Can your product engage users? Tripadvisor case.
Every company is found with a great idea — in the eye of the founders. Yet, is that the case for your customers? For a start-up to succeed, leadership’s vision must align with customer demand and this synergy needs to turn into a product. The problem is, customer demand is not singleton. It is faceted by geography, age, income and so on. Therefore, it is crucial to sense the demand of each facet of users — for the survival of your precious crystallizing company.
In this post, we will give a walkthrough on what to look in order to sense customer behavior, using TripAdvisor’s data. We have collected data on every single hotel and each hotel’s reviews as a proxy to customer behavior. Note that for our analysis, we have utilized the information of 1.7 million hotels and some million users.
Where is this product used?
Map above is colored by the average (log) number of reviews per hotel in that country. To give an example, let’s say we go to France and pick a hotel. Then we expect that hotel to have nearly 400 reviews — why: colorbar indicates the log number of reviews. France is ~2.6, and 10^2.6 is roughly 400.
Map shows that reviewing behavior is more common in the Western world along with a sample of the Mediterranean, and Gulf countries. We will talk about this in more detail. Now let’s focus on the previous example.
We said if we go to France and pick a random hotel, then we expect the hotel to have 400 reviews on average. Here, “on average” is the keyword. Because, if we actually go to France and pick a hotel, then the chances are slim that we bump onto one with exactly 400 reviews. It could be a little higher or lower.
Well I lied. Usually, it is either hell of a lot lower — like one or two reviews, or crazily higher like 2000 reviews. Why? Because social systems are full of inequalities and probably your customer behavior is not an exception. It is like income distribution. Say you are walking in the streets of NYC. Most likely, you will come across students, white collar workers, boom a multi-millionaire and if you are lucky, you can come across Jeff Bezos.
When it comes to social systems, average statistics is mostly insufficient.
Inequality in Reviewing Behavior
In the above plot, we see the inequality of the distribution of reviews per hotel. Let me be clear, these plots are very tricky to read. So let me explain it with an example.
Say we have 100 hotels and 100 reviews. Then, one weird hotel gets 73 reviews. This means the remaining 99 hotels have 27 reviews in total. Going forward, 9 out of these 99 hotels have 18 reviews in total. So remaining 90 hotels have 9 reviews in total. On and on, the final 50 hotels have only two reviews in total.
To simplify, a few of the hotels bathe in reviews where a large sum of hotels receive merely one review — if they are lucky. Sounds familiar?
Now that we grasped how unequal the reviewing behavior is let’s investigate if language use has anything to do with this effect in the following paragraph.
Reviewing Behavior and the User Language
Alright, I admit that this plot is even harder to read. Let’s exemplify on Hebrew case. If we bump into 10 Hebrew speaking people, then we expect 8 of them to review to highly reviewed hotels, and 2 of them to mid-reviewed hotels. It is not very likely to come across a Hebrew comment in a hotel that receives 1–2 reviews.
Note that the plot has three sections. First section is the top languages where most of the reviews in that language can be found in top reviewed hotels. Second section is for the Middle 40%, and final section is for the Bottom 50% — a.k.a hotels with a couple of reviews.
Final section presents an opportunity for growth. Because we see that reviews in Serbian, Slovak, Czech, and Russing (SSCR)languages are mostly found in hotels with a few reviews. Now let’s ideate on this.
If these hotels are in countries where these languages are spoken, then we can:
- Focus on making campaigns that could boost user reviewing behavior for users of language.
- Incentivize Hebrew, Arabic, Korean, Chinese speaking users to visit hotels in Eastern Europe.
- Incentivize Eastern European users to visit -and review-budget hotels that are highly reviewed.
In this post, I shared how you can utilize your data to boost customer engagement. If you have any questions or comments, feel free to ping me via LinkedIn, Twitter, or email: egemensert [AT] gmail [DOT] com.