Almost every hotel scores a 4.7. We work out what it actually is.

Review platforms sit on narrow scales — most properties land between 4.2 and 4.9, so the number stops separating anything. We re-normalise each platform against its own distribution, weigh in a score of the photography and a reading of review text, and publish the working for every property.

DATES {{ stayLabel }}
GUESTS 2
TRY:
{{ st.value }}
{{ st.label }}
HOW THE CORRECTION WORKS worked example
{{ e.step }}
{{ e.title }}
{{ e.body }}
{{ e.delta }}
PUBLISHED SCORE
{{ exampleReal }}
platform average
{{ exampleClaimed }}
{{ rg.label }} {{ rg.value }}
⌕

{{ listHeading }}

{{ correctionNote }}
LOADING MORE PROPERTIES…
End of results — {{ resultCount }} properties match these filters.
Nothing clears that bar

{{ noResultsNote }}
{{ mapScope }}
© OPENSTREETMAP CONTRIBUTORS · OPENFREEMAP
CORPUS STATISTICS · {{ corpusSize }} PROPERTIES · {{ cityCount }} CITIES

What the correction looks like at scale

Every figure below is computed from the same corpus that powers the listings. Nothing is weighted by popularity, price or commercial relationship.

{{ k.value }}
{{ k.label }}
SCORE DISTRIBUTION published corrected
{{ b.label }}

{{ histogramNote }}

MEAN CORRECTION BY CITY
{{ c.city }} {{ c.value }}
PRICE AGAINST CORRECTED SCORE one dot per property
5.0 1.0
$70$300$600+
CORPUS MEAN BY DIMENSION
{{ d.value }}
{{ d.label }}
{{ b.title }}

{{ b.blurb }}

Shortlist

{{ savedSubhead }}

{{ row.label }}
{{ h.area }}
{{ c.value }}
READING OF THE SHORTLIST

{{ savedReading }}
Nothing saved yet

Save properties from a listing and they line up here side by side, dimension for dimension.
{{ lightbox.counter }}
{{ hotel.photoCredit }}

{{ hotel.name }}

{{ hotel.meta }}
{{ hotel.claimed }} {{ hotel.real }}
{{ hotel.band }} · {{ hotel.gapLabel }}
{{ hotel.evLabel }}
02.55

{{ hotel.verdict }}

{{ d.label }} {{ d.value }}
{{ d.note }}
WHAT GUEST REVIEWS SAY

{{ m.quote }}

{{ m.src }} {{ m.effect }}
VALUE AGAINST THE {{ hotel.cityUpper }} MARKET
{{ hotel.priceLabel }} vs {{ hotel.medianLabel }} median for a {{ hotel.real }} in this band
median

{{ hotel.valueSentence }}
{{ f.label }} {{ f.value }}

{{ hotel.description }}
{{ hotel.photoCount }} PUBLISHED PHOTOGRAPHS · {{ hotel.photoShown }} SCORED BELOW PHOTO SCORE {{ hotel.vision }}
{{ p.label }} {{ p.score }}

{{ hotel.visionNote }}
HOLDS UP
{{ p.value }} {{ p.label }}{{ p.note }}

Nothing we hold scores above 3.6 for this property.
DOES NOT
{{ p.value }} {{ p.label }}{{ p.note }}

Nothing we hold scores below 3.2 here. The correction comes from rescaling the platforms, not from one weak input.
THE TRADE-OFF IN ONE LINE

{{ hotel.tradeoff }}
{{ hotel.amenityCount }} AMENITIES LISTED · CHECKED ACROSS TWO SOURCES
{{ a.name }}
{{ a.status }}

An amenity is marked confirmed when both the hotel catalogue and OpenStreetMap list it.
01Platform scoresweight 60%

Each platform is rescaled against its own observed distribution: the floor is the score below which fewer than 5% of its listings fall. A 4.6 where the floor is 4.2 carries far less information than a 4.6 where the floor is 3.4.

PLATFORMRAWDISTRIBUTIONFLOORSCALED
{{ p.name }} {{ p.raw }} {{ p.floor }} {{ p.scaled }}
Mean {{ hotel.scaledMean }}
02Room condition from photography

{{ hotel.visionNote }}

{{ hotel.vision }}{{ hotel.photoCount }} photos · weight 25%
03Review text

{{ hotel.sentimentNote }}

{{ hotel.sentiment }}AI reading · weight 15%
04Every input we used
{{ s.kind }}
{{ s.title }}

{{ s.excerpt }}

{{ s.effect }}
FINAL CALCULATION
{{ hotel.formula }}
TRUE SCORE
{{ hotel.real }}
PLATFORM AVERAGE
{{ hotel.claimed }}
{{ hotel.priceLabel }} / night
{{ stayLabel }}
{{ r.platform }} {{ r.priceLabel }} {{ r.tag }}
SCORED HIGHER IN {{ hotel.cityUpper }}
Nothing in this city scores higher at a comparable rate.