Methodology
How the destination score works
Every province gets a score out of 100 for the trip you asked about, recomputed hourly. A score is a claim, and a claim nobody can check is just an assertion with a number attached — so here is the whole method, including the parts that are judgement calls.
What the score is not
It is not a safety rating. A high score does not mean somewhere is safe, and a low one does not mean it is dangerous. It answers a narrower question: given the weather, the air, the season and what is there, is this a good week to go.
It is not for sale. No partner, advertiser or commission influences it. There is no term in the calculation for whether a province has anything bookable, and there never will be. That rule is written down in our affiliate disclosure.
It is an opinion expressed as a number. The inputs are data; the weights below are our judgement. Reasonable people would weight them differently.
The standing half
Computed from what a province is, and from the time of year. This half changes slowly.
Seasonality
−2 to +8Whether the province is in its best months right now, with partial credit for adjacent months so the score does not jump on the 1st. Being out of season costs only 2 points, because best months describe comfort rather than viability — Bangkok in August is less pleasant, not closed.
Prominence
up to +24Derived from how deeply we have curated the province. It tracks real draw better than population does: population ranks Nakhon Ratchasima and Ubon Ratchathani above Krabi and Phuket, which is not what anyone means by a destination.
Intent fit
−10 to +18How well the province’s actual mix of places matches what you asked for — beaches and islands for a beach trip, temples and museums for culture. A province with nothing matching the intent takes a 10-point penalty rather than coasting in on prominence alone.
The live half
Applied on top, from conditions right now. Only the leading candidates are enriched with live data — fetching weather for all 77 provinces on every request would be slow and pointless when the bottom sixty cannot reach the top of the list.
Rain
up to −18, plus −12 above 90%Chance of rain today. The extra penalty above 90% exists so a genuine washout can actually sink a destination instead of shaving a few points off it.
Air quality
−7 per AQI band above fairLive air quality. Matters most in the northern burning season, when the weather looks perfect and the air is not.
Upcoming event
+5A festival or event starting soon, from curated listings and the official TAT catalogue.
Official attractions
up to +6Places listed in the Tourism Authority of Thailand’s open data, after filtering out promotional records.
Official dining
up to +7Restaurants in the same catalogue, with a bonus where a Michelin listing appears.
The ceilings
Two conditions cap a score no matter how well it scores on everything else. A capped destination always reads Check first and can never read Go now.
An active weather advisory
max 70Where the Thai Meteorological Department has a live warning covering the province’s region or coast. Capped scores are compressed into a band below the ceiling rather than flattened onto it, so it is still possible to tell which of several affected places is least affected.
A standing government travel advisory
max 56Pattani, Yala and Narathiwat, which many foreign governments advise against travelling to. A strictly lower ceiling than a weather advisory gets, because the two are different kinds of fact: heavy rain clears by Thursday, a standing advisory has lasted years. Sharing one ceiling was a real bug — during a nationwide monsoon warning everything capped to the same band and a Deep South province surfaced as the top pick for the weekend.
Songkhla is deliberately excluded. Its southernmost districts are affected; Hat Yai is a normal city that Malaysian visitors travel to every weekend. These provinces keep their guides, their pages and their search results — the cap applies only to the ranking.
When nothing is good
In monsoon season an advisory can cover four regions and both coasts at once, which is effectively the whole country. When every destination we can rank is capped, the ranking says so plainly instead of presenting a top five as though it were a recommendation — the leader is labelled least affected rather than best move, and the list is described as damage limitation. This state can last for months, and pretending otherwise would be the easiest way to become useless.
Known limits
- •Intent fit rests on roughly 227 curated places across 77 provinces — about three each. The ranking is only as good as that dataset, and deepening it is the obvious improvement.
- •Live enrichment covers only the leading candidates, so a province could in principle be held down by stale standing signals despite perfect conditions today.
- •Air quality comes from a model rather than ground stations, which understates burning season in the north.
- •The weights have not been validated against anything. Nobody has checked whether people who followed a high score had a better trip, because there is not yet enough traffic to ask.
See it working