Easy assembly. Easy access. Easy on, easy off. We’re so accustomed to these clichés that the word “easy” has lost much of its meaning. Yet it is at the core of any efficient call center operation. Although processes, technologies, and agents have evolved significantly to meet changing demands, a long-standing tenet of call center management hasn’t changed: getting the right contacts to the right places at the right times, and in the easiest manner possible. Accessibility — often referred to as service level — is as important as ever, particularly in competitive organizations such as catalog companies.
What follows is a description of the ten unchanging principles of managing call center service levels. Reviewing them often and incorporating them into both strategic and day-to-day operational decisions will ensure that the call center is efficient and accessible, building high-value, customer-pleasing service.
- Categorize contacts correctly
Service level is often used in a general sense to refer to an organization’s responsiveness to customer contacts. But when applied to resource calculations, service level has a specific definition: X percent of all contacts answered in Y seconds — for example, 90% of telephone calls answered within 20 seconds. It is a concrete and stable objective for transactions that must be handled when they arrive, such as inbound phone calls or text chat. Response time, defined as “100% response within N minutes/hours/days” is the related objective for transactions that don’t have to be handled when they arrive, such as customer e-mail messages or faxes.
Differentiating between service level and response time is essential because base staff calculations vary for these two major categories of contacts. For example, to differentiate service, you may decide to improve response time objectives for customer e-mail from same-day or next-day response to, say, within 15 minutes. Most queuing experts agree that for objectives of less than an hour, base staff requirements must be calculated using a queuing formula just as you would for a service level transaction. No problem with that. But some assumptions must also change. Service level, when applied to telephone calls, is based on when the calls reach agents and the conversations begin. However, in the case of e-mail, customers’ perceptions are geared to the time they receive responses after agents have handled the transactions. If you want to respond to e-mail messages within 15 minutes and they require an average handling time of five minutes, they must reach agents 10 minutes after customers send them. That becomes the yardstick for determining staffing requirements.
- Apply appropriate queuing calculations
Because of random call arrival, for those contacts that must be handled when they arrive, you must predict base staff requirements by using either a queuing formula that takes random call arrival into account (e.g., the widely used Erlang C formula) or computer simulation, which is more flexible for analyzing complex, contingency-based environments, such as skills-based routing. Figure 1 (p. 28) illustrates the tradeoffs among staff, average speed of answer, service level, occupancy, and network load (trunk load) for service level contacts.
To calculate the staff required for contacts that do not have to be handled when they arrive, you can generally use traditional methods of industrial planning. For example, if you have 60 customer e-mail messages that require an average of three minutes of processing time, you have 60 multiplied by 3, or 180 minutes, of workload to handle. Since there are 60 minutes in an hour, the work will require 180/60, or 3 base staff hours. That’s assuming 100% efficiency (which is not realistic), so you will need to adjust base staffing estimates upward. For instance, you might try dividing minimum staffing requirements by 0.9, to allow breathers in between the transactions. The basic formula for response time is shown in figure 2 on page 28.
- Improve service level by improving quality
The notion that service level and quality are at odds is one of the most common and fundamentally damaging misconceptions in call center management. As service level deteriorates, more and more callers are likely to verbalize their criticisms when their calls are finally answered. Poor accessibility will rob the call center of productivity.
When you consider the components of a high-quality contact, the complementary relationship between accessibility and quality becomes even clearer. What if data is not entered correctly? What if the customer isn’t confident that the contact was handled properly? What if the agent didn’t capture needed and useful information about the transaction? These problems contribute to repeat contacts, escalation of contacts to supervisors and managers, and complaints.
Furthermore, service level and quality in the call center will ultimately affect business unit strategy. For example, consider the impact on the organization’s workload when the call center helps manufacturing or fulfillment pinpoint quality problems or helps IT design better self-service systems. In short, when the call center eyes the larger implications of quality, it will have a positive effect on the organization’s workload, productivity, and quality.
- Look for pooling opportunities
The pooling principle is based on the laws of probability and is well rooted in telecommunications engineering practice. It states that any consolidation of resources results in improved traffic-carrying efficiency; any partitioning of resources will yield reduced traffic-carrying efficiency. Put more simply, if you bring together and cross-train small, specialized agent groups, you’ll have a more efficient environment (see figure 3).
But pooling is not an all-or-nothing proposition. Your objective should be to pool resources to the extent possible without jeopardizing the services that customers require.
- Ensure that service levels are in parity across contact channels
This doesn’t necessarily mean being equal — for example, it doesn’t mean that you reply to an e-mail as fast as you respond to a phone call. Rather, it means operating within customer expectations across contact channels. The customer who expects a reply to an e-mail within a few hours but doesn’t get it may pick up the phone and call. Now you’ve got two contacts going, which sends call center productivity down the drain.
What are your customers’ expectations? Ask them. They may expect a 24-hour response time for e-mail messages, and no more than a minute or two in a telephone queue. Also, tell them what to expect. Your Web site, customer literature, and other sources of information can help establish expectations as well as provide information or services that may obviate unnecessary contacts.
- Manage service level by increment
As illustrated in figure 1, 30 agents will provide a service level of about 24% of contacts answered in 20 seconds. With just one more agent, service level jumps to 45%, a quantum improvement. Adding one more person yields another big improvement. But keep adding, and more agents bring proportionally decreasing benefits.
