1. Incoming call center management is the art of having the right number of skilled people and supporting resources in place at the right times to handle an accurately forecasted workload, at service level and with quality.
2. Though average call load may be predictable, calls arrive randomly--which means that they often bunch up.
3. A service level is defined as "X percent of calls answered in Y seconds", not as "X percent answered" or "Average Speed of Answer". (The ASA is skewed by the bad times when calls bunch up.) Abandonment rates matter, too, but fixing abandonment problems usually means fixing service levels.
4. Service level and quality don't conflict. If you try to fix service level with poor quality, it comes back to bite you with more calls and demoralized reps.
5. A good forecasted call load--including talk time, after-call work, and volume--is critical for budgeting people and circuits. Often, a good forecast should predict load by the half hour, using previous data, knowledge of upcoming plans, and good judgment.
6. To determine staffing needs, use a variation of the Erlang C formula. Its input is the number of reps, number of callers forecasted, and the time to serve each caller; its output is a prediction of waiting time. (Even better, add an input for response time, and you'll get the percentage who'll wait longer than that!) If agents have different skills, you'll need forecasts and calculations for each set of agents.
7. More staff, less waiting, fewer phone lines for people on hold. Less staff, more waiting, more phone lines. Formulas exist for phone lines, too.
8. Not everyone scheduled is always working on customer service. Schedule accordingly. Be clever about work schedules to get the right number working at the right time. Service level results tell you whether you got it right.
9. If you have too few reps on duty, queues get long (service level goes down), more circuits are needed, and customers get frustrated, sometimes abandoning the call. If you have too many reps on duty, you spend too much paying for them to wait.
10. Give senior managers good reports, but make sure they understand the points above.
11. Monitor the number of calls in the queue and the longest current wait. Service level and other metrics tell more about the past than the present. Be ready with plans for unexpected load (reassigning, rerouting, delay announcements, busy signals).
12. There are lots of tools and graphs to measure aspects of quality. Use them to identify root causes, not beat your employees. Reps should adhere to schedules, and do good work. Use monitoring capabilities to coach. Measuring based on "calls per hour" is unreliable, and invites cheating.
13. Customers are getting more demanding, automated systems are taking the easy calls, so reps have to be better trained and more skilled.
14. Create a good environment that uses technology well.
The book was written in 1997, and I don't know whether it's been updated. The authors have some commentary about email-based, web-based, and CTI-based systems, but the next edition might want to say more about the similarities and differences between those and the traditional call center.
Overall, I'm happy to understand more about the math and science behind this discipline. As another reviewer commented, it's clear that IT Help Desks have something to learn from the Call Center experience.