Anecdotal Data​: The Twinkie of the Data World

By Diana Aviles

Anecdotal data is defined as observations collected in a casual or informal manner that relies heavily or entirely on personal testimony. For someone like myself who works with all sorts of data on a daily basis, I find myself engaging in heated discussions about anecdotal data being used as a replacement for speech analytics solutions. You typically find this mentality within human auditors who attempt to identify potential trends in contact centers by listening to small samples of calls.

Auditors feel that if they are just able to listen to 10 calls in a month for an agent they will be able to have a pretty decent idea of what the temperature is within their call center. The main problem with this is that’s only a sample size. If the average agent takes 500 calls a month and you are only cherry picking 10 calls from that month, do you really feel that you are accurately able to see the bigger picture? If we went with the 500 calls a month sample size, you’re only viewing 2% of an agent’s activity in one month. That leaves you with the remaining 98% being completely in the dark as to possible new trends and call drivers that may be occurring.

This leaves me to my next thought on anecdotal data, numbers. I can sit here and tell you until I am blue in the face, “We observed a few agents who did not go over the E911 disclaimer upon closing a home phone sale,” and it wouldn’t mean anything to you. Why? It does not indicate to the audience if it is a fact or an opinion. There’s the famous saying “Without data, you're just another person with an opinion.” When I present findings I like to think that my data is being looked at and taken seriously. I want the audience to be engaged with the findings; therefore

I feel that you do yourself a disservice by not presenting insights without a strong backbone. When you are able to provide a percentage in your findings you are showing that you took things a step further and put your money where your mouth is. So now you can say, “23% of agents sampled were observed not advising the customer of the E911 disclaimer at the point of sale.” Your audience now has a number attached to the observation. From that observation the data becomes actionable intelligence that allows an organization to take corrective action via agent coaching to remind agents to comply with the process laid out.

At this point you may be saying to yourself, “But Diana, I’m afraid it’s going to turn people off as jargon when I go and present it to my agents and supervisors, who may not be as analytically inclined.” This is an understandable concern. While you want to be factual with what you have found, you also don’t want to come off as robotic. We are human beings after all, and there are times when SOME subjectivity is understandable.

Anecdotal data, when presented correctly and in balanced doses, can allow an organization to determine if something is worth diving into further. This is why if we look at presenting insights as a packed lunch, your sandwich is your solid data and the Twinkie is the anecdotal findings. You wouldn’t want to eat a lunch of purely just Twinkies (full disclosure: I did that once as a kid with mixed results) because it is not a complete meal, just sugar and fluff.

Operations Manager, Speech Analytics 

With more than 4 years of Quality Assurance experience in a call center environment, Diana's objective is to simultaneously promote and educate the world of Speech Analytics with a human touch; one which further emphasizes the importance of First Call Resolution, and overall customer experience.

Follow Diana on LinkedIn.