Pay for Performance
Whenever I have an opportunity to visit a business partner’s call center, I take a few minutes to conduct a rather un-scientific test, call it morbid curiosity. As I pass by cubicles and am introduced to call center staff, I always ask how agent performance is assessed. To me, the variety of responses I hear speaks volumes and perhaps helps explain the responses I get from call center agents when I pose the exact same question. Typical responses are a shrug of the shoulders, a shaking of the head and a quick glance to co-workers for reinforcement. They don’t know, feel they cannot explain the complexities or simply don’t remember.
Call center agents are expected to know and retain more and more information in today’s complex business environments. Unfortunately, our short-term and working memory capacities have not increased to accommodate this environment. Agents are also expected to generate customer perceptions that the service was excellent while managing the call to the operational metrics. Talk about feeling committed…to an asylum!
If you want your agents to feel less like they NEED to be committed but rather be committed to the customer experience, keep in mind and act in accordance with the very basics of their job expectations. And be clear about those expectations. One business partner I visited earlier in 2010 had their KPIs updated in real-time on dozens of flat screens in the call center. Another business partner created banners as a colorful reminder of the quarter’s call center initiatives. Great ideas because the agents understood why the KPIs were important, what impact they individually have on them and how they benefit from effectively performing to these. Don’t assume this is the case without testing your assumption.
A balanced scorecard can serve as a visual cue for agent success. Well-designed balanced scorecards are typically made up of four parts:
1. Metrics by which performance will be assessed,
2. Performance objectives for each metric,
3. Weighting applied to each metric (an indication of relative importance), and
4. An individual agent’s performance on each metric.
Selecting performance metrics
Traditionally, call centers have managed performance based on the goal of operational efficiency. We see this drive for efficiency continue today through the management of call center agents to metrics like average handle time, number of calls handled, after call work, etc. While these are very important metrics, the fallacy is in managing a call center to only these internally-focused metrics. Customers do not care how much time an agent has to spend filling out paper-work or electronic forms after the call ends. Rather they care that an agent is available within a reasonable amount of time when they call. Customers care even less about how long they need to spend on the phone with a (single) call center agent, as long as the problem has been resolved with that call. A 30-minute call might end with a delighted customer, a frazzled call center manager and a very confused agent. Are you seeing the disposition toward schizophrenia now?
The key to success is in selecting a variety of metrics that speak to the customer experience and balancing them with the business need for efficiency (we will speak more about this balance when we talk about scorecard weighting). Best-in-class business partners also incorporate other data sources into their agent scorecards such as internal quality monitoring data, chat, text, SMS and email data, etc.
Setting performance objectives
We’ve all been in situations where a goal was picked out of thin-air by a well-intentioned executive and then carved into stone for us to follow. In the absence of this scenario, the best set goals are based on actual historical performance. Important elements to consider in setting performance goals are:
- Mean or Median? (measures of central tendency)- In order to set a goal for future performance, we must first have an understanding of how we’ve performed in the past. Measures of central tendency indicate the point on a performance continuum where the members of a group or dataset tend to gather. While the mean (often referred to as the average) is more widely-reported in call centers, it is most useful in groups whose performance is relatively normal (normal from a statistical standpoint, that is). A normal distribution is one in which a majority of group member performance is centered around the middle of the performance continuum and the distribution of performance is perfectly symmetrical to the right and left– in short, a bell curve. Unfortunately, this type of distribution is not typical of call center performance. As such, the median (the point at which half of the group’s members fall above and below) may be a better way to determine how the call center “typically” performs on any given metric.
- Time frame of historical data– Having decided whether the mean or median will be the most appropriate statistic for determining a baselineof past performance, we must now define a time frame to represent history. At bare minimum, Customer Relationship Metrics recommends that at least three months of data be used to minimize the impact of anomalies in performance and non-normative events impacting performance. Ideally, a larger time frame would be used which encompasses all stages of a company’s business cycle or seasons (1 year). The danger in using more than a single year of historical data to establish a performance baseline is the possibility of negating or underplaying recent performance gains – essentially making the performance goal too easy to achieve.
- Predicting the future – Once a historical baseline of performance has been established, the same data set can then be used to make predictions about future performance (statistical modeling). Performance objectives can then be based around those predictions. Some business partners have also found some success if applying a 5% to 7% “lift” to historical performance and using that lift as the performance goal for the following year.
The weighting applied to each metric on a scorecard indicates its relative importance to the call center and to the larger organization. Before arbitrarily applying weighting or points to each metric, think about the organizational goals that have been set for the fiscal year and the ways in which the call center contributes to these goals. Doing so will help you make the first critical decision – whether to focus on the customer’s experience or on organizational costs. Weighting within each category of metrics (operational vs. customer experience, etc.) can then be determined based on the degree of impact each metrics has on the category outcome (ex: issue resolution has a higher impact on customer experience than courtesy, so issue resolution should have a higher point or weighting allocation associated with it).
Individual agent performance
If one of your goals in implementing a balanced agent scorecard is to keep agents informed about their performance and incite healthy competition, ensuring that your agents have ready access to accurate scorecards will be a key determinant in the success of the initiative.
