The responses to your latest survey are being tabulated, and you’re excited. The initial results show that you’ve got a 92% customer satisfaction rate, over 75% of your customers like your new product line, and only 11% are dissatisfied with your prices. And here’s another piece of exciting news: 1,358 people took the time to write in answers to your three open-ended survey questions. The only problem is, you have no system in place to analyze this massive pile of unstructured data.
This is where text analytics comes in.
What is text analytics?
Text analytics refers to the process of deriving high-quality information from unstructured text, usually by categorizing the information and then looking for patterns and trends.
Open-ended text can provide deep insights
Most companies are very comfortable with metrics. But sometimes if all you look at is the metrics, you can miss some very important information.
A great example of this is a sports stadium concession stand. Their transactional data showed that their signature “Fresh-Squeezed Lemonade” was the most popular beverage option. Game after game, they couldn’t keep up with demand for the “Fresh-Squeezed Lemonade.” Quantitative survey data, however, indicated that fans actually hated the “Fresh-Squeezed Lemonade” and felt it was simply overpriced tap water! How could that be? The answer was found in the responses to the open-ended questions on the survey. Evidently the “Fresh-Squeezed Lemonade” was the only beverage served in a unique souvenir cup, so people bought it. The fans would be much happier if the beverage served in this cup was something else, in fact they would be willing to pay MORE to have beverage options.
Text analytics has come a long way
If customers are going to take the time to give you feedback about your products and services, I feel strongly that someone ought to be reading them. In addition to that, though, it’s important to analyze all of this text to also turn it into quantitative data.
Early in my career I was given the assignment of analyzing the responses to open-ended survey questions. Back then this meant reading each response and manually assigning it category codes. Was the response about price, product quality, speed of service, or what? Was the response positive or negative? If 65% of the comments were complaints about our service, we knew this was something that needed to be addressed.
Today there are text analysis and text mining software programs that will do all of this coding and tabulating for you. The two programs that we use here at MacKenzie Corporation are Clarabridge and SPSS Text Analytics. While these programs both require a lot of work on the front end to set things up, they save you many hours of work after that.
Getting the software set up requires expertise
The way the text analytics programs work is that you start by setting up category “libraries.” What categories do you want the data sorted into? What words must the text contain to be assigned to each category? For example, if the comment includes the words “hour” or “hours” you might want to assign it to the “business hours” category.
In addition, working with these programs is not a “set it and forget it” type of thing. As new themes, current events and industry trends come up you need to go in and tweak the models and the libraries to reflect this. This month people might be talking about a recall and next month they might be talking about the impact of something that’s happening in the economy that you didn’t account for during your initial setup.
Text analysis applies to social media, too
Of course, text analytics is not just for analyzing survey questions. It can also give you a systematic way to analyze all of the things that are being written about your company and products on social media sites.
For example, how do you know the sentiment of people who have posted about your company on Facebook or Twitter? One way is to read the comments and get a gut feel for it. This is very doable when there are 100 comments a day. But what if there are 1,000 or 100,000 comments? Text analytics software can be used to analyze this text for you, too.
MacKenzie Corporation can help you with text analytics
If you’re looking at setting up a text analytics program for your organization, give us a call. Although the text analytics software programs make things quite a bit easier than the old hand coding methods, it’s still not a “do it yourself” project for most companies. We understand the questions to ask and the way to set things up to ensure that you get useful, actionable data on the back end.
