FCR is a popular topic we see on our call center consulting engagements.
Yesterday’s post dealt with the cost of ineffective call or contact resolution, citing an 80% First Contact Resolution (FCR) rate will add 25% to your average cost per contact and the importance of budgeting accurately to reflect the actual costs. In today’s post I wanted to examine a number of ways that FCR is measured in call centers and risks, benefits and various ‘gremlins’ that can influence the accuracy of your FCR statistics and present some ideas to help address or mitigate these issues.
Increasingly pundits and call center consultants like ourselves are promoting the use of FCR as the most valuable metrics for call center operations. It is difficult to argue against FCR as the perfect measure. On the surface it looks easy. We know customers and prospects are calling us to do something (pay a bill, order a product, get technical help etc.). Studies have consistently shown that when people get what they want, they are happier than when they do not.
For the time being let’s put aside the fact that successfully resolving an inquiry may not give the customer what they want: I want a refund says the customer and we quote David Spade in those old Capital One TV ads and say ‘No’. But FCR should be measuring whether the contact; call, inquiry was resolved, not whether the customer liked the resolution.
It can be challenging to measure FCR in a contact center environment. If you ask ten people how they do it you will hear a number of different responses. Some of the measurement approaches we have heard of include:
• Telephone Call Detail based – If the customer calls back within ‘X’ hours/days (48 hours, 72 hours 1 week), so the theory goes then we did not resolve the customers issue.
• IVR Survey based – What could be better than offering customers the ability to tell us how we did by offering them a post call survey.
• Agent based – The agent asks the customer if they have resolved the customers issue and this is then entered into the CRM or similar system.
Each of these approaches has benefits and potential risks or shortcomings. For example the Telephone Call Detail approach has the benefit of presenting a black and white picture of FCR. Once you have accepted the time window associated and accept the premise that the customer could have no other reason for calling again then the results have a good level of consistency.
Of course there may be reasons for the customer to call back: they ordered the wrong size, provided the wrong ship to address, received a new bill in the mail, have a second account with different issues etc. In the absence of robust analytics to provide a high level of data interrogation most centers will end up with a level of ‘false negatives’. That is to say that they will identify calls as not resolved when in fact the subsequent contact could be unrelated. This will mean a lower FCR score than they may actually be the case.
In contrast the Agent based approach of asking the customer if their inquiry was resolved before ending the original call, can also result in “false positives’. All of us who have worked as agents or with agents knows that there is a different perspective when speaking with a customer versus listening to the call or being the customer. The agents may ask the customer the question “Have I fully resolved your Inquiry” or something similar or they may not. The agent may simply check the box thinking that they asked the question or because they provided the appropriate response from the knowledgebase, so it must be resolved, right? Of course if the customer sounds unhappy or rushed the agent may choose to answer on the customers behalf etc. All of these scenarios will result in ‘false positives’ that is to say reporting that will indicate a higher FCR rate than likely exists.
One of the most prevalent solutions these days is the IVR survey, which in most cases 3 to 5 questions dealing with the call, as well as with overall satisfaction or net promoter etc. On the surface it appears to be a valid approach. What could be better than asking the customer? But depending on how it is deployed: by the agent seeking consent or before the agent answers the call by the IVR, you can have significant problems.
First as we looked at above the agents will not always offer the survey…in short they will or could play a triage role in limiting who gets into the IVR. Second if the caller agrees to participate before the call is directed to an agent, they may change their mind based on what happens with the call. Consumers and customers will ‘self-select’ whether or not to participate in any survey. If they believe it will help them they often participate, if there is little perceived value then they often will not participate. In consulting projects we have seen customers who ‘believe’ that their problem was resolved and are satisfied with the resolution they participate at a far lower rate than those who feel it was not resolved or who did not like the resolution. This illustration of ‘vested self interest’ can skew the results and reflect a lower FCR than actually exists.
Regardless of which of the above solutions is employed there are some other relevant issues that will influence the FCR reported. For example the customer believes or is promised that they will receive a credit, but that doesn’t appear on their next bill. In this case the customer and even the agent may believe the original contact was fully resolved, but it wasn’t. The same will be true if the product doesn’t arrive when expected (consumers hear 4 to 6 weeks and will expect it in exactly 4 weeks). The service isn’t restored when expected (told it would be back by 5 pm and at 5:01 they will call again); or the tech support routine they are told to run, doesn’t solve the problem. These are all examples of the customer not receiving what they expected, when they expected it. Of course the customers also contribute to FCR failures by not executing what they were to do (not following instructions) or providing inaccurate information of the original call (wrong size, incorrect address, an over limit credit card etc.) which will require a subsequent call.
As you can see effectively managing FCR is not easy and whatever approach you choose to employ, you will need to expect that it will take time for you to work out all of the exceptions, bugs and kinks. Remember that at the end of the day even if your model drives false positive or false negatives if you employ it consistently you will be able to chart improvement and declines period over period.