During one of my recent visits to a business partner, I took my usual walk through the call center and was quite pleased to see the number of agents who were logged in to Customer Relationship Metrics’ MPM real-time agent scorecards. MPM (Metrics Performance Manager) is a reporting tool that Customer Relationship Metrics uses as part of our applied business intelligence services to gather data from disparate sources. We’ve found that one of the outputs of this reporting tool that can be very motivating to agents are the scorecards. In this call center, agents were actively managing their own performance and receiving immediate feedback from the system about the changes they were making to their interactions with customers. Feedback from the ACD about their efficiency, feedback from customer satisfaction surveys, feedback from the Quality Assurance team are all in a single location, updated in real-time. Imagine the burden you would remove from your supervisors if your agents were that tuned-in to their own performance!
A few weeks ago, I was reading an interesting article about schizophrenia. It talked about the statistics, symptoms and treatment for this terrible disease. At first I was alarmed by the recent research numbers, an estimated 3.2 million Americans suffer from this mental illness. Wow. As I read on, I learned that four types of “delusions” exist in schizophrenics, and from that list of four, “Delusions of Control” is one that really struck a chord with me. Naturally, I started to draw some parallels between this particular symptom and people I know, myself and those in my line of work. I do believe it’s fair to say that based on the delusion of control alone, we all have a touch of schizophrenia from time to time. Perceived control is a way of life in the call center.
When reflecting on the life inside a call center, it’s easy to believe that we are patients that are often not medicated to control our delusions. The call center as an asylum may not be a stretch! Not only is it insanely intense, it is also a place of constant contradiction. We often have expectations of our employees and our call center agents to adhere to a specific model intended to produce a controlled response (a great service experience). In the same breath, we also expect that model to produce the opposite results (do it fast, right and cheap). Isn’t this setting your team up to feel schizophrenic? We allow agents to believe they are in control, but in reality, they are not.
I was reminded of this parallel when speaking with one of our partners last week. This particular client had three service centers that were using the “Pay for Performance” model with their agents. As he elaborated on the damages this was causing, I began to recall the correlation between my recent revelation on call center schizophrenia and the “Pay for Performance” model (particularly in service orientated call centers.) In this particular model, agents are being paid based on metrics such as number of calls handled and number of minutes spent on those calls. This is the expectation set forth. At the end of the month, organizations are left scratching their heads as to why customer satisfaction scores are so low. Well, the innate service component is being squished out of the agent as they are trying to hurry on to the next caller. But yet, we are expecting an outstanding customer service experience to come from our service orientated call center, right? Insanity in its true form and we’ve all had this conversation with ourselves and everyone on the management team.
This will be the first in a two-part series focusing on designing the perfect, or as-perfect-as-you-can-get, model for service call centers. Part One will discuss the “Pay for Performance” model, how it has been incorporated in the service call centers and how it is affecting your agents and your customer service scores. Part Two will discuss how to build effective balanced scorecards and, in turn, a more appropriate model to your service call centers. We need to control the insanity!
What is “Pay for Performance?”
“Pay for performance” also known as incentive pay, rewards workers based on the outcomes they achieve as opposed to the traditional model of paying for time worked. These models have been wildly popular in outbound telemarketing for many years, advancing the earning potential of skilled salespeople while “weeding out” those who in a conventional pay model would largely rely on their base salary to pay their bills. More sophisticated (sales) incentive pay models financially penalize agents for “buyer’s remorse,” encouraging quality sales acquisition methods.
Sales vs. Service
While time and outcome-pressured compensation models may work in a sales environment, they represent the antithesis of what is needed in the service world. Conventional wisdom states that in a sales environment, there is only one outcome that matters — sales that “Stick”. Certainly there are complexities in how a call center agent reaches a “Yes,” but that does not negate the fact that there is only one outcome that is guiding the call flow. A customer service call center is far more complex. Customer service call center agents are tasked with resolving calls in a manner that is pleasing to customers, builds brand loyalty while remaining sensitive to everyone’s time – the customer on the phone and the one waiting to be helped. That is quite a tall order especially when a case can be made that the Sales team is often responsible for the call to the Service team. At the end of the day, incenting agents based on a single outcome may expose your organization to a very high level of business risk.
“Why are my customer service scores so low?”
The correlation between time spent and outcome is much more fluid. Let’s examine some of the unfortunate outcomes of ill-conceived pay for performance models in customer service centers:
As an assignment, add your metrics to this list and evaluate them against the delusion of control construct.
“If not “Pay for Performance”, then what should we use?”
In a service center, a balanced agent scorecard is a far more effective way to pay and incent agents. Balanced scorecards force agents and their managers to focus their attention on more than a single Key Performance Indicator (KPI). Some of CRM’s existing customers have access to an important tool which assists them in determining the relative importance of agent skills, from the perspective of the customer – predictive (regression) modeling. In the figure below, the beta levels on the left indicate the level of impact each agent skill had on the customer’s overall perception of the agent’s performance. The right side of the figure indicates the current performance level of that skill (on a 1 to 9 scale).
One business partner uses this regression output to not only set priorities within their agent scorecard, but to also set priorities for ongoing / developmental training for the upcoming quarter. The figure below indicates the degree of improvement in customer satisfaction this business partner has been able to achieve by linking customer, agent and training priorities.
Companies using the “Pay for Performance” model in their service call centers will remain to be at war with themselves. If you are paying your agents on the number of calls they take, then you will get a high number of repeat callers and lower FCR rates due to customers being rushed off the call. If you are currently using the “Pay for “Performance” model in your service center, have you experienced similar results?
Now that we identified a much more suitable, more effective model to adopt in your service call centers, it’s time to discuss “how.” Part two will discuss just how you can build effective balanced scorecards to incentivize your agents.