English

How You Should Perform a Sentiment Analysis on Your Reviews

By Alex Gaggioli, August 8, 2016

Introduction:

As a seasoned property owner, you understand the significance of online reviews. They help travelers discover your property and can ultimately be the deciding factor. Reviews aren’t only useful to travelers, they can also offer you a lot of value. There’s a lot of information lurking in reviews, which may require an additional analysis. Beyond simply reading reviews, it is helpful to take a part each review to identify problem areas, context, and bias.

For that reason, we suggest property owners conduct a sentiment analysis to uncover hidden information. A sentiment analysis is a process where you attribute a positive, neutral, or negative rating to each review. This analysis will help you identify what your guests love and what may need some more attention.  

Why You Should Perform a Sentiment Analysis

How-to-perforam-a-sentiment-analysis---graphs

Everyone has their own opinion concerning online reviews. Some people think they’re unreliable and biased, while others enjoy the constant feedback stream. We know that reviews are often untrustworthy and unfair to property owners, but there’s a lot of information behind the simplistic ranking system that defines most user-generated platforms (i.e. Yelp, TripAdvisor. etc.).

Star and number ratings are pretty worthless unless you look at what influenced a customer’s rating. No platform uses a consistent rating scale and one person’s five is another person’s three. For example, if you only look at a guest’s two-star rating, you’d categorize their rating as negative. But in reality, the guest’s review may have been predominantly positive, but one issue dragged the overall score down.

A sentiment analysis will help take some of the mystery out of reviews. The goal is to see what you’re doing well and identify areas where you can improve. Although guest reviews will continue to have bias, it’s easier to spot when you’re analyzing a review line by line. We’ll walk through how to conduct a simple sentiment analysis and what you should do with the info, next.

How to Run a Sentiment Analysis

First and foremost, creating and running a sentiment analysis manually is a timely task. But, the benefit is getting a much more in-depth look at your reviews that can help you get a real sense of what people think. Here we’ve created a very basic sentiment analysis. It will give you a sense of what people feel about different aspects of your property.

We’ve created a sample sentiment analysis which we will discuss in more depth later. But, if you want to follow along as we walk through the steps, you can access it here.

1. Take a sample of your reviews

We suggest you take a sample of 50 to 100 reviews in order to get an accurate idea of your guests’ sentiments. It’d be great to do an analysis on all your reviews, but that’s a timely, tedious task that you most likely don’t have the time to complete. You can source your reviews from as many different sites as you want. The number rating will become irrelevant once you break down the reviews into smaller pieces, but may be helpful to compare sentiments later on.

2. Export your reviews to a spreadsheet

Enter your reviews into a spreadsheet and include a column that identifies the source for reference. I would also include the number or star rating for reference. In the future, if you want to try and pull insights based on what influenced ratings the most, you can use the same data. We’ll show you in our example later on how to best set up your spreadsheet.

Note: a sentiment analysis works best when there’s a large number of reviews, usually in the hundreds or thousands. If you only have a handful of reviews, it will be difficult to pull accurate insights from the data.

3. Highlight positive, neutral, and negative tones

Go through each review and highlight each phrase if it’s either positive, neutral, or negative. Highlight positive reviews green, neutral yellow, and negative red. The colors will help you identify how people talk about different areas of your property. You’ll soon start to notice patterns and gain insight as the colors start to take shape.

Sentiment Analysis Review Highlights

4. Attribute a score to each review

Using the highlighted phrases, you can then numerically quantify your reviews. For each positive phrase, give the review one point, neutral receives a zero, and a negative review receives a negative point.

Once you’ve attributed a score to each phrase, add it up and determine the general sentiment of the overall review. I suggest creating another column that says “review sentiment” and highlighting it the appropriate color. If your score is positive, it’s green. If it’s zero, it’s yellow, and if it’s negative, it’s red.

