Home / Case Studies / Corporate Traveller Segmentation
Aviation

Predictive Modelling Uncovers Millions in Hidden Corporate Revenue

An airline knew they had contracted corporate customers — but suspected a significant portion of business travellers were booking as regular passengers, invisible to their commercial team. We built a predictive model to find them.

Predictive Analytics Revenue Growth Customer Segmentation Data Science

The Challenge

This airline had a well-established corporate travel program — companies registered with an ABN, negotiated contracted rates, and their employees booked through managed channels. That segment was visible, tracked, and commercially managed. But there was a blind spot:

  • A significant volume of passengers exhibited corporate travel patterns — frequent weekday flights, short-stay returns, premium fare classes, consistent routes — but booked as regular individual travellers
  • These travellers had never registered for a corporate deal. They weren't linked to an ABN. To the airline's systems, they looked identical to leisure passengers
  • The commercial team had no systematic way to identify these hidden corporate travellers, quantify the segment, or target them for conversion to contracted business accounts
  • Without identification, these high-value passengers received generic marketing, no loyalty incentives tailored to business travel, and no outreach from the corporate sales team
  • The airline was leaving significant revenue on the table — both from the travellers themselves (who would likely fly more frequently under a corporate deal) and from the companies behind them (who may have been splitting bookings across multiple carriers)

Our Approach

We built a classification model using predictive modelling and regression analysis to score every non-contracted passenger on their likelihood of being a corporate traveller. The model doesn't guess — it identifies statistically significant behavioural patterns that distinguish business travel from leisure.

  • Feature engineering: We extracted over 30 behavioural signals from booking and flight data — day-of-week patterns, booking lead times, fare class preferences, route frequency, trip duration distributions, return flight timing, and booking channel characteristics
  • Model training: Using the airline's known contracted corporate travellers as the positive class, we trained a classification model to learn what corporate travel behaviour looks like. The model was validated against hold-out data to ensure it generalised beyond the training set
  • Regression analysis: Alongside classification, we used regression techniques to quantify revenue differences between segments — measuring how much more corporate travellers spend on ancillaries, upgrades, and rebooking flexibility compared to leisure passengers
  • Probability scoring: Every non-contracted passenger receives a corporate probability score from 0 to 1, allowing the commercial team to prioritise outreach by confidence level rather than working from hunches
  • Revenue opportunity modelling: We estimated the revenue uplift from converting identified corporate travellers to contracted accounts — accounting for increased booking frequency, ancillary spend, and reduced churn under corporate agreements

Data Science, Not Black-Box AI

This project deliberately uses interpretable statistical methods rather than opaque deep learning. The airline's commercial team needed to understand why a passenger was flagged as likely corporate — not just that they were. Here's why that matters:

  • Explainable predictions: Every classification comes with the contributing factors — "this passenger flies the same route every Tuesday, books 2-3 days in advance, and always selects flexible fares." The sales team can reference these patterns directly in outreach
  • Actionable segmentation: Regression coefficients tell the commercial team exactly which behaviours correlate most strongly with corporate travel and higher spend — informing both targeting and product design
  • Auditable methodology: The model can be reviewed, challenged, and refined by the airline's own analysts. No vendor lock-in, no proprietary algorithm they can't inspect
  • No data sent externally: All modelling was performed on the airline's own data infrastructure. Passenger data never left their environment

From Identification to Conversion

The real value isn't just knowing who the corporate travellers are — it's converting them. Once identified, the airline's commercial team can approach these passengers (or their employers) with tailored corporate deals: negotiated rates, priority booking, loyalty accelerators, and dedicated account management. A traveller who was previously invisible becomes a contracted corporate customer generating predictable, recurring revenue.

The Results

90% Classification Accuracy
18% Non-Contracted Bookings Flagged Corporate
Millions Annual Revenue Opportunity
3x Higher LTV for Converted Accounts
Booking Segmentation Breakdown 18% 82% General ■ Identified Corporate ■ General Bookings

The model revealed that nearly one in five non-contracted passengers were exhibiting corporate travel behaviour — a segment the airline had no visibility into before this project. The revenue opportunity from converting even a fraction of these travellers to contracted corporate accounts runs into the millions annually. The commercial team now has a prioritised, scored pipeline of conversion targets instead of relying on inbound enquiries and guesswork.

"We always knew there were corporate travellers hiding in our general bookings — we just had no way to find them. Now we have a scored list and our sales team is converting accounts we never would have found manually."
— Data Analytics Manager, Airline (name withheld)

Frequently Asked Questions

How does predictive modelling identify hidden corporate travellers in airline bookings?

The model analyses over 30 behavioural signals from booking and flight data — including day-of-week travel patterns, booking lead times, fare class preferences, route frequency, and trip duration — and compares them against known contracted corporate travellers. Each non-contracted passenger receives a probability score from 0 to 1, allowing the commercial team to prioritise outreach by confidence level.

What is the business value of identifying uncontracted corporate airline passengers?

Uncontracted corporate travellers represent significant hidden revenue. Once identified, airlines can approach these passengers or their employers with tailored corporate deals — negotiated rates, loyalty accelerators, and dedicated account management — converting invisible bookings into predictable, recurring revenue streams with up to 3x higher lifetime value.

Why use interpretable statistical models instead of deep learning for customer segmentation?

Interpretable models let the commercial team understand exactly why a passenger was flagged as likely corporate — for example, weekly same-route flights with short booking lead times and flexible fares. This transparency makes predictions actionable for sales outreach, auditable by internal analysts, and free from vendor lock-in that comes with proprietary black-box algorithms.

Sitting on Data You're Not Using?

We'll show you how predictive analytics can uncover hidden revenue and sharpen your commercial strategy.

Book Your Free Assessment