An Erwho? No, it’s an Erlang!

And other fundamentals about forecasting and staffing a contact center

By Shye Louis

Photo by Jordan Ladikos; Unsplash.com

If you manage a crisis center you probably struggle with scheduling crisis counselors — do you have the right number of counselors on duty at the time that they are most needed? You probably have some strong intuition about when your busiest times of year, month, week and day are, but do you know how to actually forecast your future call volume? Perhaps, if you have the good fortune to have an ACD and Workforce Management software tools. But even if you are a small center, you can benefit from learning more about the science of forecasting and staffing.

Forecasting:
In order to best figure out how to use your limited resources of crisis counselors, it’s important to figure out how much work they will be needed to do and at what times of day. Much like your local meteorologist we need to look at history and patterns to make a forecast to predict future workload. The first step is to gather historical data about your incoming calls — 2 years worth of data is best if you have it, but use the past year or whatever you’ve got to help you look at patterns. The more granular you can sort the historical data, the better, down to the half-hour intervals if you can. This is easiest if you have an ACD that can generate these reports for you, but it’s not impossible to figure out using data from whatever tool you use to document your calls (if your software documents call start and end times and your staff enters their calls in close to real time). You may need to adjust for anomalies (like a day your phone system was down for repairs for example).

Predicting Monthly Volume
You can predict monthly volume based on your historical call data using a few different methods:
• A point estimate: assumes any point in the future will match the corresponding point in the past. (April 1st = August 1st). This method does not account for any trends up or down or atypical data from the past date.
• Averaging: simple averaging of past numbers to a moving average where older data is dropped out, or a weighted average where more recent events are given more weight. Weighted averages are harder but more accurate.
• Time series: most accurate approach, looks at historical info in the context of trends and seasonal differences. Not really possible without a workforce management software.

Creating Daily/Half-Hourly Forecasts
A few weeks of clear data is usually sufficient for creating these types of forecasts. Compare each day of the week’s numbers to the weekly total to get percentages that reflect your day of week patterns. Repeat the same process for your half-hourly patterns — for example look at several Mondays by half-hour to see how each half-hour compares to daily totals.

Workload:
What do you do with all of this information? You use it, in conjunction with what you know about your average handle time for calls (including any time after the call the crisis counselor will take before the next call to complete documentation, etc.) to calculate your workload (number of calls forecast for the hour or half-hour, multiplied by average handle time). Remember though, that average handle times can vary but time of day — especially at crisis centers, with the longer calls often falling later and later at night. This is why the forecasts by hour or by half-hour are important to take the time to do (instead of just calculating an hourly forecast as an average across the entire day’s volume).

Having workload information lets you figure out how many base staff you might need to handle the calls. But contact centers, and crisis center especially, need to also take into account that our workload isn’t able to be tackled sequentially, our call volume arrives whenever people decide to call us, it’s random (feast or famine, right?), so we need to have more crisis counselor hours on our schedule than hours of actual call time would suggest. How do you figure this out?

Erlang C:
An Erlang C is a traffic engineering model that helps us figure out how to use our staffing resources to accomplish our defined workload. It’s a complicated formula (see pic below, eep!!), named for a Danish telephone engineer, that allows contact centers to determine how many agents they need to staff, based on the number of calls per hour, the average handle time of calls and the average delay before answer.

No one expects you to memorize this formula, so don’t be scared! If you have a workforce management tool this calculation is built in. If you don’t, there are many free Erlang C calculators online — just Google “free Erlang C calculator” to find what you need. Here is one example:

And here is a sample chart showing different results from the Erlang C calculator with the number of staff required to manage workload depending on different call volume, average handle time and average delays:

Relationship of Staffing and Service:
Adding more and more staff results in diminishing returns, with less and less impact as the numbers get higher. Service will worsen dramatically the closer the number of staff gets to the number of hours of workload. One person extra or one person missing from your coverage can be enough to dramatically improve service, or send service spiraling out of control.

This was overwhelming — how do I learn more?
If this stuff is new to you, it takes a while to wrap your brain around how to use it. Especially if your systems don’t have all the bells and whistles that make it easier. It can still be worth it to take the time to understand more about how to predict your call volume and staffing needs. Having this kind of information at hand can be helpful to defend the need to hire more staff or recruit volunteers. It can also be a big to help for staff to be able to understand necessary scheduling changes you might need to make to better manage your call volume. For a deeper dive on this subject, with lots of examples and opportunities to try it out yourself, check out the courses on workforce management available from the Lifeline’s new member benefit, the ICMI Online Training Pass. Details on the benefit here: https://medium.com/@CrisisCenterBlog/a-new-benefit-for-the-new-year-b2343ce8c00c#.sx8wli86g


Shye Louis is the Coordinator of Best Practices in Suicide Prevention at the National Suicide Prevention Lifeline, a program administered by the Mental Health Association of New York City.

If you’re involved with a crisis center and interested in joining the Lifeline, a network of over 160 crisis centers around the country, please emaillifelineinfo@mhaofnyc.org.

If you or someone you know is struggling with depression or thoughts of suicide, reach out. The Lifeline is available 24/7 at 1–800–273-TALK (8255).