Review Analysis: How to analyze customer & product reviews

Product reviews are one of the most comprehensive and useful sources of insider info available, anywhere. These often have gold nuggets that can guide development, troubleshoot new initiatives, and improve customer experience. They provide a helpful benchmark to compare your offerings to the competition, showing both your strong points and areas in which you need to close the gap. Think free customer research, there before you’ve even thought to ask.

But customer reviews, helpful as they are, are plagued with one big problem. Human experience described in qualitative terms is always going to be messy, and this is no exception. There’s no structure, no data you can pull out with a simple command +F search. Sometimes, trying to make sense of customer experiences, shared in thousands of unstructured reviews can feel like tidying up the shells on a seashore.

Often, operations management teams end up looking only at the star ratings. Those, at least, are easy to manage quantitative data. Go a bit further, and you might read through a small sample of positive and negative reviews. You know you should go deeper, but who has the time-- or extra staff-- to even begin such a time consuming manual review exercise?

What you need to do is a review analysis.

What is review analysis?

Review analysis is the process of transforming unstructured review data to structured data that can be used to guide decision-making.

A few of the primary uses include:

• Product feature ideas: scanning product reviews for sentiment on desired features
• Roadmap prioritization: determining what the dev team should focus on first
• Bug tracking: scanning new reviews in real-time
• Customer care ratings: determining what customer service departments are delivering good service

For instance, Amazon review analysis might tell a tablet manufacturer they need to invest in better packaging and more complete  documentation for their Amazon product. Alternatively, it could suggest that customers won’t be happy till the startup time is less than 30 seconds.

Analysis of online travel reviews could be used to provide market analysis for service providers, or build a recommendation system for future travelers.

Analysis of  app store reviews could tell developers they need to work on better GPS connectivity, or let them know that their map design is much appreciated (and any changes should be rolled out slowly!)

Why is review analysis important?

Review analysis is important because it helps you understand what customers think of your app, site, or experience. At its heart it is a customer centric activity. It enables you to make decisions on what to improve based on what your customers value. A company that doesn’t analyze their reviews might end up basing development decisions on alternate criteria, such as:

• What is easiest to do first
• What the CEO or founder would personally like to do
• What a product or marketing team is most excited about.

There’s nothing wrong with starting with easy fixings, or following your passions when it comes to developing a new product. True success, though, comes when a company matches its offerings to the felt needs of our consumers. That’s why review analysis is critical to any business which hopes to see better ratings, increased adoption, and a loyal user basis. It allows a company to understand what its customers care about and prioritize, and provides a comprehensive body of information that can inform future development.

How do you analyze reviews?

Review analysis is important, but how is it done? The manual, time-consuming process of a decade ago is no longer needed, and modern tools take the grind out of turning your database of raw reviews into actionable insights.  The actual procedure is a simple three-step process that could hardly be easier:

1. Find a review analysis provider and integrate your reviews into their platform
2. Explore the analyzed results to find key issues, bugs, or feature requests
3. Find specific examples of key issues and share results with stakeholders.

Step 1: Find a review analysis provider and integrate your reviews into their platform

If you're stuck at #1, here's a hint: you're looking for a feedback analytics solution that uses thematic analysis paired with sentiment analysis to make sense of reviews at scale.

Call it review analysis software, or a feedback analytics platform. Our top recommendation: Thematic. Yes, we may be slightly biased. But it works like a dream, it's easy to use, and it's even got a free trial!

The beauty of Thematic is the way it works for any type or number of reviews.  Machine learning algorithms natural language processing enable it to make sense of almost anything. In fact, it can now even process chat logs!

Step 2: Explore the analyzed results to find key issues, bugs, or feature requests

Here's an example of a basic comparison of two map services: Google Maps and Waze. The person running the analysis wanted to know which was a better choice, and the data analysis showed that Waze offered better voice directions, Google Maps a more consistent GPS connection, and both of them had a loyal user base.

But that's only the beginning. Essentially, the feedback analytics platform sorts through the review data. It pulls out all common themes, no matter how they're worded. Sentiment analysis allows these themes to be tagged as positive reviews, negative reviews, or neutral reviews. Then the software  groups themes into trends and summarizes the data into simple color visualizations.

These visualizations make it immediately obvious which app has an edge-- and where.

Click through anything you'd like to know more about, and you can view both subthemes and the actual reviews which form the dataset.

This is especially important when you come to surprises in the meta-analysis-- for instance, take the trend 'android auto'. This theme is mentioned in 4.7 of Waze reviews-- and only 0.2 of Google Maps reviews. These reviews are all negative, but why?

Clicking through to the sub themes and then to one or two sample reviews allows us to discover the source of the problem: 9.1 of the reviews that mention android auto also talk about losing GPS. Google maps seems to play nicer with android auto, and the phrase is almost absent in negative reviews.

For another example, here's a review analysis of the differences between Power BI and Tableau, two corporate data visualization software options that have the same target options.

Again, Thematic is used. Public reviews are analyzed, trends discovered. The app uses natural language processing to turn qualitative reviews into quantitative data that can be graphed.

Tableau gets points on attractive visualizations and the ability to handle large datasets. Power BI is appreciated for affordable pricing and a lower learning curve.

Review analysis isn't limited to public customer reviews. Good analysis software is able to strip sensitive data from private customer correspondence.

Step 3: Find specific examples of key issues and share results with stakeholders.

It's a two-click procedure to turn confidential crash reports or chat logs into metrics and visualizations that can be shared with the whole team as well as external stakeholders.

For instance, Atom bank used Thematic review analysis on over seven feedback channels and three product lines, and was able to reduce call center volume by 40%.

Greyhound integrated their station and passenger data into the program and used Thematic to identify themes in user reviews, resulting in 20x reduction in analytics time.

Doordash used review analysis to analyze employee engagement feedback, and saw a 12 pt increase in employee NPTs.

What happens in each of those cases is that review analysis turns an enormous amount of inaccessible data and customer feedback into real-time actionable insights.

Knowledge is power, and each of these companies found themselves with far more power to grow, expand, and improve CX and product experience across the board.

Making review analysis work for you

So next time you look at an unorganized database of review data, remember: this isn't a collection of disparate, chaotic data points. It's not just a list of qualitative data, and it means a lot more than the star ratings.

With appropriate analytics and a review analysis tool like Thematic, this dataset can become a goldmine of knowledge.

It's a goldmine that can guide decision making in every part of your company, from customer service to development to your ten-year plan.