This principle underscores the importance of getting the “right people in the right places at the right times” during each increment (typically, a half-hour) of the day. Being even slightly understaffed will cause big problems in terms of low service levels, high agent occupancy, and heavy telecommunications network usage. Also, those half-hours producing a service level of 100% in Y seconds may indicate that you have far more agents than you need during those times of the day. Getting workforce planning — forecasting, staffing, and scheduling — tuned up is a sure path to better performance.
- Don’t force occupancy to unattainable levels
Occupancy is the percentage of time during a half-hour that agents on the phones are in talk time and after-call work. The inverse of occupancy is the time agents spend waiting for inbound calls, plugged in and available. In the example in figure 1, a service level of around 80% of calls answered in 20 seconds equates to an occupancy of 86%. If the service level drops to 24% of calls answered in 20 seconds, occupancy goes up to 97%.
When service level gets better, occupancy goes down. Therefore, the average number of calls taken per individual also will go down. Some managers can’t stomach this reality — heaven forbid they should have any “unproductive” time in the call center! However, the time agents spend waiting for calls is sliced into 12 seconds here, two seconds there, and so on, the result of random call arrival.
In most call centers, agents handle various non-phone tasks when the inbound call load slows down. In fact, “blended” environments make a lot of sense because no one has a perfect forecast all of the time, and schedules don’t always perfectly match staff to the call load. But don’t be misled. When non-phone work is getting done, there are either (a) more agents on the phones than the base staff necessary to handle the call load at service level or (b) the service level objective is sacrificed. Don’t try to force occupancy higher than what base staffing calculations predict it will be.
- Support service level with multichannel workload planning
The Incoming Calls Management Institute (ICMI) defines call center management as “the art of having the right number of skilled people and supporting resources in place at the right times to handle an accurately forecast workload, at service level and with quality.” This definition can be boiled down to two major objectives: (1) Get the right people and supporting resources in the right places at the right times (service level) and (2) do the right things (quality).
Multichannel planning involves forecasting, staff and system calculations, scheduling, and real-time management. The most successful call centers have an established, systematic planning process (see figure 4). Systematic planning necessitates communication about issues such as resource allocation, budgeting, and workload priorities.
- Ensure that budgets reflect today’s workload dynamics
Although the convergence of Internet, telecommunications, and computer technologies enables organizations to automate many aspects of contact, it is also creating new types of services and multiplying the connections among customers, organizations, suppliers, and industry interest groups.
Organizations in virtually every industry have invested a growing portion of their budgets in call centers in recent years. The seemingly incessant growth in operational budgets (although declining when viewed in terms of cost per overall volume of contacts) is driven in large part by the age-old economic principle of elasticity: The more channels of access you provide and the better you make your service, the more services of all types your customers tend to use.
It’s important to do everything possible to provide and encourage customers to use automated support alternatives such as Web-based services. But it’s also necessary to be realistic and acknowledge the demand for agent-assisted services.
- View service level as an enabler
In the end, service level must be viewed in the context of a much larger objective: customer satisfaction and loyalty. Service level does not guarantee a satisfying customer experience. But it cannot be minimized, either. It is an enabler — an enabler of more important business to which you can then get on.
Brad Cleveland is president of the Incoming Calls Management Institute (ICMI), based in Annapolis, MD. He can be reached by phone at (410) 267-0700, ext. 958; by fax at (410) 267-0962; and by e-mail at [email protected].
Figure 1
| INPUT | Average talk time in seconds = 180 |
| Average after-call work (wrap-up) in seconds = 30 | |
| Calls per half-hour = 250 | |
| Service level objective in seconds = 20 |
| OUTPUT | Seated Agents | Average Speed of Answer (seconds) | Service Level (%) | Occupancy (%) | Network Load (hours) |
|---|---|---|---|---|---|
| 30 | 208.7 | 23.5 | 97 | 54.0 | |
| 31 | 74.7 | 45.2 | 94 | 35.4 | |
| 32 | 37.6 | 61.3 | 91 | 30.2 | |
| 33 | 21.3 | 73.0 | 88 | 28.0 | |
| 34 | 12.7 | 81.5 | 86 | 26.8 | |
| 35 | 7.8 | 87.5 | 83 | 26.1 | |
| 36 | 4.9 | 91.7 | 81 | 25.7 | |
| 37 | 3.1 | 94.6 | 79 | 25.4 | |
| 38 | 1.9 | 96.5 | 77 | 25.3 | |
| 39 | 1.2 | 97.8 | 75 | 25.2 | |
| 40 | 0.7 | 98.6 | 73 | 25.1 | |
| 41 | 0.5 | 99.2 | 71 | 25.1 | |
| Source: ICMI Inc. Note: Based on Erlang C, using half-hour data |
|||||
Figure 2
Volume/RT/AHT = Minimum Agents
Volume = Expected number of contacts
RT = Response time
AHT = Average handling time
Source: ICMI Inc.
Figure 3
| Calls in 1/2 Hour | Service Level | Seated Agents | Occupancy (%) | Avg. Calls per Agent |
|---|---|---|---|---|
| 50 | 80/20 | 9 | 65% | 5.6 |
| 100 | 80/20 | 15 | 78% | 6.7 |
| 500 | 80/20 | 65 | 90% | 7.7 |
| 1000 | 80/20 | 124 | 94% | 8.1 |
| Assumptions: Calls last 3.5 minutes. Calculations are based on Erlang C, for a half-hour of calls. | ||||