5. Create a list of everything mentioned

Go through each review and create a list of everything mentioned in the highlighted pieces. For example, the phrase “the staff was amazing” would be highlighted in green and you would write staff in your “mentioned” list. While you create this list, record the same score you gave in step number four, but this time to the word. So, if you write down staff, record one for the positive review. When you’re finished you’ll be able to see what, generally, people feel about a certain aspect of your hotel–i.e. your staff. 

6. Determine if people feel positive, neutral, or negative about each aspect of your hotel

After you’ve created your “mentioned” list, determine how the general public feels about each aspect. Then, order them according to score. So if your staff was mentioned positively six times, neutral twice, and negative once the math would like Score = 6(1) + 2(0) + -1(1) resulting in a score of 5.

The higher the score, the better people felt about that aspect of your property and vice versa. This analysis isn’t perfect due to the fact that many people won’t mention everything that was good or bad about their stay. However, it does give you better insight than just reading the reviews or taking note of the number rating.

Example of a Sentiment Analysis

How-to-perforam-a-sentiment-analysis---graphs

For our example, we chose a random property off of TripAdvisor and analyzed five of their reviews. We followed the instructions above and first went through each review and found phrases and sentences that were either positive, neutral, or negative. Then, using those same highlighted phrases, I made a list of everything mentioned from decor, to the pool, to the room’s carpet. I gave them each a score depending on how they were mentioned in the review.

After I added up the scores, I was then able to sort the list by positive, neutrals, and negatives. I know now, based on the five sample reviews, that people enjoy the hotel’s decor, rooms, and staff attitudes, but staff knowledge, noise, wifi and pool parties have a serious damper on people’s experience. Here’s what the sample looked like:

How to perform a sentiment analysis - mentioned list

You can view our entire spreadsheet example here to get an idea of how we analyzed the reviews. Obviously, you’ll want to analyze far more than five reviews in order to get a broader view of how people felt about your property.

Shortcomings of This Analysis

This analysis will give you a basic view of your reviews. Depending on how you categorize a review, as positive or neutral, for example, will be determined by how you interpret the reviews. Such as, there were several phrases in the five reviews I used that could have swayed either neutral or positive, but I had to use context clues and my own intuition to make a decision.

However, even with those shortcomings, diving into the details of each review and understanding them line for line will help you. For example, it was quite obvious from only five reviews that WIFI was a serious issue. Documentation of a both positives and negatives instills a sense of urgency to fix and address issues.

Alternatives to This Sentiment Analysis

There are several companies out there who offer more structured and streamlined sentiment analysis solutions. A good sentiment analysis software system will pull in your reviews from the most popular sites and then automatically sort them based on algorithms. Most of them will even give you specific data that will help you make operational improvements and make a real difference for your property. Two such companies that have such products are Revinate and TrustYou.

If you’re interested in running a more in-depth analysis of your reviews, we highly suggest you take a look.

Conclusion

User-generated reviews are an inevitable part of any business in this day and age. But, reviews offer so much more than a headache and biased opinions. If you take a deeper look at what your guests write about you, you’ll gain invaluable insight. As always, a proactive approach to your reviews is always beneficial in the long run. If you keep a close on your reviews, you’ll be able to spot trends faster and make more meaningful changes to your property.

You might also be interested in...

7 Tactics For Successful Hotel Reputation Management

If you own a hotel or any other type of property which accommodates guests, there’s an excellent chance that those […]

PillowTalk: Cloudbeds Product Updates September 2019

Grab a pillow! Let’s talk… As Q3 has come and gone, it’s time to highlight some of the new and […]

Meet Cloudbeds Market Manager: James Spicer

G’day! Kia Ora! That’s how you may hear people greet each other in Australia and New Zealand. I’ll refrain from […]

Worried about SCA Compliance? With Cloudbeds, you’re covered

The new Strong Customer Authentication (SCA) regulation will affect most of the Travel industry in many ways, both directly and […]

Better Together

Cloudbeds loves helping Hoteliers and Hosts grow
their business. Let us show you how.

Book a Demo