Hotel reviews are useful, manipulable, and badly read. Studies estimate 15-30% of reviews on the major sites are fabricated. Among the real reviews, most are read for the wrong information. The framework below explains how to extract signal from review pages.
How fake reviews are produced
Three primary sources of fake hotel reviews:
1. Hotel-commissioned positive reviews
Hotels pay third-party services to generate fake positive reviews. The most common source is "review farms" in countries with low labour costs — workers post 30-50 reviews per day across multiple sites.
Signals: short reviews, generic praise, accounts with limited posting history, similar phrasing across multiple reviews.
2. Competitor-commissioned negative reviews
Less common, but real. Competing hotels (or their agencies) post negative reviews of nearby properties.
Signals: focus on a single major issue (rather than nuanced criticism), accounts with reviews of only that hotel and a few competitors, posted within a short window.
3. AI-generated reviews
Increasingly common since 2023. Generated text is plausible but lacks the specificity of real experience.
Signals: no specific names of staff, no specific room numbers or types, no specific dates of incidents, generic descriptions that could apply to any luxury hotel.
How to spot fake reviews
Five signals to watch for:
1. Account history
Click on the reviewer's profile. Real travellers have a history of reviews across multiple properties, restaurants, and locations. Fake accounts typically have:
- Reviews of a single property only
- Reviews of 5-10 properties all in the same area, all posted within a week
- Profile pictures that are stock photos (reverse image search will confirm)
- Names that are generic ("John Smith", "Sarah Lee")
2. Specificity
Real reviews mention specific things: the name of the doorman, the number of the room, the time of an incident, the dish at dinner. Fake reviews are generic.
A real review:
"The breakfast at the Pavilion Restaurant was outstanding — particularly the housemade brioche and the fresh-pressed orange juice. Our waiter (David) remembered our names by the second morning."
A fake review:
"Amazing breakfast! Service was excellent. Would highly recommend!"
3. Length and structure
Real reviews are typically 100-300 words and discuss multiple aspects of the stay (room, service, food, location). Fake reviews are typically very short or oddly long, with focus on one aspect only.
4. Posting cadence
Real reviews appear at irregular intervals. Fake reviews appear in bursts — 5-10 positive reviews in a single day are usually paid placements.
5. Language patterns
Real reviews have natural variation in vocabulary, sentence structure, and tone. Fake reviews (especially AI-generated ones) have unnaturally consistent style. They often use industry jargon ("the property", "the team") rather than natural language ("the hotel", "the staff").
Reading the real reviews
Once you have filtered out fake reviews, three rules for reading the real ones:
1. Sort by most recent
The hotel's current state is what matters. Reviews from 2 years ago may describe a different property after renovations, ownership changes, or management transitions.
2. Read the moderate reviews, not the extremes
5-star reviews are over-positive (the guest had a great moment). 1-star reviews are over-negative (the guest had a single bad experience). 3-star and 4-star reviews are usually the most informative — these are travellers describing the average experience honestly.
3. Look for repeat issues
A single review mentioning slow Wi-Fi is anecdote. Five recent reviews mentioning slow Wi-Fi is signal. The pattern across reviews is more useful than any individual review.
A hotel review tells you what one traveller experienced. Hotel reviews tell you what the hotel actually is. Read the pattern, not the individual review.
Aspect-by-aspect review reading
Different aspects of hotel quality are best assessed from different review types:
Service quality
Read the most recent 3-star and 4-star reviews. Look for: were complaints resolved, were preferences remembered, did the staff anticipate.
The 5-star reviews about service are usually about a single excellent staff member, which is not generalisable. The 3-star and 4-star reviews describe the actual service standard.
Cleanliness
Read the most recent 1-star and 2-star reviews. Cleanliness issues are over-reported in negative reviews and under-reported in positive ones. The negative reviews are usually accurate about specific cleanliness problems.
Food
Read the longest reviews. Food is a complex topic; long reviewers typically engage with it. Short reviews give limited food information.
Look for: specific dish recommendations, comments on breakfast quality, whether dietary restrictions were handled well.
Wi-Fi and tech infrastructure
Read business-traveller reviews. Business travellers test Wi-Fi seriously and report on it. Leisure travellers often do not engage with the tech.
Noise
Read the most recent reviews from longer stays (3+ nights). Single-night reviewers may not have noticed noise issues that matter on longer stays.
Location
Read recent reviews from people who walked the area. The review will mention specific restaurants, walking distances, and the surrounding neighbourhood character.
What to ignore
Three things in reviews that are not useful:
1. The number itself
A 9.4 hotel and a 9.0 hotel are the same hotel. The 0.4 difference is noise. Use the score as a coarse filter (above 8.0 vs. below 8.0); ignore fine differences.
