Forecast accuracy

Forecast accuracy is a workforce management (WFM) metric that quantifies how closely a contact center's predicted contact volume — or handle time, or staffing requirement — matched what actually occurred over a given period. High forecast accuracy means the operation was staffed close to what demand required; low accuracy means either chronic over- or understaffing, each with distinct cost and quality consequences. Forecast accuracy is foundational to contact center planning because virtually every other staffing and scheduling decision depends on the volume forecast as its primary input.

An illustrative case: a WFM team forecasts 1,200 inbound contacts for the 9:00-9:30 AM interval on a Monday. The actual volume turns out to be 1,440. The percentage error for that interval is 20% — the team understaffed by roughly one in five contacts, driving up hold times and degrading customer experience for that half-hour window. Multiplied across dozens of intervals per day and hundreds of days per year, systematic forecast error is one of the highest-leverage cost and quality problems in contact center operations.

How forecast accuracy is measured

The most common formula for forecast accuracy in contact centers is derived from Mean Absolute Percentage Error (MAPE):

MAPE = (1/n) x sum of (|Actual - Forecast| / Actual x 100)

Where n is the number of intervals being evaluated. MAPE is expressed as a percentage of the actual volume. Forecast accuracy is then stated as the complement: Forecast Accuracy = 100% - MAPE. An operation with 15% average MAPE has forecast accuracy of 85%.

Some teams use Symmetric MAPE (sMAPE) to address a mathematical asymmetry in standard MAPE: when actual volume is very low (near zero), even small absolute errors produce very high percentage errors that distort the average. sMAPE uses the average of actual and forecast in the denominator rather than actual alone. The choice between MAPE and sMAPE should be consistent within an organization so that benchmark comparisons are valid.

Accuracy benchmarks by interval granularity

Forecast accuracy targets vary by the time interval being measured. Daily forecasts tend to be more accurate than interval-level forecasts because random variation averages out over longer periods:

• Daily accuracy: Well-run WFM operations typically achieve 90-95% accuracy at the daily level — meaning daily volume predictions are within 5-10% of actuals on average.
• 30-minute interval accuracy: The operational unit that scheduling runs on. 80-90% accuracy at the interval level is considered good for established operations with stable volume patterns. Accuracy below 75% at the interval level will produce visible service level failures even with good scheduling execution.
• Intraday accuracy: Forecasting volumes for the current day as it unfolds, using early interval actuals to revise remaining-hour projections. This real-time adjustment capability is where modern WFM platforms differentiate from legacy tools.

Key inputs and what drives forecast error

Contact volume forecasts are built from several input streams. Historical volume data is the primary input: patterns of contact volume by day of week, time of day, and time of year (seasonality). Operations with less than 12 months of history cannot fully account for annual seasonality — holiday spikes, summer slowdowns, or fiscal quarter-end surges will not be captured in the baseline model.

Marketing and product event calendars are the second critical input. A promotional email campaign, a product launch, a price change announcement, or a service outage will generate contact volume spikes that historical patterns cannot predict. Organizations that integrate their marketing calendar directly into the WFM forecasting process — adjusting baseline forecasts with explicit volume multipliers for known events — systematically outperform those that treat contact volume as a pure historical time series.

Average handle time (AHT) forecast accuracy is a separate but related dimension. Even a perfect volume forecast will produce incorrect staffing requirements if AHT is misforecast. New product launches often temporarily elevate AHT because agents encounter unfamiliar issue types; post-system-migration periods often see AHT spikes as agents adapt to new tooling. WFM teams must forecast both volume and AHT to produce accurate staffing requirement calculations.

Forecast accuracy and downstream metrics

Forecast accuracy connects directly to schedule adherence as a paired metric: even perfect adherence cannot compensate for a systematically inaccurate forecast. If the forecast under-predicts volume by 20%, agents can adhere perfectly to their schedules and the operation will still be understaffed by 20% during those intervals. Service level failures attributed to "adherence problems" are sometimes actually forecast accuracy problems that only become visible at the scheduling level.

Occupancy rate is similarly affected: systematic forecast underestimation leads to understaffing, which drives occupancy above target thresholds and degrades agent quality and retention. Shrinkage forecasting accuracy is a component of the overall staffing requirement calculation — if shrinkage is underestimated, net available agent time will be lower than forecast even when gross staffing is correct.