Methodology · research method 1.0 · updated 18 November 2026

How the index works

2,099 hotels in 23 destinations, 1,987 measured walks and 1,728 official websites. Here is exactly where it comes from, how we check it, and what the order of each list does and does not mean.

What this index is

An independent index of hotels, guesthouses, resorts and aparthotels in ski resorts, beach towns, spa towns and cities. For each destination we list the places that qualify, map them, measure how far they are on foot from the main lift, beach or centre, and give their official contact details. Every fact links to its source and the date we checked it.

This is an early version: the lists are in a provisional order and guest scores are not in yet.

How hotels get onto a list

We start from open data (OpenStreetMap, Wikidata), official hotel registers and tourism-board accommodation lists, then read each place’s own website for facts. Duplicates are merged; unclear cases are checked by hand. This time we looked at 6,180 candidates and found 2,718 that qualify.

A place qualifies if it is a hotel (including motels and family hotels), a resort run as one property, a guesthouse, pension or B&B letting rooms by the night, or an aparthotel with a reception or staffed front desk. We leave out flats, rooms, villas and chalets let by individuals, apartment blocks without a reception, hostels, campsites, huts without serviced rooms, and places that are closed or not yet open.

A destination needs at least 60 qualifying places before it can be ranked. With 150 or more, the published ranking will be a “Top 100”; with 60 to 149, a “Top N”, where N is the number of qualifying places rounded down to a multiple of 5. We never pad a list. Until the full score exists we do not call any list “top” or “best”: each destination page shows up to 100 places in the provisional order. 45 smaller destinations show every place we found and are marked as small.

The current order is provisional

Provisional order (prominence & fit proxy). Not a quality ranking. No guest scores are used.

Version pfp-0.1 scores seven public signals from 0 to 100 and weights them:

  • Official stars (20): 1 star scores 20, 5 stars 100.
  • Distance to the main anchor (20): walking route to the main lift, beach or centre; 150 m or less scores 100, 2 km or more scores 0.
  • Independent mentions (15): Wikidata 50, Wikivoyage 30, tourism board or register 20.
  • Range of facilities (15): share of up to 14 facilities (spa, pools, restaurant, ski room…) confirmed present, out of those we could check.
  • Map detail (10): how many of 12 key details are filled in on OpenStreetMap.
  • Official website (10): working 100, broken 40, none 0.
  • Size (10): room count on a log scale, from 10 rooms or fewer (0) to 300 or more (100).

If no walking route can be calculated, we use the straight-line distance plus 30% and flag it. A missing value takes the destination’s median, so missing never counts as bad.

It shows how established and how well placed a property is, not comfort, cleanliness, service, food or value. It favours large, starred, well-mapped hotels near the anchor, while small guesthouses often do very well with guests. It also decides where we look first: the leading 10 to 20 places in each destination get an in-depth check on their own websites.

The full score that is coming

The final ranking will use about 97 measures in ten categories. Starting weights by destination type (ski / beach / city / spa):

  • A Guest score, 6 measures (overall, recent, trend, volume): 38 / 38 / 40 / 35
  • B Guest views, 23 (staff, breakfast, beds, noise): 12 / 12 / 12 / 15
  • C Location, 12 (walk and climb to the anchor, shops, transport): 20 / 14 / 24 / 8
  • D Facilities, 14 (spa, pools, kids’ club, parking): 8 / 10 / 4 / 22
  • E Rooms, 8 (size, family rooms, balconies, renovation): 4 / 5 / 6 / 5
  • F Price and value, 7: 10 / 12 / 12 / 10
  • G Cleanliness and upkeep, 4: 2 / 3 / 2 / 3
  • I Ski or beach, 12 (ski-in/ski-out within 100 m, first line, Blue Flag): 6 / 6 / – / 2
  • H Families, 6, and J Accessibility, 5: weight 0, used for family and accessible lists

Data-reliability measures never add points, and no measure except the overall guest score can be worth more than 12% of the total. A hotel needs at least 30 guest reviews and data for at least 60% of the weighted measures, or it shows as “not ranked: missing data”. Weight changes get a new method version, listed below.

