Round 8% of the 31.1 million critiques submitted to Tripadvisor in 2024 had been pretend, in keeping with the corporate’s “Transparency Report 2025.”
That is greater than twice the quantity detected in 2022, in keeping with firm studies.
However that does not imply the variety of pretend critiques on the location doubled, stated Becky Foley, vice chairman and head of belief and security at Tripadvisor.
Whereas submissions to the web site elevated, Tripadvisor’s insurance policies on pretend critiques additionally modified, she stated, citing the corporate’s extra aggressive stance in opposition to “incentivized critiques.” These happen when firms provide clients reductions or freebies in alternate for critiques, or present incentives to staff whose names are talked about in critiques.
“The staff get their mother, finest buddy or cousin to submit critiques, mentioning their names,” she stated. “It finally ends up resulting in companies having critiques that are not really helpful to our neighborhood.”
What’s a pretend evaluation?
Based on Tripadvisor, it’s “any evaluation submitted by somebody who’s knowingly submitting biased or non-firsthand content material, in an effort to control a property’s repute.”
However there is a greater cause detections are on the rise, she stated.
“Our system is at all times getting higher,” she stated, citing Tripadvisor’s three-pronged course of that depends on auto-detection, human evaluation and neighborhood suggestions.
Some 7% of submissions in 2024 had been auto-rejected earlier than being posted, the report stated. Auto-detections additionally flagged one other 5% of submissions for human evaluation.
Nonetheless, Tripadvisor’s belief and security workforce finally moderated greater than 4.2 million critiques, amounting to greater than 13% of all submissions in 2024, in keeping with the report.
One other 244,000 critiques had been disputed by members on the third stage of evaluation, the report confirmed. Of those, some 72% remained on the location, and 28% had been eliminated, it stated.
Easy methods to spot a pretend
Tripadvisor has 4 classes of pretend submissions: boosting, vandalism, member fraud and paid critiques, Foley stated.
There is a false impression that vandalism accounts for many pretend critiques, Foley stated, including that boosting (54%) and member fraud (39%) have lengthy made up the majority of these critiques.
Paid critiques (4.8%) are a smaller however “extra pernicious” class of fakes, she stated.
This consists of “evaluation farms,” she stated, including that these authors are sometimes concerned in different forms of on-line fraud too.
Most paid critiques originate in Asia, Foley stated, regardless that solely 17% of actual submissions got here from the continent final 12 months.
In 2024, greater than one-third of all paid submissions detected by Tripadvisor got here from Indonesia and Vietnam, whereas in 2022 most paid critiques got here from India, studies present.
A ‘cat and mouse’ sport
Rooting out pretend critiques is a continuing “cat and mouse” sport, Foley stated. However Tripadvisor is getting higher at detecting them yearly, she stated.
“We’re the primary to confess that we’re by no means going to achieve absolute perfection,” stated Foley. “We would not catch (a pretend) the primary time, however we’ll catch it will definitely.”
Tripadvisor makes use of its personal expertise, developed over 25 years, to flush out fakes — a system that more and more depends much less on what a evaluation says, and extra on the way it’s posted, she stated. The corporate makes use of synthetic intelligence and behavioral biometrics to search out patterns, which may detect abnormalities like submission spikes and IP tackle masking makes an attempt.
To catch paid reviewers, Tripadvisor investigators pose as pretend evaluation brokers, Foley stated. When a foul actor posts their first paid evaluation, “we’ve got all the knowledge … a whole bunch of knowledge factors related to it,” which the corporate makes use of to construct patterns to establish different critiques submitted by the writer up to now.
Violators aren’t pulled off the location, however their rankings are penalized for a 12 months, Foley stated. Repeat offenders obtain a crimson badge on their itemizing “that claims this property is making an attempt to deceive you,” she stated.
AI critiques: not pretend, however not allowed
Most critiques written by AI aren’t pretend, Foley stated, calling that “one of many myths I like to bust.”
Most are written by common vacationers that use the expertise to make their writing extra polished, she stated.
Nonetheless, the corporate does not enable them — at the least not for now.
In 2024, Tripadvisor eliminated greater than 200,000 critiques that it suspected had been written by AI, in keeping with the corporate’s 2025 transparency report.
“We’ll proceed to watch the tendencies and patterns. However proper now, we simply don’t need vacationers to come back to TripAdvisor for a sea of sameness.”
