MSHR 6610 Week 3 Contact Center Demand Analysis Example

What Arrives, When and How Long It Takes: A Demand Analysis of a Utility Contact Center's Calls by Half-Hour, Weekday and Billing Cycle, Separating Recurring Patterns From Storm Spikes

Student Name

Walden University

MSHR 6610: Aligning Human Resources with Business Operations

Instructor Name

Month Day, Year

What Arrives, When and How Long It Takes: A Demand Analysis of a Utility Contact Center's Calls by Half-Hour, Weekday and Billing Cycle, Separating Recurring Patterns From Storm Spikes

Why Demand Comes First

Requests for more agents at Northfork Power & Light's contact center usually begin with a supervisor's sense that the floor is busy. This analysis asks what the pattern behind that feeling is. It reads thirteen weeks of illustrative call history by half-hour interval, day of week and billing cycle, converts volume into workload and separates demand that recurs, which a roster should be built around, from demand that does not, such as storm outage calls. Aksin et al. (2007) describe forecasting call arrivals by interval as the foundation of call center staffing, since every later decision depends on it.

Volume by Interval and Day

Three patterns recur. Monday mornings carry the week's heaviest volume, as customers call about weekend problems. The two days after bills post bring a second peak, about 12 percent above an ordinary day, driven by billing questions. And the three hours after the dinner-time peak begins at five carry steady, substantial volume that falls off only slowly, since many residential customers call after work.

Table 1

Average Calls Offered per Half Hour, Selected Intervals, Thirteen-Week Average

Composite utility. Illustrative thirteen-week averages excluding the two storm days identified below.

From Volume to Workload

Calls do not take the same time at every hour. Daytime calls average about six and a half minutes, including after-call work. Evening calls are slightly longer, about seven minutes, because more involve payment arrangements. Multiplying calls by handle time gives workload in agent-hours per interval. A Monday 8:00 a.m. interval carries about 23 agent-hours of work in a half hour, needing about 46 agents' worth of talk time before any allowance for queuing; a Monday 6:00 p.m. interval carries about 16 agent-hours, or about 32 agents' worth. The evening workload is smaller than the morning peak but larger than the evening roster, which was built when evening volume was lower.

Outliers

Koole (2013) recommends removing such one-off events from the base forecast and planning for them separately. Two days in the thirteen weeks carried outage-driven volume three to four times normal, after a windstorm. Those calls were short, mostly people reporting outages and asking about restoration times, and most could be handled by the automated outage line. Including them in the averages would raise every interval's forecast and lead to a roster staffed for storms on ordinary days. The analysis excludes them from the recurring pattern and treats storm response as a separate plan, using cross-trained back-office staff and the automated line.

Seasonality and Trend

The thirteen weeks run from late winter into spring, when heating bills are high and disconnection notices peak. Summer volume is about 10 percent lower, except during heat waves. Over the past two years, total volume has grown about 3 percent a year, mostly in evening calls, as more customers work daytime hours away from home. That trend is part of why the evening roster has fallen behind: the roster was last rebuilt three years ago, when evening calls made up a smaller share of the day, and nobody has revisited it since.

What the Analysis Supports

Two smaller patterns are worth noting even though they do not change the roster. Lunch hours from noon to 1:30 p.m. carry a modest bump as customers call during their own breaks, and Friday afternoons are the lightest weekday interval, which makes them the natural time for training and team meetings that pull agents off the phones. Placing that off-phone time on Friday afternoons rather than Monday mornings would cost almost nothing and remove a self-inflicted shortfall the current schedule creates at the week's busiest hour. The analysis supports three conclusions for the staffing work that follows. Demand in the early evening is recurring, substantial and growing, and the roster should be built around it. The bill-posting peak is predictable and can be covered with scheduled overtime or shifted breaks on those days. Storm spikes should not shape the regular roster at all. Gans et al. (2003) caution that forecasts based on averages understate variation from one week to the next; the staffing model will therefore add a margin for normal variation but not for storms.

References

Aksin, Z., Armony, M., & Mehrotra, V. (2007). The modern call center: A multi-disciplinary perspective on operations management research. Production and Operations Management, 16(6), 665-688. https://doi.org/10.1111/j.1937-5956.2007.tb00288.x

Gans, N., Koole, G., & Mandelbaum, A. (2003). Telephone call centers: Tutorial, review, and research prospects. Manufacturing & Service Operations Management, 5(2), 79-141. https://doi.org/10.1287/msom.5.2.79.16071

Koole, G. (2013). Call center optimization. MG Books.