Why capacity forecasting is the missing piece

You can't optimize what you can't predict

Revenue management gets discussed as a pricing discipline. But before you can set the right price, you need to answer a more fundamental question.

How much space do you actually have to sell?

For most carriers, that question is harder than it should be. And the answer is often wrong in one direction or the other.

Two costly ways to be wrong

If you overestimate available capacity, you sell space you cannot deliver. Shipments are offloaded, customers receive calls they did not expect, forwarders remember the disruption at contract renewal, and operations absorbs the consequences. 

If you underestimate capacity, you leave space unsold on a flight that departs with room. That revenue does not return. It is rarely recorded as a visible loss, which is exactly why it persists.

Neither outcome is unusual. Both happen regularly, and the second is far less visible.

Why belly capacity is difficult to estimate

On a widebody passenger aircraft, available cargo capacity on any given day depends on passenger load, checked baggage weight, fuel uplift for the route and expected winds, aircraft configuration, ULD availability and position count, and operational restrictions on the day.

Several of those variables are not known until close to departure. As a result, many carriers price against a planning figure: A seasonal average, a rolling historical mean, or an experienced planner's estimate. And then discover the real number later.

The result is a recurring gap between the capacity you priced against and the capacity you actually had.

The number only few carriers forecasts: What actually gets tendered

This is the forecasting challenge the industry still tends to understate.

Booked weight is not the same as tendered weight. Shipments arrive light, arrive heavy, arrive at a different density than declared, arrive late, or do not arrive at all. Depending on lane, customer and season, the gap between what was booked and what actually showed up can be material. And it varies predictably by customer, which means it is forecastable.

Many carriers manage this with a flat overbooking factor applied broadly. That is a blunt response to a problem with clear structure. A customer with a consistent tendered-to-booked profile behaves differently from one that routinely books ten tonnes and tenders six, and a uniform factor treats them identically. Closing that gap means forecasting the shortfall itself: Only then can overbooking move beyond a uniform load factor and become calibrated to actual tender behaviour. 

Forecasting tender behaviour, not just available capacity, is where cargo revenue management is heading rather than where it stands today: Sold volume is the number the discipline has learned to model, while forecasting what physically gets tendered against those bookings remains the next frontier. And a meaningful share of unbooked upside may sit behind it. 

There is another distinction that abstract discussions of capacity often flatten.

A flight can be volume-constrained and weight-available, or the reverse. A 0.4 cbm/kg shipment and a 0.15 cbm/kg shipment consume fundamentally different resources, and yield per kilo alone does not tell you which shipment should have been accepted.

If your capacity forecast is a single weight figure, you are pricing against only part of the constraint. Density-aware forecasting changes acceptance decisions, not just rate levels.

From estimates to intelligence

The variables driving cargo capacity are numerous, interdependent, and in most cases already recorded somewhere in your operational history. That is well suited to a machine learning problem.

Models trained on years of actual operating data can learn what a planner learns over a career and apply that learning consistently across every flight, every day. For example, that a specific routing runs heavy on fuel in November, that baggage load on a lane spikes around particular holidays, or that an aircraft’s practical usable position count differs from its published one.   

What better forecasts change in practice

Forecast accuracy is not an operational nicety. It is an input into every downstream commercial decision; and a flawed input can make even a sophisticated pricing strategy confidently wrong.

When teams trust the forward capacity picture, four things change.

1. Acceptance decisions get sharper. Turning away a shipment because a flight looks full is only the right decision if the flight is actually full.
2. Sales stops hedging. Reps can commit to space with greater confidence instead of leaving margin for error in both directions.
3. High-value space gets protected deliberately rather than accidentally.
4. Overbooking can move from a uniform factor toward a calibrated one.

Forecasting capacity, forecasting opportunity

The most advanced cargo organizations no longer treat forecasting as a planning exercise handed to commercial teams as a constraint.

Every improvement in forecast accuracy is also a pricing improvement, a sales improvement and a yield improvement, because commercial performance is built on top of it.

Forecasting is not purely about knowing how much space you have. It is also about knowing what that space is worth.