WFS improves air cargo forecasting tool

Worldwide Flight Services's (WFS) has improved its machine learning tool to better forecast air cargo volumes and support better strategy development for services.

The Performance Management Platform - Machine Learning Forecast (PMP MLF) uses machine learning algorithms trained on 10 years of operational data to deliver forecasts of cargo volumes by flight, truck, and day, providing each warehouse with precise data to align workforce and resources in advance.

The tool helps WFS to accurately forecast volumes using intelligence based on the processing of over 3m air waybills and historical flight and truck movement records, incorporating seasonality, holidays, and cargo types.

Currently providing forecasts across 9,842 flights and 6,216 truck movements per week across 75 warehouses in 13 countries, the system produces daily forecasts of tonnage, ULDs and piece count, broken down by transport mode (freighter, passenger, and Road Feeder Services), flight or truck number, customer, and warehouse location.

These forecasts feed directly into station-level planning tools, giving every location clear and reliable forward-looking data.

Using the PMP MLF tool, WFS can detect and preplan for volume surges early and adjust resources proactively, shifting labour between teams or sites with greater agility. This reduces Service Level Agreement breaches due to understaffing or overloading and avoids unnecessary overtime or idle time.

This summer saw the roll out of phase 2 of the tool, with further digital improvements, including enhanced dashboards and visual analytics; tighter integration with workforce management and rostering tools; and customer-level forecasting to co-plan volume peaks.

Data collected shows the tool has a 92-98% accuracy range, even during irregular demand periods, said WFS.

The air cargo industry has long struggled with accurate forecasting due to volatile volumes. Labour planning has often relied on manual estimations and historical averages, which can result in a 10-15% gap between staffing levels and actual workload, causing inefficiencies, reactive operations, and inconsistent service quality.

Jimi Daniel Hansen, senior vice president operational excellence, said: “For many years, cargo handlers have relied on manual scheduling, Excel spreadsheets, or basic rolling averages for forecasting – and we know some still do.

"By leveraging machine learning within a complex operational network, our goal was to replace reactive guesswork with data-driven clarity to optimise workforce allocation, enhance service levels, and reduce operational waste across our global air cargo network – and we are inspired by the results.

"Predictive planning and precision forecasting means we have achieved a fundamental transformation in how cargo handlers plan and operate."

He added: “All of these benefits are meaningful to our customers. They translate into fewer delays due to staffing issues, improved service consistency, and transparent, data-backed capacity shared in advance. This is the type of digital innovation they want to see."