Trends Surprising results from analyzing the sentiment of 4,000 online hotel reviews.

You probably monitor your online reviews pretty carefully. You may even apply some advanced tools like sentiment analysis to see how good or bad a guest’s experience really was. But what happens when you’re not looking at just one review, but thousands. What can we learn? Some pretty interesting stuff it turns out.

The Goods
To find our answer we ran reviews for over 4,000 hotels through Bitext‘s sentiment analysis API to find an average sentiment score for each one. We then grouped reviews by their star rating and found the average sentiment score for each group.

As expected, one-star reviews saw the lowest sentiment score with an average of -0.51. Five-star reviews come out on top with an average score of 1.59. The biggest gaps were 0.77 between one- and two-star reviews and 0.56 between three- and four-star reviews.

What This Mean
What surprised us most was how high one-star reviews scored. As we manually audited some of the results we saw the reason: it doesn’t always take a completely terrible stay to get one star. Instead, many of these reviews started by saying something positive before naming the dealbreaker(s) that ruined the trip.

Three-star reviews, that we expected to be right at 0.0 on the scoring scale, were further away from 0.0 on the positive side than one-star reviews were on the negative side. Even more interesting was how flawless and extremely positive five-star reviews all seemed to be. In other words, “sorta bad” is enough for a one-star review, but “sorta good” may only get you three. If you want five, the experience has to be incredible.

Next Week, More
We’ve been publishing new hospitality marketing insights every Tuesday since 2012. Don’t want to miss the next round? Stick your email address below.