

    The toll is a per-kilometre average by country, not the fare at the booth. A single figure would be wrong the first time somebody checked it — so you get a band, with the rate used printed next to every line.

  • It breaks down country by country

    Fuel, tolls, vignettes, and the tunnels and bridges it recognises — Mont Blanc, Fréjus, Øresund — each on its own line. When a total looks wrong, you can see which piece is wrong.

  • It stays out of the ranking

    Deliberately. Putting a network call inside the scoring would tie a ranking that always works to a free service that may not answer. The day it's down, this page stops and the rest of the tool doesn't notice.

// where it stops

A tool for deciding,
not a shop.

Half of what makes a recommendation trustworthy is knowing what it won't do. These are limits by design, not features waiting to be built.

Books nothing

No availability, no checkout, no payment system — and it stays that way on purpose.

Searches no flights

Flight cost isn't modelled. A budget you set here covers accommodation, food and local transport.

No accounts, no profiling

Nothing to sign up for, nothing to log into, and nobody watching what you searched for.

Sends your data nowhere

Criteria, favourites and corrections stay in your browser's local storage. There's no server to send them to.

// frequently asked

Questions.

How does Beacon understand a sentence written by hand?+

By written rules that run in your browser, with no network call. They cover months and dates, trip length, budget, the nine interest axes with their intensity, vetoes such as “but not Sardinia”, and the destination type. The criteria are shown to you before they are applied, each next to the word it was inferred from. For the phrases the rules don't reach you can plug in a language model — including one running on your own machine — but it only turns the sentence into criteria: it never decides the result.

How accurate are the costs?+

They are estimates, and they are always shown as a range rather than a single figure, because a precise number would communicate a precision that does not exist. Accommodation, food and local transport are covered, per person. Flight cost is not modelled at all.

How many destinations are there?+

158 destinations across 35 European countries: cities, areas and islands. Their climate is measured, from ten-year monthly means published by Open-Meteo covering 2015 to 2024. The scores and costs are estimates, and the interface says which is which.

Is Beacon free? Do I need an account?+

Free, with no account and no sign-up. There is no server: search criteria, favourites and your corrections live in your browser's local storage. The code is public under the MIT licence, at github.com/memphis90/beacon.

Can I correct scores I disagree with?+

Yes, and that is the point. Every axis score and every cost is editable from the Parameters panel, and the change is kept in a separate layer: the catalogue file is never rewritten. You can export your corrections as overrides.json and keep them.

Write where you'd like to go.

FREE · NO ACCOUNT · MIT LICENCE · PUBLIC CODE ON GITHUB