WFM Survey Results

Each quarter SWPP surveys the workforce planning community on critical workforce planning topics. Over 200 call center professionals representing a wide variety of industries participated and provided insight into this  quarter’s survey on forecast accuracy.

Participant Profile

Almost half of the survey participants (45%) are from call centers with over 500 agents. Sixteen percent are from centers with between 101-200 agents, and 13% from centers with 201-300 agents. Another 16% are from centers with less than 100 agents. The largest group of participants are from financial, insurance, and the health care industry.

Frequency of New Forecast


Over half of the respondents indicate that they create a new forecast weekly while 25% do it monthly. Slightly over 10% combined indicate that new forecasts are created quarterly, annually, or rarely. More frequent forecasts are needed in centers that experience significant fluctuations and that have the flexibility to adjust the staffing to match it. Those reforecasting less frequently may represent the smaller and more stable environments that  experience little change and are also unlikely to have much scheduling flexibility.

Variance Between Forecast and Actual Volume


Variances at the daily, weekly, and monthly levels are reported to be lower than for the half-hourly level. Many centers focus on forecasting accuracy measurement at a daily or longer level, which can conceal some significant fluctuations in the intervals. However, analysis of the interval variance can be revealing and assist the center in spotting some consistencies in the times or days of the bigger variances. For example, if the analysis shows a consistently high variance between 10AM and 11AM on Tuesdays, digging into the reason may be easier.

Drivers of Call Volume Changes


When asked what drives changes in the call volume, the respondents provided a wide variety of answers and were able to select more than one driver. Weather, system outages, and marketing efforts lead the choices with billing cycles and mailings close behind. Only a few reported that there are no known drivers.

Variance Between Forecast and Actual AHT


As with the variance on volume, the AHT variance is lower for daily and longer periods than it is for half-hour intervals. The highest variance reported is in the 16 to 20% range.

However, the variance is reported at over 20% at the monthly level. With volume and AHT equal partners in the calculation of the workload to be handled, it is just as important to achieve AHT forecast accuracy as it is for volume. However, it is not common to see the WFM department concentrate as much attention on the AHT accuracy measurement or improvement strategies.

Drivers of Changes in AHT


When asked to indicate as many drivers of change in AHT as appropriate, the largest number of respondents chose call center systems. Product releases and marketing offers are a distant second. It is clear the reasons for volume changes and AHT changes are not the same and each needs to be addressed as a separate issue in efforts to improve forecasting accuracy overall.

One Thing That Could Help

And finally, the participants were asked to name one thing that would help the WFM team to achieve a more consistently accurate forecast. There were many responses, but quite a few focused on communication and technology. Here is a sampling of the answers:

  • Operations partners communicating more frequently with WFM on upcoming events that could impact call volume and/or AHT.
  • Knowing when mail drops occur at all times.
  • Teamwork.
  • Currently we don’t have WFM software tools. All forecasting is done via ACD reports and Excel. I feel confident with the addition of WFM software we would be able to have a more accurate forecast across all channels and intervals.
  • Establishment of a central communication hub to filter information from all departments that could have an impact on calls so that when they have a new initiative it is properly communicated to the WFM with enough lead time instead of posting as an afterthought.
  • One thing that would improve our forecast would be increased information from agents in regards to call trends/drivers. We currently get some trends from floor agents, however most of our call data comes from pulling Natural Language tags in the IVR.
  • A true focus on planning, as opposed to a combined focus on planning and operational delivery on the plan.
  • It would be helpful if the WFM tools were more accurate in their automation. We use a tool and end up building the forecast outside of the tool and then pasting it in the tool.
  • Time. More time to devote exclusively to forecasting.
  • A workforce management solution that can better mimic our actual ACD skilling environment. Our organization has a switch that routes calls based on “greatest need” rather than skill level at the agent level.
  • We usually come within a percent or two so I can’t think of anything else. Our AHT forecast is off due to the business wanting to reach a goal that, at this time, is unrealistic.
  • Speech analytics!
  • We currently have multiple queues that make forecasting volume and AHT more difficult. Next year we are planning to move to a multi-channel smart-routing system that will allow us to track total call volume more accurately.
  • A crystal ball!

Closing Comments

Based on the responses, it appears that forecasting accuracy is measured by many centers and the ability to look at volume separately from AHT is also in place. For most, the purpose of the measurement is to support analysis that will lead to improvements. Digging into the details of both volume and AHT fluctuations separately is important. However, it is also key to look at the interval level data to find any consistent times of variation where the causes might be more easily identified.