Guest scores will come from public review scores and counts. We will match listings by name, location and address, use only confident matches, put everything on one 10-point scale and never count the same reviews twice. Scores based on few reviews are pulled towards the destination’s typical score. We never publish or keep the reviews themselves. Positions will come with a likely range, such as “#6 (4–9)”; the provisional order has no ranges, because it is not a measurement.

What gets its own page

A page has to earn its place, for readers and for search engines. A destination page opens only when at least 30 hotels have official or sourced data, at least 40% of the listed hotels are at verification level A or B, at least 10 hotels have been checked in depth, hotel claims have been checked, the areas are described and the airports and seasons are confirmed. Today 23 destinations pass; the others are listed on the home page with what is missing.

A hotel gets its own page only when it has an in-depth profile, a confirmed website and phone or e-mail, a full address, at least 15 sourced facts across at least three topics, at least one claim we could check and at least two answers built from evidence. Today 183 hotels pass. Every other hotel is shown in full on its destination page; hotels at level C (open data only) are listed without contacts until they are checked.

Verification levels

  • Official site checked Checked on the official site: in the in-depth check set; official website checked; map position and address agree within 50 m; full address; website plus email or phone confirmed on the official site; at least 80% of profile fields hold a value or an explicit “unknown”. (160 hotels)
  • Sourced Sourced: qualifies; position from OpenStreetMap or Wikidata; address with street, postcode and town; website present; property type backed by a source. (1,441 hotels)
  • Open data only Open data only: qualifies or probably qualifies; name, type and position only. (498 hotels)
  • Held: conflicting sources, unclear eligibility or a suspected closure; kept off the lists and the map until resolved. (11 properties)

On every fact, a green tick means confirmed on an official source or measured by us; a hollow ring means it comes from OpenStreetMap only or is the hotel’s own claim.

Data sources and licences

  • Maps, places and addresses: © OpenStreetMap contributors, ODbL 1.0.
  • Routes: OSRM (FOSSGIS routing.openstreetmap.de) on OpenStreetMap data; some address checks use Nominatim.
  • Wikidata (CC0 1.0); Wikivoyage contributors (CC BY-SA 4.0), listing names only.
  • Hotel websites, registers, tourism boards, operators, airports and councils: facts only, linked, quoting 25 words or fewer.
  • Weather data by Open-Meteo.com (CC BY 4.0): modelled 1991–2020 monthly averages; a few places use MET Norway / Yr (CC BY 4.0).
  • European Central Bank euro reference rates, with their date.
  • Interactive map tiles: OpenFreeMap, © OpenMapTiles, data © OpenStreetMap contributors. Overview map: Natural Earth (public domain).

How often data is refreshed

“In season” starts 8 weeks before a destination’s main season.

  • New openings and removals: every 3 months, plus a check 6 weeks before the season.
  • Closures: monthly; every 2 weeks for the leading 20 in season.
  • Contacts and facilities: every 3 months (the leading 20 monthly).
  • In-depth profiles: monthly; hotel offers every 2 weeks in season.
  • Lift-pass prices: monthly from September to December, then every 2 weeks until the new table appears.
  • Distances and climate averages: yearly. The order, and guest scores once added: monthly.

Every page shows when its data was checked.

What we never do

  • No hotel can pay to be listed or to move up, and commission or online bookability is never part of the score. Partner links (Booking.com via Stay22) may earn us a commission; a hotel gets one only when we hold its exact Booking.com page.
  • We never publish review texts, quotes from reviews or reviewer details.
  • No personal data: only business mailboxes such as info@ or reservations@ and reception phones; no staff names, no photos of people.
  • No facts, descriptions or photos taken from booking sites or map apps.
  • We obey robots.txt, make at most one request per second per site (one every three seconds for hotel websites) and never get round a CAPTCHA.
  • We never pad a list, guess a missing number or call monthly data “live”.

Corrections

Coming soon: a “Your hotel? Correct a fact” form, open to anyone with a link to evidence. We will re-check the official source ourselves. Corrections that change a position, and our own mistakes, will be logged publicly. Facts supplied by a hotel will count only with evidence, be marked “according to the hotel” and move a score by no more than 3 points in total. For hotels

Method versions

  • pfp-0.1 (17 November 2026): provisional prominence & fit order; research method 1.0; schema 1.0.