Reviews are a critical part of the online travel economy, with online shoppers poring over feedback to decide which companies to patronize. But false reviews can devastate a business’s reputation and bottom line if left unchecked. In the latest Fraud Decisioning Playbook, James Kay, TripAdvisor’s director of issues management, tells PYMNTS how the travel platform detected and eliminated one million fake reviews with its fraud detection system, leaving only legitimate reviews for customers to peruse.
Reputation is essential for all businesses, especially those focused on bringing in new patrons. Companies that accumulate positive reviews can more easily appeal to fresh consumers, after all.
This principle holds for travel market businesses that rely on positive online feedback to attract customers. The process’s high stakes mean travel-adjacent businesses, such as hotels, restaurants and tour groups, need to feel confident that the reviews available are accurate. Fake reviews can present merchants in a misleading light, prompting legitimate customers to either engage or avoid companies based on false statements.
Travel platform TripAdvisor has been taking fake reviews seriously, employing a data- and human insight-heavy strategy to prevent them from being posted to its platform. James Kay, the firm’s director of issues management, recently spoke with PYMNTS about how TripAdvisor’s system defends the “content integrity” of posted reviews against misinformation and other types of fraud.
“Fraudsters are very inventive,” he said. “They are constantly trying new techniques and it’s absolutely a game of cat and mouse in trying to stay ahead of what the fraudsters are doing. That’s really the aim of our fraud team — to stay one step ahead of the fraudsters — and I think by and large we’ve been pretty successful in our efforts.”
Keeping Reviews Genuine With Data Insights
Reviews are the lifeblood of TripAdvisor’s business model, and trustworthiness and reliability are essential for the global travelers and businesses that use its platform. The company recently published its first review transparency report, which details its reviews’ editorial guidelines and notes that the 66 million reviews posted to TripAdvisor’s website were monitored by advanced fraud detection technology.
The company’s processes have prevented approximately 1 million fake reviews from getting published, Kay said. TripAdvisor uses machine learning (ML) to conduct deep data analysis that gauges the authenticity of submitted reviews, considering factors like the reviewer’s internet protocol (IP) address as well as the device used and its location. TripAdvisor’s fraud team then analyzes flagged reviews to evaluate their trustworthiness.
“Ours is a very data-driven approach,” he explained. “Someone can write a review and very carefully write it in a way so that, to the naked eye, it looks like a genuine review, but the data doesn’t lie. There’s going to be elements in the data that give away the true nature of that review and that’s what we can track.”
Tapping A Data Pool To Stay Ahead Of Fraud
TripAdvisor’s fraud detection system can also tap into the data the company has already collected from reviews, helping the system decide whether certain behaviors should be considered unusual.
“We’re talking about an arsenal of review submissions made over the lifetime of the site,” Kay said. “That’s a very good data pool in which to be able to identify patterns and trends.”
He compared the company’s anti-fake review efforts to techniques the banking industry uses to detect financial fraud: Data collected from every encounter can enhance the system’s ability to push back against fraud and respond quickly to suspicious activities.
“The system has been built to be able to adapt and flag [reviews] when it sees patterns and connections that might be on the rise,” Kay said.
How Content Integrity Benefits End-users
Staying ahead of fraudsters and fake reviews is essential for TripAdvisor to continue serving those who rely on the site. These customers need to be able to make informed travel decisions based on reliable information.
“As far as content integrity goes … ultimately, what we’re delivering to consumers [is] reviews that provide useful and accurate information so they can feel confident and can plan a great trip,” Kay said.
Content integrity is just as crucial for the various businesses that appear on the TripAdvisor site, he added, noting that sometimes such entities might even be behind the fake reviews. TripAdvisor’s evaluation process organizes fraud into three specific categories. The first, biased positive reviews, are the most common, occurring when business owners encourage friends and family to write favorable reviews on their behalf. The second scheme uses biased negative reviews, where competitors post unflattering things to harm others.
The third misleading type involves paid reviews, in which users “sell” favorable evaluations to business owners that will be then posted to popular sites like eBay or Facebook. This type of activity is very rare and only represents 3 percent of all fake reviews, according to TripAdvisor’s research. The scheme can elevate businesses’ rankings based on faulty information, so the company curbs the activity by assigning “red penalty badges” to those suspected of such tactics.
Paid reviews can also result in legal trouble for reviewers. TripAdvisor assisted Italian prosecutors in a case last year against a paid reviewer that resulted in the reviewer receiving a nine-month jail sentence. Working with law enforcement to crack down on fake reviews can send a powerful message that the issue is serious, Kay added.
“A fake review isn’t useful to anyone,” he said. “It’s important for both [consumers and businesses] that we’re maintaining the platform in the way we are.”
Reviews are essential to travelers who are planning where to stay, where to eat and what to tour, as well as to the businesses providing each service. Safeguarding publicly available reviews’ integrity is key to making informed travel-related decisions, letting providers win consumer trust.