Reviews Opinion Mining

Opinion Mining: Digging Deep Into App Store Reviews

A Treasure Trove of Data

Amidst the all-caps rants and troll infested app store reviews is a mine full of market research data. A lot of app publishers and marketers actually know about the potential of app reviews. It is not only important in the context of the app the reviews are for, but also for future app development endeavors. But in order to harness the potential of app store reviews, a process called opinion mining or sentiment analysis is necessary.


The terms opinion mining and sentiment analysis are often used in different contexts. But they both used to describe a process of systematically extracting subjective information from a body of text. Nowadays, there’s a niche for opinion mining in social media networks. Twitter, for example, uses opinion mining to determine people’s reactions on a trending topic. These reactions can be classified into three polarity aspect: positive, negative, and neutral.


In the context of an app store review, opinion mining seems the straightforward analysis of users’ satisfaction. If a majority of the reviews is positive, then great, right? If it’s negative, what else can we do about it? But most of the time, these users actually want to tell you what they want, not only with your app but also in apps with the same functionality in general.


Why Bother with Opinion Mining?

Aside from the aggregation of important feedback, opinion mining can also benefit marketers in different ways:

  1. Opinion mining is helpful in the early stages of app development, especially as you are building the app’s concept. It can give you insights on the market or niche of your app.
  2. Digging deeper, you can find information on specific features or functionality that don’t work for your target audience. That is even before you get into the beta-testing phase!
  3. You can evaluate the cause of both your success and failure. You can actually use app store reviews as some sort of benchmark that can give a picture of how a successful app should look like in the perspective of users. For example, app store reviews and ratings are important considerations in downloading an app. For every negative review, the app’s chances of download decrease. Also, a game app with a lot of downloads doesn’t necessarily have satisfied users. This can result in poor retention and therefore, low revenue in the long run.
  4. You can generate previously overlooked keywords using app store reviews. You can find targeted keywords well-hidden within positive reviews. There’s also the added bonus of these keywords having less competition.


Tools for Opinion Mining

One basic strategy in opinion mining is to group app store reviews according to sentiments. These sentiments are basically keywords that fall under the category “unambiguous affect words”. Words like “happy”, “sad”, and “bored” fall under this category. This approach to opinion mining or sentiment analysis is called a “knowledge-based technique”.


More advanced knowledge-based techniques are not easy to perform and evaluate, though. You can trawl for these keywords manually with the use of filters. This technique can take some time and effort so it’s not practical to use in analyzing real-time sentiment.


There is also a statistical technique that uses machine learning, specifically deep parsing of texts. Automated tools offer more accurate results and more versatile processes. You can create your own tool which can cost a lot of time and money. There’s also App Annie where you can export a spreadsheet of reviews from a specific date range, app version, and country. There are also opinion mining tools for a fee. Most of these tools are geared towards social marketing, though.


You can, of course, use a hybrid technique that combines both the knowledge-based and statistical techniques.

How to Respond to Bad Reviews

Bad reviews, just what I want – said no one ever. This statement could’ve even produced a bitter aftertaste for some, especially us app developers. App stores were rumored to be infested with one star ratings and all caps “reviews”. This is bad news since good reviews and high ratings heavily influence app installs. About 59% check app ratings and reviews before downloading an app. The difference in conversion between a one-star rating and a five-star rating is a staggering 770%! A majority of users won’t download apps that have a three-star rating or below.


So what will you do when bad reviews keep on coming, dragging your app’s star rating down?

  1. Don’t be too defensive. It’s natural to get defensive whenever we receive critical feedback. But this isn’t about you anymore, no matter how personal app users can get. Set aside any strong emotions and encourage the app user to look at the problem objectively. Instead of defending your app, ask for proof that the problem really indeed comes from the app and not from other factors on the side of the user.
  2. Ask them what really is the problem. Some reviews can be quite vague. Ask users why they that your app is ‘the worst ever’ or why they are discouraging other users from downloading it. You can learn a lot from user input and you can save some time from diagnosing app performance problems.
  3. Acknowledge any shortcomings and give apologies for the inconvenience. Even the best apps have “buggy” moments. Don’t deny any mistakes on your (and your app’s) part. Even if you can’t see any mistake, don’t tell your app users that they are downright wrong. There could be some unforeseen reasons why your app malfunctions – just acknowledge the possibility of that mistake and assure them that you are trying to get the root of it.
  4. Don’t just promise solutions, make it happen. Apologies and assurances go a long way in appeasing app users but if you can’t offer any solution, your app will be uninstalled. If the problem is something that can’t be easily patched up, offer users short-term solutions or in-app rewards that would lessen the inconvenience brought on by any technical problem.
  5. Don’t take the bait of trolls. Of course, you can’t answer all of your app’s reviews, but there’s also a good reason why you shouldn’t answer some of them. There are trolls that just want to get a rise out of you. Avoid them, especially if they look like they want to draw attention to themselves. If you feel the need to answer any allegations, make it brief and polite. Do not go down to their level. Explain everything in the first reply and if there’s a need for another reply, make it clear that it would be the end of the conversation.
  6. Don’t let false information spread. It’s difficult to tell if users that spread false information on reviews are trolls or just well-meaning but ill-informed individuals. Whichever of the two they may be, you should immediately correct them. If possible, address all your users in a way that the one who left the review won’t feel like they are being singled out.
  7. Report spammy reviews. Fake reviews are a reality in the app stores then, until now to some extent. Competitors may pay not only for positive reviews of their apps but also for bad reviews on your app. Fake ratings usually don’t come with a review and fake reviews are often quite short and generic. A few of reviews like this are legitimate, especially if your app is still gaining traction in the app stores. But if bad reviews flood your app without reason, you can report these reviews. In Google Play Store, you can flag a review as spam or unhelpful. You can report reviews in the Apple App Store through iTunes Connect. Click the “Contact Us” link under the “Support” tab. On the contact form, select “App Store Questions,” then “Customer Reviews,” and then “Report a Fraud Concern”.