If the purpose of directional KPI for contact centers, such as FCR and CRM, is to tell us movement and trend, exact accuracy is unnecessary.

By Turaj Seyrafiaan

There is a classic quote by John D. Rockefeller that says, “Don’t be afraid to give up the good to go for the great!”. But is that good advice? As a consultant I have learned to say it depends!! It very much depends on the situation and what we are trying to achieve.

There are times that we should aim for excellence. As an example, we all know how important customer service and customer experience is and for those situations I would say yes, just being ‘Good’ is not good enough as customers expect more and more. We need to go for ‘Great’.

On the other hand, there are times when going for great can actually hold you back and stop you from achieving higher performance and that is what I like to talk about today. Perhaps the question should be: when is a good time to consider ‘good enough’? It all comes down to what is the purpose and expected outcome. Let me explain by giving you an example.

Working with a new client, we started discussing benefits of First Contact Resolution (FCR) and how important it is in improving the contact centre performance. My client agreed with all the points that I was making but then said that they do not measure First Contact Resolution. Looking at my puzzled face, he said they have looked into measuring FCR but have not been able to find a way to measure it accurately. “Accurately? What exactly do you mean by that?”, I asked. He said that he knew any FCR measurement will not be 100% accurate but he was looking for accuracy in the high 90’s!

Here is the problem. Certain KPIs, such as First Contact Resolution, are meaningful both in absolute and directional metrics. By directional metric, I am referring to a metric that can tell us if the performance has moved and in which direction (is the performance improving or deteriorating?). In that regard the level of accuracy is secondary. If we are comparing FCR performance in April versus March and we see that we have improved from 73% to 75%, does it matter if that 73% is not 100% accurate?

If we are planning to offer IVR self-serve for the most common call type, and we find the most common call type occurs twice as many as the second most common (42% versus 21%), does it matter if these percentages are 100% accurate? The clear answer is NO, it does not matter. As long as we are using the same methodology in measuring, the relative relationship between those numbers is correct and meaningful. We can take action based on that information. In trying to achieve high accuracy, many organizations miss, or ignore, valuable information that can assist in managing the overall performance.

Going back to the question of First Contact Resolution, there are number of quick and relatively easy measurement methodology to consider. One way to measure FCR is to ask your customers! Keep in mind that only a small percentage of customers complete surveys (higher percentage if calling). So, you do need to initiate enough surveys to get a reasonable sample size. At this point you are not looking for a statistically valid sample but a sample that is indicative. A much smaller sample with lower degree of confidence would still be good enough to provide reasonable measurement (of course the sample size relates to the size of the centre and the overall number of contacts). On top of the sampling issue is also the cost of collecting data directly from customers.

Another method, could be using internal data from CRM system (assuming that the organization has employed such system). Using CRM and disposition codes (call/contact types), one can look at the calls coming back from the same customer regarding the same topic in a pre-defined timeframe. The assumption is that these calls are not FCR. The benefit here is that there is a much larger population of data to work with. The shortfall, on the other hand, is the assumption itself. What if a repeat call was coded incorrectly? Or the repeat call came after the defined timeframe? Again, one can argue that even though the results are not 100% accurate, they are good enough to be used in our analysis; and that errors of coding or other exact definitions should wash out over the sample. Remember, when looking for indications and not absolute accuracy, good enough will work. As we can see, neither of these methods result in 100% accurate measurement but what is important is that we get a correct picture of relative performance.

I want to be clear, we are not talking about, and accepting, incorrect calculations (that would be wrong), but rather correct methodology with reasonable accuracy that matches the purpose of the measurements. In practice, any time that we look at a measurement, we need to ask what this metric is going to tell us, and how we are going to use such information. If the emphasis is directional and we are more interested in relative movement and trend, then ‘Good Enough’ is exactly what it is, GOOD ENOUGH!


Please Contact ApexCX for all your customer experience needs.

(May 19, 2021)