Customer feedback clouds

Review analysis from customers

Syntactics Development
September 18, 2020
5 min read

Analysing customer reviews to understand free feedback

Any organisation can apply big data analytics to understand customer feedback. Interested how this works? We will show you how we did this in the following blog post.

Gathering data

So where to begin? It all starts with data creation also known as data mining. If you are an e-commerce organisation you probably already have some sort of customer review system in place. If not, that should be your starting point. Often websites as Trustpilot have a bunch of reviews online about you and your organisation. It is easy to ask customers for reviews, however, keep in mind the way you ask can bias the response. For example, many reviews are reward based. This means you will receive a discount on your next purchase or you are able to win something. These responders will be incentivised to review more positively. An incentivised review is not always positive for your marketing. New potential customers may respond negatively.

Processing data

The next step is getting your data ready for analysis. Depending of course on what kind of data you have this process differs. However, the next step is converting your reviews into text and into time-series data. Reviews are often only seen as text, while they contain very much time series information. For example, if you look at the past 100 reviews and you find a upward trend it shows you you are improving.

Interpretation of data

After applying several analyses, among others: LDA or NLP approaches and time-series analyses on the meta data, you can start interpretation. This may the most important step, turning your data into insights. This step may also be known to many as the Business Intelligence part. You will be able to identify main topics used in reviews by customers, think of shipping time, customer service, product and experience. After identifying topics, with a topic modelling or natural language processing technique you will be able to also consider these topics as time-series information. Find out how you have performed over time on each topic and whether or not your service is still up to par.

Automating insights

By now you will have gotten the hang of it. This is were the actual fun starts. You now know the feedback in the past, but what if a new review drops in? You can start training a review sentiment analysis algorithm to both identify topic and sentiment. Even cooler, you will be able to mark reviews as important to respond to or as must read for employees. Because what is more fun than to read positive feedback from customers.

Are you interested in analysing customer reviews or you want to start using your data for innovation or business intelligence? Just let us know and we will see what opportunities your organisation has.

Follow us on...
Return to Blog