2. The "loved it" reviews
A 5-star review with "amazing!" and no detail is unhelpful. Skip them.
3. Reviews from people whose travel style is different from yours
A 3-star review from a budget traveller complaining about the rates at a luxury hotel is not signal. A 5-star review from a non-business traveller praising the family pool is not signal for a business trip.
Cross-platform reading
The same hotel often has different review profiles on different platforms:
- Booking.com tends to skew slightly positive
- Tripadvisor is more discriminating
- Google reviews skew toward business travellers
- Hotel-specific review sites (often the hotel's own) are heavily skewed positive
Read reviews on at least two platforms. The disagreements are interesting:
- High Booking.com score, low Tripadvisor score: usually a hotel that performs well on short stays
- High Tripadvisor, low Booking: usually a hotel whose long-stay quality exceeds short-stay
- Both high: probably a genuinely good hotel
- Both low: avoid
Read: Understanding Hotel Reviews
The full framework for reading hotel ratings.
Read the pillar guide →What about review aggregators
Sites like TrustYou and ReviewPro aggregate scores across multiple platforms into a single "sentiment" score. These are useful as quick filters but lose nuance.
The aggregated score is a starting point. Read the underlying reviews to understand what the score actually represents.
A note on professional reviews
Professional reviews (from journalists, hotel inspectors, travel agents) have advantages over user reviews:
- Trained observers
- Multiple-night stays at the same property
- Comparison to other hotels in the same category
- Editorial standards for accuracy
Professional reviews are less manipulable than user reviews. The mainstream luxury travel publications (Conde Nast Traveler, Travel + Leisure, the New York Times Travel section, the Financial Times How to Spend It) are reliable.
The trade-off: professional reviews are less timely. A property that has declined since the review may still hold its old positive coverage.
The platform-specific reading framework
Different review platforms have different strengths and biases. A specific framework:
Booking.com
Strengths: Volume of reviews. Most active platform globally. Strong filtering by trip type.
Biases: Reviews are slightly inflated (the reviewer self-selects to review only positive stays). The 0-10 scale produces compression at the high end (most hotels score 8.5+).
How to read: Filter by recent reviews, filter by trip type, ignore the score, read the text.
Tripadvisor
Strengths: Long review history, more detailed text reviews.
Biases: Skews toward leisure travellers. Older reviews dominate the front page (sort by recent).
How to read: Sort by most recent. Read the 3-star and 4-star reviews. Ignore the 5-stars and 1-stars.
Strengths: Integrated with maps, easy to find nearby alternatives.
Biases: Skews toward business travellers. Reviews are typically shorter than other platforms.
How to read: Useful for quick verification but not as primary source.
Hotel-specific review pages
Strengths: None for unbiased evaluation.
Biases: Heavily skewed positive. The hotel curates which reviews appear.
How to read: Read for marketing language, not for evaluation. Move to independent platforms.
The specific signals of fake reviews in 2026
Fake review patterns evolved with the rise of AI-generated content. Specific 2026 signals:
Signal 1: AI-generated specificity
AI-generated reviews now include specific names and details. They sound real. The signal is the absence of negative detail. AI-generated reviews are uniformly positive; real reviews include at least one criticism, however minor.
Signal 2: Emotional language without specifics
Phrases like "absolutely magical", "truly transformative", "an experience we will never forget" without specifics about what made the experience magical or transformative. Real reviews ground emotion in detail.
Signal 3: Posting cadence
Five reviews of the same property in 24 hours, all positive, all from accounts with similar profile pictures and registration dates. This is a coordinated posting campaign.
Signal 4: Generic profile pictures
Reverse image search any reviewer profile picture that looks like a stock photo. If it appears on multiple websites, the reviewer is fake.
Signal 5: Inconsistent travel patterns
A reviewer who has reviewed a $40 motel in Ohio and a $4,000/night Maldives villa in the same month is unlikely to be real.
Reading negative reviews productively
Three rules for negative reviews:
- Look for the specific issue, not the emotional reaction
- Check whether the issue is a property fault or guest fault (the "the room was small" complaint at a hotel with published room sizes is usually guest fault)
- Look for whether the hotel responded — and how the response reads. Hotels that respond defensively are hotels that handle issues poorly. Hotels that respond constructively are hotels that handle issues well.
The hotel's response patterns tell you more about how the hotel will handle your issues than the reviews themselves.
The five rules
If we were forced to compress review reading:
- Filter out fake reviews using account history, specificity, and length signals
- Sort by most recent and read the 3-4 star band, not the extremes
- Look for repeat issues across multiple reviews — patterns matter
- Read aspect by aspect, using the appropriate review type for each
- Cross-reference at least two platforms; the disagreements are signal
For more, read the hotel reviews and ratings pillar and the booking sites comparison.
