Actionable Insight Wanted: Why whole comment sentiment is not enough


Share on LinkedIn

How many times have you been to a job interview, had a good discussion with the interviewer and been contacted the following week to be told that they “decided to go with another candidate”? Pretty frustrating, isn’t it? It seemed like everything went well: you were able to mention all your relevant accomplishments and apply them to the job in question, the interviewer didn’t ask you any technical questions you couldn’t handle and you had a good rapport with the interviewer. You even found out that your daughters played on the same soccer team and spent an extra fifteen minutes talking about the upcoming season’s competition. So what went wrong? Why didn’t they choose you?

Now, image that after the same interview (and, in fact, all interviews) you were given a score from one to ten to rate how well you did in the interview. You would know if you made a positive or negative impression, and you could compare your scores over time to see if you were improving or not. Sounds great, right? But wait a second…how would that really help? Sure, it would be nice to know what kind of impression you made, but how would you know what to work on to improve your chances the next time? Wouldn’t it still just be a shot in the dark, and couldn’t you end up focusing your efforts and your resources completely on the wrong things?

This is the problem with most social CRM solutions today and a complaint we hear often. Google the term “social CRM” and you’ll find a host of companies offering social media monitoring and sentiment analysis. These services monitor social media channels like Facebook and Twitter, as well as the internet and other feedback channels, for discussions and comments about a company’s brand and products. And they aggregate all this data into easy-to-use, interactive dashboards. Customers are drawn in by with these features, excited to learn that they can see all discussions going on about them at glance. But, at the end of the day, many are still coming away unsatisfied. Why?

Most social CRM solutions do a great job of comprehensively counting brand and product mentions. They are also very good at summarizing data in interactive dashboards that display various graphs, charts and infographics showing current and historical trends broken down by brands, products, and various demographics, Still, the data they provide can pretty simply be summarized like this:

“4000 people talked about your Brand X this month compared to 3500 last month. 40% of them had a positive opinion, 20% neutral and 40% negative. Women like your products more than men, and people in France dislike them the most.”

These are interesting facts, but what actions can be taken based on this data to improve a company’s products, marketing campaigns or customer service?

To be really useful, feedback, tweets, and Facebook comments need to be broken down into multiple discussion topics, and sentiment needs to be evaluated for each topic within a comment. After this type of breakdown and evaluation, a summary of the analysis results would look more like this:

“4000 people talked about your Brand X this month compared to 3500 last month. 2400 of them were specifically discussing Brand X’s price – 60% had a negative opinion of the price compared to 30% last month. Women were more dissatisfied with Brand X’s advertising, and in France people were most concerned about Brand X’s environmental impact.”

There is a lot of insight here for a company to work with:

Action 1: Investigate why there was a jump in dissatisfaction with the price of Brand X: Was there a price increase? Did this price increase adversely affect sales?

Action 2: Investigate recent advertising campaigns that may have been offensive to women. Why and how did this happen? What action can be taken to remedy the situation?

Action 3: Look into the environmental impact of Brand X. Do customers in France have a misperception (informative marketing campaign needed), or are their concerns valid (possible study on how to reduce this impact)?

Of course, this type of analysis is technically far more difficult and can take years to develop. There are only a handful of services on the market today that can offer this level of information. But, as the glow begins to wear off so many current social media monitoring and sentiment analysis solutions, their customers are starting to ask what they are really getting out of these services. Many companies are already turning elsewhere to look for more in-depth analysis that can provide them real actionable insight.

What will happen to current social CRM solutions as their customers demand more actionable insight? Will they be reduced purely to data gathering tasks with other companies providing the analysis? Or will they be able to develop (or purchase) their own topic-level sentiment analysis?

Matti Airas
My main mission during the last three years have been trying to figure out how free-form text feedback (social media, forums, NPS, transaction queries etc.) analysis can improve overall customer satisfaction. Specialities: Customer Experience Management, Text Analytics, Internet applications and services, mobile applications, customer development and sales, management and startups.


Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here