Leveraging Text Mining to drive Consumer Story and Insights
Leveraging Text Mining to drive Consumer Story and Insights
Text mining sounds like a sophisticated word, but it's quite a basic idea behind it. In short, this includes using the technology of machine learning to derive useful knowledge from text-based data (product feedback, emails, data from call centres, and more). Customer analysis teams use the text mining technique extensively, and it can help product marketers and analysts understand the pain points of their clients, how consumers use those products, and more.
Text Mining Techniques: A Quick Introduction
Text mining, as we have described, deals with the use of technology to extract information from text data. Perhaps the most obvious source of data is product reviews you find on websites like Amazon — but that apart from that, other data that can be analysed include:
· Point of sale data
· Social Data
· Data call centre
· Data from loyalty programs
· Data from apps
· Feedback via email
· Focus group statistics
Text mining is based on a three-part method for extracting information, with the three parts being entity extraction, categorization and analysis of sentiments.
Part One: Entity Extraction
The first step is to define which text to extract. If you're mining a database, the tech team will need to define the particular area that contains the document. If the text data is stored in several files, the files would need to be collected at a single location.
From here you can mine the text to extract structured data and apply the algorithms of machine learning to the source text.
Part Two: Categorization
Before you go ahead and analyse your data, definition and category models will have to be built that will help you make sense of your data.
Customer analysis teams usually find that the data that they have on their hands is connected to thousands of topics, many of which go well beyond the reach or jurisdiction. If that is the case, one approach is to hone in on the most common or commonly discussed concepts and ignore the other concepts that only appear in all of the data in one or two instances.
Part Three: Sentiment Analysis
Here's the hardest part — analyse your results. Customer analysis teams here use data mining techniques to reveal connections between the previously established concepts.
Clustering, grouping, and predictive modelling are the main techniques employed. Companies may either stop here, or go a step further in integrating their derived principles with other data sets to predict future behaviour.
The Power of Text Mining Techniques for Consumer Insights Teams
Customer intelligence teams may use text mining to accomplish a variety of business goals like learning about the pain points of their customers, knowing whether their product / service lives up to the standards of their customers and more.
Pain points and Unique Selling Proposition
First, consumer intelligence teams can use text mining to understand the pain points of their customers and understand why customers are choosing to use their product / service. (The latter is defined in business terminology as the Specific Selling Proposition of a Company or its USP).
For example, a company that sells vacuum-insulated lunch boxes may realize that their product fits many types of consumers — including those who want to eat better, those who want to save money and more.
Now, say the company is reviewing their data and finding out that the vast majority of their clients are struggling with a problem of weight loss. Now that the organization understands that its customers want to prepare their lunches so they can have more control over the food they eat, they can use this knowledge to fine-tune their marketing and sell accordingly.
Aside from copying and updating blogs, if the organization is trying to broaden its content marketing campaign, they might also build more brand collateral (blog posts, eBooks, brochures, etc.) that will teach customers how to cook healthier meals and use their lunch boxes to maintain a balanced lifestyle.
Expectations vs reality
Furthermore, text mining is also useful for businesses who are seeking to see if their product / service fulfils the requirements of the customers.
Picking up positive or negative feelings here is fairly simple — if your product reviews contain a lot of optimistic words like "fantastic," "perfect," "useful," then you'll know you 're on the right track. On the other hand, if your product reviews contain derogatory words, then it's worth taking a closer look at why your consumers are unsatisfied and your product changes accordingly.
Understand what customers like/dislike about your product
Working in the same vein, text mining often lets you discover what your customers like about your product and hate it.
Here you will probably find some contradictory details — Consumer A may say they like Feature A, but Consumer B may claim Feature A is useless and does not do anything for them. Considering this, look at what the data as a whole is showing, instead of being distracted by outliers and anomalies.
If you have more data (e.g. data on the Lifetime Value of your customers) at hand, consider reviewing the text along with these additional data. Will you find this:
· Most customers who don't like Feature A seem to have a low lifetime value and aren't going to hang around long, and that;
· Your consumers who make transactions regularly have no problem with Feature A.
If that is the case, obviously there is no need to change your product to please your once-off clients.
Benchmark your performance against the competition
Text mining will also allow you to test your results against your competitors on the move.
Consumers prefer to equate the product in hand with similar products which they have seen in product reviews. Here they may point out how advantageous your product is in certain respects, but otherwise it does not perform well.
Customer analysis teams get a better understanding with this knowledge on how their brand matches up against the other products on the market and determine whether they need to adapt by improving their product. If the brand operated by the team is priced at a premium and the organization supports the price by saying their product is the most successful on the market, then they would certainly want to look at improving the product in order to live up to those claims.