Are Online Reviews the Next Generation of Guest Satisfaction?
By Michelle Wohl VP of Marketing & Client Services, Revinate | April 07, 2013
The popularity of online reviews and the importance of social media are changing the way that hotels are approaching how and where to gather customer preference data. Historically hotels have relied on extensive post-stay surveys and mystery shoppers to ascertain service levels, customer satisfaction, and areas for improvement. However, online reviews have yielded an interesting by-product, a wealth of data for hotels who eagerly want to please customers.
In Spring 2012 the Wyndham Hotel group switched its focus from traditional guest satisfaction surveys to mining online customer review data for satisfaction insights. With the bold introduction of their WynReview platform earlier this year, Wyndham Hotels & Resorts retired their traditional guest surveys and began directing all guests to provide feedback exclusively on TripAdvisor. This disruptive move was backed by extensive research, and Wyndham now focuses on guest feedback in public reviews as the central source of guest satisfaction indices. Sophisticated data mining tools are used to extract specific preference information about a number of operational areas such as room service, housekeeping, front desk, and parking. By combing hundreds or even thousands of reviews for sentiment, the customer preference data contained in reviews can overcome some of the standard bias in surveys and offer a fuller picture of ways to boost satisfaction.
GSS vs. Sentiment Analysis
Traditional guest satisfaction surveys have been the main source of customer feedback for many years. While email has made them easier and cheaper to administer and track, surveys have a number of fundamental issues that can limit their effectiveness. Sampling bias can occur when time constraints make leisure customers more likely to respond than business customers. This skews your survey results by providing information about only what is important to one customer segment. Closed-ended questions such as "please rate your satisfaction with your room from level 1 to 10" will not yield rich data about what a customer liked or disliked about their room. Surveys aim to collect information in as uniform a manner as possible by asking the same questions in the same way so that the answers are most influenced by the respondents' experiences and not by the wording of the question. In order to keep response rates high only a limited number of questions can be asked of each customer. Questions are formulated by hotels to focus on areas that they feel need to be measured, rather than what is most important to customers.
Good survey methodology can limit this weakness, but customers have limited time and attention for surveys. According to Survey Gizmo, the average response rate for an external survey like a customer survey is between 10 and 15 percent which means that it can take a long time for surveys to yield reliable data. Medallia, a leader in formulating and administrating traditional surveys, lists five steps to get the most out of surveys; capture data, share results, recover from feedback, discover root causes, and improve. However, hotels like Wyndham are finding that focusing on online customer reviews can streamline the process, which leads to quicker results.
Sentiment analysis technology has given hotel managers a variety of ways to track customer preferences. A great advantage is that you can delve into the data of a sentiment analysis tool at any time, rather than having to craft a question on a survey and then wait for a response. For example, a traditional survey question about guest rooms may provide a level of satisfaction on a 1 to 10 scale, which is helpful. But an analysis of customer sentiment about guest rooms becomes a great way to learn highly specific preference information. While sentiment analysis technology will never achieve complete accuracy (two humans can often differ in their analysis of feedback), a system tuned and optimized for hospitality feedback produces a high degree of confidence over sizable data sets. For example, a sophisticated sentiment analysis tool can show the following: