September 26, 2019 | Curt Waugh

The Ultimate Open End: Unlock Insight with Text Analytics


Share on Facebook Share on LinkedIn Share on Twitter

Please tell us anything else you’d like to tell us.

Before we talk about that statement, let’s talk about analytics and CX more broadly.

“What would you say it is we do?” One of our senior managers asked me that during a rare idle moment at a trade show. I gave it some thought and replied, “I think we give organizations a ‘to do’ list around CX improvements.” While that’s a pretty accurate definition, having given it more thought, I see something more fundamental in our value to clients:  We structure CX data in a way that enables them to act on it. The “to do” list is just an artifact of that activity that allows organizations to socialize the work.

Looking Under the Analytics Hood

Verint ForeSee possesses the tools, techniques, and support that can structure almost any CX data and put it to work for a service-minded organization. ForeSee in particular brings two decades of experience in structuring incoming survey data through rigorous customer experience models. Among the solution’s many impressive qualities is the ability to pre-structure survey respondent data with an automated first-pass analysis, which quantifies the connection between customer experience and business outcomes. Note that in this process the data structuring is built into the survey design. It’s a gate through which the absolute unstructured thoughts and feelings of survey respondents are turned into actionable insight.

Another tool we use to structure data is Text Analytics (and its big brother, Speech Analytics). The power of natural language processing has reached the point where we (“we”) can read whole comments, determine an overall category for that comment, assign a sentiment score to it, and, in the case of CX Suite and modeled surveys, tie customer satisfaction measurements to the entire schema at any level. Go back just a handful of years and the best we could do was a keyword frequency report with no sense of sentiment in any of it. Now, the Text Analytics tool becomes the absolute gate through which unstructured open ends (or “squishy word data” as I’ve called it probably too many times in the past) are converted to insight.

All Data Is Good Data, But…

It’s important to note here that this Text Analytics gate is becoming or might already be the most important Voice of Customer (VoC) input we have. I mean, surveys are never going away. We need to ask questions and get direct feedback if we ever hope to make a customer’s experience everything it can be. But surveys will evolve and other data sources will complement and supplement them. Given this survey evolution and the rising importance of unstructured word data, we need to give more thought to how we handle the data, design it correctly, and feed the machine to get the most value out of it.

Look, Text Analytics can structure any open-ended data source. Have an old database with too many comments that you never read? Any comment cards laying around from an event? Set up a social media sniffer? Throw all of these feeds into Text Analytics and be amazed at what you can learn. But we can do better. Why not take advantage of the gate by using this critical moment to bring in fully unstructured data and let the system do its work? Why not feed it with responses to The Ultimate Open End?

The Power of Open Ends

By way of quick review, an open-ended question is one that does not elicit a yes/no answer. Did you get what you needed today? is a closed-ended question. How was your day? is open-ended. In court, closed-ended questions are called leading questions. That’s important to us here because we don’t want to lead anybody. The whole point of a mature VoC program is to let the customers lead us to a better customer experience. Let them speak to us.

The Ultimate Open End is a universal question that doesn’t bias the response. We often see questions on surveys similar to, “Please tell us how we could have improved your experience on our web site today.” This seemingly innocuous and friendly question is biased in at least three ways:

  1. It wants improvement suggestions. What if I don’t have one? What if I just want to criticize? I don’t know anything about web design. Why would you ask me that?
  2. It limits the response to the web site experience. But what about the support agent I talked to that sent me here? I don’t care about the search feature—where is my size?
  3. It only asks about today. I’ve been coming to your store weekly for as long as I can remember. What about what happened last week? What about my general improvement idea?

Let’s instead fashion an open end with no bias (or as little bias as possible). We’ll take out any hint of telling the customer what they should talk about. We’ll remove any sense that we’re requesting positive or negative feedback. We’ll extract the mention of time. My suggestion:

Please tell us anything else you’d like to tell us.

One could likely think of countless other ways to wordsmith this short statement, but that’s the basic idea. Put this question at the end of every survey throughout an entire VoC program.

With that in place, we are now feeding the Text Analytics engine perfect, raw, fully unstructured word data. It passes through our highly intelligent app and comes out categorized, scored, measured, and organized in an information hierarchy.

Wow—So Much Value!

Through this process we’ve introduced several value-adds to our CX program, all because we took advantage of the Text Analytics gate. People are now telling us about related experiences throughout their journeys, not just the website. Our customers and site visitors don’t speak marketing speak and don’t really care about the divisions within our organizations. Journey point this and silo that don’t exist in their world. They’re just dealing with your whole company and they want to tell you about whatever it is they did or failed to do. The minute you try to put it in a box, you will frustrate them.

People are now giving us both positive and negative feedback. Too often, we think of customer feedback as bug-report and find-and-fix. That’s part of it, but people are also delightfully positive sometimes and like to name names when somebody goes the distance to help them. Plus, sometimes everything works, there is nothing to fix, and folks still aren’t converting.

We also are now listening to the whole customer experience and not just some incident from today. We all have experienced service blips at our favorite store or watering hole and we keep going back. Let customers talk about their whole experience so you can do what it takes to build customers for life.

And, as discussed above, we can also run the exact same question across all our surveys. Text Analytics tools allow us to reach across all our listening posts and, in essence, find out what everybody’s talking about. We’ll see connections between journey points and service opportunities we might not have noticed – critical indirect feedback to complement the direct feedback of our surveys and inferred feedback analyzed in other data sources. We’ll get insight into how to funnel common service functions into less expensive spaces. We can learn information about ourselves that we didn’t think to ask.

The Ultimate Open End is the fuel that modern Text Analytics solutions need. It’s incredibly easy to implement and will add a ton of value to a solid CX program with little effort. Just as we design surveys to make them inherently actionable, so we can design other forms of customer feedback to do the same.

About the Author

Curt Waugh is a Sales Engineer and Solutions Consultant at Verint ForeSee. He partners with sales teams to develop custom designed CX programs that support business outcomes across all customer journey points.

More by this author


Share on Facebook Share on LinkedIn Share on Twitter