Keep up with all of the essential KM news with a FREE subscription to KMWorld magazine. Find out more and subscribe today!

What do your customers really think about you?

Using knowledge gained from studying customer sentiment is helping companies retain ongoing customers and land new ones. The idea of customer sentiment is nothing new. Customers can be the biggest promoters of a product or service, or they can be the main reason a product or service fails.

For years companies have looked to satisfaction surveys with ratings of 1 to 5 or 1 to 10 to determine how well they are meeting client needs. As helpful as those surveys can be, they still fall short in understanding the true customer sentiment because they include only structured data. Any unstructured data, like comments at the end of a survey or replies to open-ended questions, aren’t included. Without that information, it’s difficult to get at the root of some customer complaints and try to correct any issues.

That was the case with Sage North America, providers of 23 lines of business management software. The company conducted more than 100,000 surveys a year. If the average score on customer satisfaction was a 6 or less, the customer was considered a net detractor. If 7 to 8, the customer was neutral, and if 9 or 10, the customer was a net promoter of the company’s products, according to Hal Bloom, VP of market research.

Those surveys included customer comments, but with so many of them, it was difficult to draw conclusions. Bloom says, “We would have two to five people working on the surveys and they would sometimes come to different conclusions about what people were saying.”

About four years ago, the company sought a text analytics firm that could help automate and better compile that unstructured information. Sage chose a solution from Clarabridge. The software collects all sources of customer feedback, transforms it using natural language processing, categorizes the content, performs sentiment scoring and delivers customer insights enterprisewide through a variety of interfaces.

“We were impressed with their modeling and their technical point of view,” Bloom recalls. “At the time, text analytics was still in its infancy. The main thing is that they weren’t complacent with their solution. They talked to our technical staff to determine what our needs were, and they didn’t have a static package. They keep changing the model and better refining the sentiment measurements.”

Those measurements include not only major categories, like customer support, but also drill down to more refined areas like time on hold. Each customer’s experience is plotted in one of four quadrants—positive experience, positive sentiment, negative experience and negative sentiment. From there, Sage actively reaches out to customers in the negative quadrants to find out why they feel that way. Among the issues the company discovered, for example, was that many customers were upset when the company changed to a new platform for its ACT product in 2008 due to some underlying technical issues. As a result, the company put the disgruntled customers back on the previous version of the software, or upgraded them to the next version when it debuted, according to Bloom.

As a result of the more detailed customer experience information provided through Clarabridge, Sage North America’s number of net promoter scores (net promoters minus net detractors) has doubled in the last three years, and the number of customer recommendations has risen as well. Bloom says, “We attribute that to the use of text analytics.”

The hoagie and consistency

While Sage has been collecting customer sentiment information for several years, other firms outsource that function to professional survey companies. One of those companies, MindShare Technologies, had customers who wanted to get a better handle on true, insightful customer sentiment.

“We often asked structured questions, and the answers to those are easily quantified,” says Kurt Williams, Mindshare CTO. “The problem is that you have to go read those and the information isn’t very quantifiable.”

For example, a handful of customers might say service was slow, while another set of customers might say they waited a long time—another way of saying service was slow. Others might use different wording to mean the same thing. Each group might be relatively small, but pulling together all of the different terminology that means slow service can show the company has a significant problem that needs to be addressed, Williams says. His company’s customers want to have that kind of knowledge behind the customer sentiment surveys in order to make necessary changes.

KMWorld Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues