Analyze Customer Satisfaction
With Standard Call Center Tools
For many contact center executives, answering this question is the call center equivalent to finding the Holy Grail: What is the real relationship between customer satisfaction and service level?
We manage very expensive operations and measure our operation’s throughput with metrics that are akin to manufacturing metrics: Service Level or Average Speed of Answer is similar to Manufacturing Queue Time, Agent Handle Time is similar to Manufacturing Cycle Time, and finally, Abandon Rate, which is similar to Manufacturing Defect Rate.
But, because we know that managing a contact center is, at its core, a process of managing a series of human interactions, we know that we are missing, by focusing on production metrics, the measure of how well we are doing on a touchy-feely score.
This is a recurrent theme, and there are many (expensive) ways that we have seen companies try to get in touch with their softer side.
All of us monitor our agents and provide quality scores. Many of us enlist third party customer polling firms that will gauge customer satisfaction by conducting a survey. Some of us will take that information and try to correlate the individual customer wait time with the overall satisfaction using mathematical techniques like regression modeling.
What we inevitably find is that the primary driver of satisfaction is agent demeanor and call resolution.
When I was a kid, I worked as a cook in a restaurant. It was a hot, hurried job. One day, after working up my courage, I asked the owner if I could move to a position waiting on tables (we only had waitresses) in order to make more money. The owner kindly stated that his policy was to use only women as wait-staff, because in a food service breakdown, with a smile, a woman could make everything better for the customer. I didn’t feel great about my smile, but took the words of wisdom back to the kitchen. In essence, all can be made better by a good interaction (and a smile).
But even that doesn’t mean people are happy with their wait. I might like the waitress and seem satisfied with the overall experience, but I can still be irritated at the wait. Similarly, our customers can be satisfied with the call resolution but be unhappy with the wait time. And the next time I call, as I am dialing, I will be irritated at my expected wait. As contact center managers, we know this too. We know we need both well-trained agents and decent production.
But there is one production metric that most of us already report, that is a direct measure of customer satisfaction. Rather, it is a direct measure of customer dissatisfaction. That measure is, of course, abandonment rate.
Service level and ASA are interesting because they give us good operational productivity measures. But those of us who have taken care to research the correlation between service level and customer satisfaction, usually find that there is only a very weak correlation between the two. Some of us have concluded that service level doesn’t matter nearly as much as we had anticipated (although we continue to set our company goals using service level).
Abandon rate is the only true metric that can judge, directly, some form of actual satisfaction; every hang up (after some minimal threshold time to abandon) is a somewhat unhappy customer. These customers actually voted with their feet.
You can argue that customer satisfaction surveys, that do not include abandons (again, past a minimal threshold), are truly underreporting the customer dissatisfaction. You cannot charm a customer who has hung up the phone.
So how do you use the bits of customer satisfaction information we have?
Our task is to determine the operational goals of our contact centers and to develop the long and short term plans that achieve these goals. Our job is to develop in our agents the skills required to turn out satisfied customers (as well as provide the resources and flexibility to allow them to do their jobs).
The former issue will be the focus of the rest of our discussion; if only because it is an area that I believe is less understood and is also in line with our expertise.
So the problem to work through is this: for a customer service department, what production goal should we staff to, keeping in mind that we must maintain customer satisfaction?
There are at least two ways to work through this: we can build a regression model that relates service level to customer service, and/or we can build an accurate model of abandons and use the model to analyze our service at varying staffing levels.
Like previously mentioned, regression models usually show a weak correlation between satisfaction and service levels. We believe that the better metric to focus on is abandons, and some very neat analysis can be done now, without having to hire a polling/survey company.
Many companies are now investing in simulation technologies to help with their long-term planning. Assuming that you have such a tool, and you have validated your simulation model against actual ACD performance data, you can create the following graph:
Graph 1: Service Level Versus Abandonment Rate
This graph demonstrates the relationship between service level and your abandonment rate, in effect drawing out the patience of your customer base. So which service level goal should we choose?
Clearly, this is a qualitative question. If we are managing a customer service function, we will want to balance many things:
- My costs
- Customer expectations
- Corporate mission/brand identity
- Availability of alternatives/competition
- Availability of alternative channels
Center executives have told me that, almost by definition, customer service functions have no profit component to their economics. They are cost centers only.
Our answer to this is that there is always an implied profit structure associated with your customer service function. You may not know what it is, but it is there nonetheless.
Let me explain. Associated with every service goal is a required staffing level, a required infrastructure cost, and an overall contact center cost required to hit that service goal. Once we set a goal, we really set a cost.
Every service goal also has an implied profit. That is, in a theoretical sense, we would like to manage our contact center as effectively and efficiently as possible, and hence we would like to manage to an optimal value.
Let me introduce the concept of POF. In contact centers, your POF is the implied cost associated with abandoning a call. It represents how much money it costs you to have an unhappy customer, and you can assign every abandon (past some short threshold) a POF value. You set your POF to compensate for the anger or dissatisfaction of a customer who abandoned.
Very few of us have ever tried to actually determine this cost. But by assigning, for example, a POF of $25 per abandoned call, I can plug that value into my simulation-based long term planning tool to draw this graph:
Graph 2: Service Level Versus Profit at a POF of $25 per Call
In this example, if we were managing our contact center like a business, and every abandoned call cost me $25, our profit would lead me to manage, optimally, to a 90% within 20 second service level standard (the star represents the profit maximizing point). This is the goal that optimizes our business.
A corollary to this logic is that, if we were managing our center with a service level goal of 90% within 20 seconds, our implied call value is $25. We are staffing our contact center as if each abandoned call costs us $25 in goodwill.
Let’s try another one.
Let’s say, instead, that we value each abandoned call at only $2 per call, keeping everything else about the contact center the same. Given the economics of our center and the patience of our customers, we can plot profit versus service level, and determine the profit maximizing service goal:
Graph 3: Service Level Versus Profit at a POF of $2 per Call
In this example, the profit maximizing service level is 62% within 20 seconds. If our service level goal was 62%, then our implied POF is $2. We are staffing our contact center as if each abandoned call costs us $2 in goodwill.
With our accurate center simulation model, costs, customer patience, contact center efficiencies, economies of scale, etc…
I can produce, in just a few minutes, the following graph:
Graph 4: POF Value Versus Optimal Service Level
This simple graph gives our decision makers a real intuitive view into their goal setting. They can choose whatever service standard they like, and this graph will describe the value per call they are implying. It works great to clarify service discussions.
By choosing a service goal, we truly are assuming a value per call and a cost of not satisfying our customers. And we all know what they say about the word “assume”.
To really understand customer satisfaction, it’s better to lay bare the assumptions we make and sic our simulation models on the analysis. As always, when we turn on the analytics, we make better decisions.
Ric Kosiba is a charter member of SWPP and co-founder and President of Bay Bridge Decision Technologies. He can be reached at EDK@BayBridgeTech.com or (410) 224-9883.
Copyright ©2008 Society of Workforce Planning Professionals. All rights reserved.