Methodology · PSS 3.0.1
The Pokhara Stay Score
The Pokhara Stay Score (PSS) is a 0-10 comparison score calculated from the evidence available for each property. It is not a guest review rating, an official star classification, or a promise about an individual stay.
How PSS works
We compare six things: guest satisfaction, sleep and noise context, room comfort, views, location, and power/connectivity. Stronger evidence carries more weight than hotel marketing claims.
Missing information does not automatically mean a hotel is bad. Every score carries a separate confidence label. Current records are desk-researched, so room-level Wi-Fi, noise and backup coverage have not been independently tested.
Read the full methodology → · Our data standards
- Score describes the comparison
- 9.0-10 Exceptional · 8.0-8.9 Excellent · 7.0-7.9 Very good · 6.0-6.9 Good · 5.0-5.9 Fair · below 5 Poor.
- Confidence describes the evidence
- High or medium confidence: stronger evidence. Low confidence: provisional. Pending: insufficient evidence to publish a score.
How to interpret current scores
Every property page shows the score status and confidence. “Provisional” means the evidence is limited. “Pending” means there is not enough information to publish a score. Use the score to compare the records shown here, and review the property details before booking.
Guest-rating evidence contributes when a source has at least 25 reviews. Category ratings require at least 30 reviews to contribute. Noise ratings are estimates: they count the bars, pubs and nightclubs mapped in OpenStreetMap within 150 m and 300 m of each hotel's verified location, and can only be raised by that count because the map under-records Pokhara nightlife. No completed onsite measurement rows are currently recorded, so current results do not include measured noise, Wi-Fi speed, backup-power, or standard-room-view audits.
Current record counts: 50 properties · 1 provisional · 0 pending · 49 scored.
Six weighted pillars
A pillar with no usable evidence is omitted from the weighted calculation. Evidence strength and freshness affect how much it contributes.
Guest Consensus
Weight: 30%
Sleep & Quiet
Weight: 20%
Room & Comfort
Weight: 15%
Views & Setting
Weight: 12%
Location & Access
Weight: 12%
Power & Connectivity
Weight: 11%
A pillar with no usable evidence is left out of the weighted average. The remaining pillar scores are averaged using their published weights, then the result is pulled toward 6.0 according to evidence coverage. Even full-strength evidence stops shrinkage only when coverage reaches 65%, while thinner evidence retains at least 30% of its own signal.
Coverage combines each pillar’s weight with the strength of its evidence. Evidence strengths are: measured 1.0, guest reviews 0.9, computed 1.0, third-party listing 0.65, hotel-listed claim 0.5, and editorial inference 0.35. Non-computed evidence stays at full freshness for 12 months, then loses half its weight every 24 months, down to a 0.4 floor.
Guest ratings are normalized from each source’s scale, then blended with a neutral 6.0 prior weighted like 60 reviews. Each source needs at least 25 reviews to contribute. Cleanliness, comfort, location, and Wi-Fi category ratings need at least 30 reviews, whereas staff, facilities, and value ratings are stored but are not currently used by the scorer.
A result is pending when usable pillar weight is below 40% or evidence coverage is below 25%. Otherwise, low-confidence results are provisional. High confidence requires coverage of at least 70% plus a large review sample or measured evidence with solid reviews, while medium confidence requires at least 50% coverage plus solid reviews or measured evidence. Lower evidence levels are labeled low confidence.
Caps apply when guest consensus is below 3.5 with sufficient review strength (overall cap 5.0), measured night noise is at least 55 dBA (cap 6.0), an eligible cleanliness category rating is below 4.5 (cap 6.0), measured cleanliness is 2/5 or lower (cap 5.5), backup lighting is explicitly recorded as unavailable (cap 6.5), or an in-room smoke detector and a second escape route are both absent (cap 6.5). Unknown backup lighting does not trigger a failure cap.
Evidence sources
The scoring engine distinguishes onsite measurements, guest-review snapshots, third-party sources, hotel-provided claims, and editorial estimates. These evidence types do not receive equal strength. Older evidence loses influence over time. Geographic proximity is managed by a separate canonical place layer and is not currently scored.
The location pillar currently uses a recorded guest location rating where available. Legacy lake, airport, and bus-park distances are excluded until property coordinates and canonical place relationships have been independently verified. The location pillar is omitted when that evidence is missing.
Hotel-listed facts are currently assigned one hotel-level evidence tier in the scoring adapter. This means individual claims do not yet receive separate source-specific strength inside the scorer, so the published record and its field-level source notes should be checked alongside the score.
Price does not affect the overall PSS. Value for money and traveler-type fit are shown separately. Guest-rating categories for staff, facilities, and value are currently collected as evidence but are not inputs to PSS 3.0.1.
Percentiles are shown only for scored hotels with medium or high confidence. Pending, provisional, and low-confidence records are excluded both from percentile labels and from the comparison group.
Method changes
2026-10-01 (PSS 3.0.1): A blank backup-lighting audit no longer counts as a confirmed failure. Current published records have no measured audits, so their overall scores are unchanged. The Best Hotels main list now requires medium or high confidence, and provisional records are shown separately.
2026-09-28 (PSS 3.0.0): Removed unverified legacy geographic distances from the location pillar. Scores were regenerated, so earlier engine versions are not directly comparable.
The score engine version is 3.0.1. A change to its rules may make later scores incomparable with earlier versions.
Corrections and commercial links
Use the contact page to report a factual error. Corrections are reviewed before publication. Affiliate commissions do not enter the PSS calculation. Read the affiliate disclosure.