Customer Satisfaction & Text Analytics

Customer Satisfaction & Text Analytics: "

What the new co. car has to do with text analytics…

I believe this may well be the first Cavalry Blue 2011 Toyota FJ Cruiser to be delivered to the North East. I waited for it to come off the truck, untouched by any Gaijin.

If you’re from around here you can appreciate a good SUV in the winter. As a bit of an outdoorsman, the FJ caught my eye when it first came out a couple of years ago. Unfortunately they discontinued the Sunburst Yellow this year, my first choice.

Today’s post is not about cars, it’s about customer satisfaction and text analytics. As soon as the purchase was complete I was told I would be receiving a survey, and that if I wasn’t able to give them a perfect 10, please let them know now. Later that evening I received a phone call on my mobile with the same message.

Market research purists wince when they hear of surveys being administered in this way. Customers feel uncomfortable giving honest feedback. They know that the staff at the dealership will see the rating. What arises in the mind of customers like myself is more in line with game theory than honest feedback. “Shall I give them a 10 so they are happy if I need to see them for a problem?” or “Perhaps I should give them a 9 so they think they have to earn my trust/respect by doing a good job next time I come in for service?”. Either way, with auto dealership NPS (Net Promoter Scores) or OSAT (Overall Satisfaction) ratings, the distribution of ratings are not likely to be in the shape of a nice healthy poisson curve, but severely skewed.

Having worked with clients’ large scale customer satisfaction programs for several years now, I do understand that the benefit of these programs are often just as much or sometimes even more about a way to manage staff as about collecting the most reliable metrics. But obviously, the more a company allows this type of interference with what should ideally be an anonymous and unbiased process, the less valuable the results become as a management and quality monitoring tool. So why is this behavior allowed? Because while they may share some data with corporate, car dealerships are independent franchises. Therefore they are hard if not impossible to control.

One solution I believe is implementing text analytics into customer satisfaction programs [Full disclosure, Anderson Analytics has specialized in text analytics since 2005]. While in situations such as the one above, most customers will feel obliged to give a much higher rating than they otherwise would. However, if after the 10 or 11 point “Overall Satisfaction” or “Recommend” rating, an open ended comment question follows (which is now quite common), customers who may have given an untruthful overall rating are still very likely to be relatively truthful in their comment.

These comments will often contain specific mentions of parts of the sales process, and just as importantly in this case, may also give us a hint of actual sentiment and emotion. Considering the inherent bias in the survey administration process just described, this sentiment and emotion derived using text analytics can still give researchers a way to track and compare differences between locations over time.

Toyota, in case you’re listening, there’s a better way to you listen. ;)


[Post to Twitter] Tweet This Post


Popular posts from this blog

How to Build Your Brand Through Customers On Your Site

Plan de comercialización produtos tradicionais Galiza e Norte de Portugal

7th Master Class on EU Cohesion Policy for PhD Students & Early Career Researchers