WFM Performance Measurement, Part I: Forecast Accuracy

This is the first of a 3-part series that will tackle the somewhat elusive application of effective performance measurements for your workforce management team.  I call it 'elusive' because it is apparent in almost every workforce management role I have served, or for every client I have consulted with, that these measurements have been in great need of implementation or improvement.

Workforce management professionals, by nature, are analytical, scrupulous, and (at times) frustrated with a lack of tangible, numerical data--they need to be graded, and assessed on some sort of a black-and-white, mathematical scale.  Telling them that they're 'doing a good job' is less satisfying and motivating than telling them (as an example) that they're not doing a good job because their forecast accuracy performance was 5% out of acceptable variance range.

For those of you who are not by nature or work experience WFM professionals, this series (hopefully) will help.  The series is broken into three parts in order to tackle the major aspects of workforce management service and support in the contact center: (i) forecasting, (ii) scheduling, and (iii) intraday/real-time management.

FORECAST ACCURACY

With regards to forecasting accuracy, it is important to for organizations to break up the forecasting responsibility into parts.  Here's an example:

1. Budget & Yearly Capacity Planning Forecast (Fiscal Year)
2. Long-Term Forecasting & Capacity Planning (Rolling 15-month, e.g.)
3. Short-Term Forecasting for Scheduling/Intraday Management

The reason why it is important to separate the forecasting responsibility when measuring performance is because each of the above-mentioned forecasts are (a) completed at different times, with a different degree and amount of information available, and (b) because the forecasts themselves are different.  For example, you do daily and interval forecasting in the Short-Term, but you do not do so for the Budget Forecast.  Another example is that, because the Budget Forecast is usually completed at least 3 months in advance of the beginning of the fiscal year for the entire year, there may be more lenient and flexible targets applied to the Budget Forecast than the Long-Term Forecast which, as an example, gets refreshed every 30 days with new information, with only the immediate 1-3 months' worth of forecast getting locked in for measurement purposes.

Yet, we're getting ahead of ourselves here.

The Overarching Authority on Forecast Accuracy: Performance vs. Indicators

Regardless of the forecast, the difference between measuring performance, and measuring indicators, needs to be understood.  When it comes to the performance of your forecast analysts, there is only one defining factor that should be considered: the variance of actual requirement to forecast requirement, exhibited in hours and/or FTE (full-time equivalent).  

In its simplest incarnation, WFM's responsibility is to accurately anticipate and provide coverage for business requirements.  Therefore, when measuring performance, indicators such as volume, AHT, shrinkage, or any other factor do not matter as the end result of those indicators is the requirement (either forecast or actual).

Comparing actual requirement to forecast requirement for performance purposes provides the complete picture--it combines all of your indicators into a succinct, ultimate result, and it reduces the impact of differing economies of scale.  For instance, based solely on volume accuracy, an analyst would have great difficulty achieving a forecast accuracy of +/-5% for a business line with volume of 100 calls or less in a week or monthly period--literally 5 calls can mean failure!  Conversely, a difference of 10,000 calls on a monthly volume of 500,000 may not seem like much...until you look at the difference in requirement that said volume represents when combined with AHT and other load factors.

This is not to say that indicators are not important--they are important, as they indicate reasons for variance of actual requirement to forecast...but should not be used when grading WFM quality of forecast.

Frequency & Consistency of Forecast Performance Measurement

You may have heard an idiom-spewing colleague say at one point or another that "the Devil is in the details"...and when it comes to forecast accuracy, they are especially correct.  Accuracy without a measure of consistency is ineffective.  Think about it in terms of measuring the speed of service that your organization offers to customers.  Merely measuring your average speed of answer (ASA) for a month, week, or even daily period is ineffective, in that you do not have a sense of what proportion of your customers are receiving the desired speed of service.  Your details are hidden in the average.  Take a look at an example comparison of service between two competing organizations below, who have an ASA target of 20s:

Company 'A', based on the target ASA of 20s, would be considered more successful than Company 'B', even though Company 'B' offered the desired speed of service to more customers!  60% of A's customers received the desired speed of service, versus 80% of B's customers.

The same mindset is applicable when measuring forecast accuracy for your WFM department.  It is insufficient to merely state your daily, weekly, or monthly variances were within acceptable thresholds--the consistency of such performance needs to be determined and monitored as well.

What does this mean in relation to the different levels of forecasting that were mentioned earlier?  Take a look below at some recommendations:

Budget Forecast Accuracy Standards

• Entire Year requirement variance must not exceed +/-X% AND X number of months out of the Entire Year must have variances not exceeding +/-X%.
• (You may want to get deeper, and additionally measure the variance between the budget forecast on a monthly basis, and what gets generated at all other levels of forecast as a quality control measure, too.   Why was the yearly budget so far off compared to other forecasts?  Check your indicators!)

Long-Term Forecast Accuracy Standards

• "Locked" Monthly requirement variance must not exceed +/-X% AND X number of weeks and/or X% of days within each month must have variances not exceeding +/-X%.

Short-Term Forecast Accuracy Standards

• Weekly requirement variance must not exceed +/-X% AND X number of days within each week must have variances not exceeding +/-X%.  ALSO
• X% of intervals within EACH DAY must have variances not exceeding +/-X%.

The actual level of performance expected of your WFM team wherever "X" exists above is up for you to decide, and should also consider current performance.  Where is the team at right now?  How long will it take for them to get to a desired level of performance?  What is holding them back?  When it comes to performance management, you want to challenge your employees, but also make goals and objectives attainable and reasonable given all impacting factors.

This concludes Part I of the series.  What methods of measuring forecast accuracy do you currently use?  What challenges do you experience?  Do you have doubts in regards to what has been presented?  I would love to hear your feedback!  Comment below, and stay tuned for Parts II and III.