Thanks to the computer age information has become so abundant that managers are having a hard time staying up to date and in focus on how the daily business is doing. That’s where metrics come into the picture. They provide a quick and reliable acid test to help us know where we are going and how fast we are getting there.
Deciding on which metrics to use, however, can be tricky. Textbook examples might prove a bit too theoretical. At the same time metrics based on gut feeling might be unreliable. A good approach is to set up customer metrics with the help of the customers themselves. In the following we discuss a very efficient way to create metrics that are both reliable and valid.
Coming up with your metrics
Say you want to have people rate their “customer experience” when they visited your store. Surely you couldn’t ask people to do so on a scale from 1 to 5, since it’s hard for people to understand the question and people would consider lots of different things when giving the rating. Needless to say it will result in low validity (the question isn’t measuring what it’s supposed to) and low reliability (the results will vary depending on who gives the answer). If you wanted to make a scale question with separate components (such as store cleanliness, customer service, product range, etc), which ones would you choose out of potentially millions of things?
Instead you should start out with a pilot survey intended to (now we’re using fancy research jargon) operationalize the concept. In plain terms it means that you try to find out which the things are that contribute to and make up the experience of visiting your shop. Once that’s clear, you can create a metric that’s measuring what it’s supposed to.
Setting up and fine-tuning the concept
Now you want to find out which things affect customer experience positively and negatively. The best thing to do this is to run a pilot study with two open-ended questions:
“When visiting our shop, what things did you find positive?”
“When visiting our shop, what things did you find negative?”
Do the pilot study and have people pour their hearts out. Run both questions on text mining software (such as that of Webropol) and find the biggest topics from both questions. There you have it – those are the things you should have your customers rate!
Say you get 10 things that are talked about a lot. Then you make a scale question where people are asked to rate these ten things. The sum of these will be the “customer experience” you were looking for in the first place. Then you can do the actual survey with this scale question.
Wait! There’s more than just the metric!
Once you’ve completed your pilot study you have actually got a lot of interesting information besides the metric. What you just did is that you:
a) Found out which things your customer experience is made up of
b) Found out which things add to or subtract from the experience (remember, you asked positive and negative things separately in the pilot study!)
c) Quantified your experience concept, i.e. you now have a numerical tool to use. This allows for studying customer experience trends over time, compare different customer segments, stores visited etc.
d) You have quantified the sub-components (e.g. product range) that contribute to the experience. You can also study these over time or between segments, stores etc.
For more information on text analytics and text mining, please contact us! Don't forget to visit our website!