You Are Viewing

A Blog Post

Analysing Customer Feedback for Restaurants, Hotels, and Travel

travel and tourism global

In any services industry, and specifically in the restaurants, hotels, and travel industries, Web 2.0 fostered a shift into transparency with pricing information and user generated reviews easily accessible to anyone that searches. On top of that, social media channels allowed users to share pictures and opinions and to converse with others about  their experiences in almost real-time. The customer feedback that was once limited to a small group of acquaintances or colleagues is now open to the millions of online viewers. This shift into more transparent interactions, while beneficial to customers, is putting pressure on the business and hence customer experience managers to up their offering or fail.

The need for automation

In this SAS study, Kelly McGuire, explained how negative ratings swayed hotel choice, even  more so than price (negative reviews take the hotel out of the game at which point price is not relevant), whereas the content and the language of the reviews had little impact. While customers do only look at aggregate ratings, service providers do not have this luxury. The article states clearly that to stay in the game, service providers have to improve their ratings. Though It does not mention that to improve these ratings, managers have to know what the issues are and which to prioritise. To make these decisions, they have to analyse the comments people post.

Manual review of a couple of hundred comments is possible; but doesn’t scale when customer experience managers receive  2000+ customer reviews a months,  call-center logs, and user generated reviews (Tripadvisor, Yelp, Twitter, Facebook, blogs, forums), in addition to the constant barrage of inquiries on the different web channels.

Value of sentiment/opinion analysis

The restaurants, hotels, and travel industries are seeing several disruptive waves from price comparison websites, food ordering aggregators, and the AirBnBs that are sidestepping the traditional players in the market. These companies succeed because they rely on processing large amounts of data to make almost real-time decisions about and for their customers. While traditional businesses are waiting for reports from partner DMC (which are at best a couple of months old and everyone gets them) or their survey results; competitors are working on knowing their customers. Automated text analysis bridges this information gap without requiring huge investments in “Big Data” solutions. Knowing the customer opinion helps specifically in three areas:

  • Revenue management and investment decisions:

With all the influences previously mentioned,  revenue and investment management in a dynamic market is one of the main challenges businesses face. To be competent enough, managers have to evaluate enormous numbers of forecasts and inputs. Simplifying these assumptions by listening to the target consumers, globally, can help them produce accurate results in lesser time and make accurate investments whether in improvements or innovation.

  • Guest experience and retention:

As mentioned in the in the SaS study, a reported negative guest experience not only means the business lost that guest’s custom but also  any potential guest reading that feedback.  Detailed sentiment analysis of all customer comments – not a sample – allows businesses to identify and prioritise the positive and negative elements of their service and their competitors’ at a global scale; while being able to hone down on specific regions too. With the always up-to-date information, managers can decide what to offer, what to change, and who to offer it to. Improvements that lead to better guest feedback, better retention, higher acquisition, and the coveted high ratings on review websites.

  • Targeted Social Advertising (or Social Selling):

Social media channels offer an attractive means of reaching specific groups of people anywhere in the world. With this big opportunity, the challenge becomes what to tell them. People on social media channels are not looking to be advertised to; so any message, to be effective, has to talk to them individually. Targeted advertising does not mean sending a generic message to a specific group of people but rather sending a specific message to a specific group of people. Audience analysis through automated text analysis gives businesses the knowledge at the local and group level of the audience’s interests and aspiration, whether that’s the video game player in Tokyo, the new mother in Dusseldorf, or the  teenager in Bogota.

generic sentiment anlysis

Aggregate sentiment analysis showing positive, negative, neutral ratios

How it’s done

Outside specific specialties, automated text analysis is seldom used or even thought possible. In the last 5 years, social media analytics, a form of automated text analysis, has become mainstream. Social media analytics tools provide descriptive statistics about the audience activities on the main social media channels: Twitter, Facebook, Instagram, etc. This information answers questions about how many times the brand was mentioned vs. the competitors; how many reviews the brand received in the last month and what’s the aggregate opinion( 60% positive …); or, what was the response to the brand’s campaigns. All this information is required to help assess the business’s tactical moves and its environment.

metrics for social media channel

Channel metrics for activities and engagement as compared to the industry.


This kind of analysis is flashy, and useful, but not sufficient to answer strategic questions like what are the priority areas I need to address to improve the customer experience; or, how do the preferences of my customers from South America differ from those of my customers from Asia. The answers to these questions are extracted from detailed textual analysis of the customers’ feedback on digital channels.

Until recently, manual coding of text was (and still is) used in some domains to get a very nuanced segmentation of the audience opinion – but to do it at scale, in a business context, is almost useless due to the time/resources required. New commercial tools, such as Crimson Hexagon, focus specifically on addressing this need. They build on academic research in the Social Sciences by using human trainable systems that can analyse documents in any language based on a set of training documents. Relying on a human trainer marries the benefits of manual coding with those of computers to process large amounts of data in reasonable time.

detailed sentiment/opinion analysis

Detailed sentiment/opinion analysis from a customer experience project at 5W Consulting.

Big Insights without the Big Headache

The true value to the restaurants, hotels, and travel industries is in the ability to analyse any and all sources of textual content. With expert help, the business can load the 2000+ customer reviews/month, call-center logs, user generated reviews, social media feeds and then process all the feedback to identify the themes, priorities, and detailed opinion and preferences.

It’s true that to maximise the value from such an initiative you require support. But equally,  you no longer have to content with general evaluations of your customer feedback through vague ratings or generic sentiment graphs – simply, because your competitors aren’t. Your customers are giving their feedback for free; now, you can listen to them – all 1000s of them – from across the globe.

Subscribe to receive similar articles and webinars.

Name: email: