Car park revenue per passenger has been falling at most major airports for several years. The top-line numbers may have recovered post-2020, but unit economics have not. Something has been eroding the per-traveller value of the parking line, and most commercial teams know it intuitively, but cannot describe it with the precision required to act.

The intuition has names: ride-hailing, public transport improvements, off-airport parking aggregators, changing booking patterns, shifting traveller demographics. The honest answer is that all five are happening at once, in different ratios at different airports. Which means the question “why are we losing parking revenue?” has no single answer - only a portfolio of answers, each requiring different commercial response.

This is a diagnostic, not a prescription. Where has the revenue gone? Who is the customer who used to drive and park, and what are they doing instead? The data trail to answer those questions is already in the airport's own systems. It just isn't being looked at the right way. This post is about airport car park customer switching - what it looks like in the data, and what AI changes about your ability to see it.

The structural shift hiding inside the headline numbers

Headline non-aeronautical revenue numbers are rising globally - but the mix is changing underneath. ACI World data shows car parking has increased its share of non-aeronautical revenue from 21% to 24% in recent years, even as concessions revenue share has fallen. Parking is strategically more important to airport finances than it was five years ago, not less.

Underneath the rising importance, however, sits structural volatility. Parking revenue is significantly more traffic-elastic than aeronautical revenue. Industry analysis of North American airport financial reports shows parking revenue fell 32% in the COVID impact year - compared with an 11% fall for aeronautical revenue and a 40% drop in passenger numbers. It has recovered more slowly. As of early 2026, large hubs sit at approximately 95% of 2019 parking revenue levels, while medium hubs remain near 85%. Even as passenger numbers have caught up, the parking line has not.

The conclusion that follows: a recovering passenger line does not automatically mean a recovering parking line. Something is intercepting the customer between booking the flight and arriving at the airport. The data trail to identify those interception points is the substance of what follows.

The rise of rideshare and what it actually costs you

The most visible structural pressure is ride-hailing. Los Angeles International Airport raised its per-vehicle TNC pickup and drop-off fee from $4 to $6 in March 2026 - a structural acknowledgement that ride-hailing is now competitive ground transport, not a side stream. Other major hubs are following with similar fee adjustments, and the European trajectory is broadly the same with some local variation.

The customer-behaviour shift is asymmetric. A meaningful fraction of business travellers and short-stay leisure travellers who would historically have parked have switched to rideshare. The economics for the customer depend on trip length, party size and home distance from the airport, but for short-haul, single-traveller business trips inside a 48-hour window, rideshare often wins on total cost. The same customer who would have parked for three days at €18 a day (€54 of parking revenue) now pays a fixed pickup/drop-off fee that yields a fraction of that figure to the airport.

The migration is real and the margins are not equivalent. Lost parking revenue is partially recovered as TNC fees, but not at the same yield. The deeper commercial problem is not the migration itself - it is that most airports cannot quantify it. They see the parking line move and the TNC line move, but they cannot tell you which specific customer segments are switching, in which booking-horizon windows, and at what price-sensitivity thresholds. That is the visibility gap AI can close.

The five switching signatures the data can surface

Five distinct patterns of customer churn the data trail can identify when it is brought together properly. Each signature has a different underlying cause - and a different commercial response.

Signature 1 - The “almost-booked” customer. Started a parking booking on the airport website, abandoned mid-funnel, then completed a TNC booking within the next 48 hours. Visible only when the parking booking funnel data is joined to downstream ground-transportation telemetry - most airports have both, but rarely look at them together.

Signature 2 - The price-comparison defector. Booked a competitor - an off-airport parking aggregator, a hotel park-and-fly bundle, an off-airport long-stay lot - after the airport rate appeared in a price comparison engine. The shift is now structurally large. Digital pre-bookings accounted for approximately 54% of all airport parking reservations globally in 2025, up from 31% in 2019, with aggregators like ParkVia, ParkCloud and SpotHero materially boosting visibility of off-airport alternatives. This is the segment dynamic pricing was built for.

Signature 3 - The mode-shifter. Customer-account history shows three or four parking bookings followed by a switch to rideshare or public transport. Often correlated with a public transport improvement (a new rail link, a more frequent shuttle), or a TNC promotional period in the airport's catchment. Highest-value behavioural signal in the dataset - these customers had a relationship with the airport, and they cancelled it.

Signature 4 - The shortened-stay traveller. Same customer, same destination, same flight pattern - but a parking booking that is hours shorter than the equivalent trip a year earlier. Caused by changing arrival behaviour at the airport (faster check-in, more efficient security flow once the EU Entry/Exit System rollout stabilised, more frequent off-airport rental car returns), and it directly affects yield per booking even when booking volumes hold.

Signature 5 - The leisure substitution. Family travel that used to park (peak summer, two-week stays) increasingly going by family-member drop-off or by ride-share-plus-luggage. The most price-sensitive segment and the largest segment by volume. Losing share here moves the P&L visibly, because the displacement is concentrated in the same peak weeks that drive annual yield.

The framework matters because each signature requires a different response. Price-comparison defectors respond to dynamic pricing and rate parity. Mode-shifters respond to retention products - loyalty bundles, frequent-parker tiers, integrated mobility passes. Shortened-stay travellers respond to off-peak pricing of short windows. Leisure substituters respond to family-rate bundles and product redesign. You cannot design the right response if you cannot see which pattern is causing the loss.

Why this analysis is hard without AI

The data exists. Booking funnel telemetry exists, customer-account history exists, referral data exists, TNC fee transaction data exists, public transport mobility data exists. The problem is that these datasets live in different systems - the booking engine, the CRM, the analytics warehouse, the ground transportation revenue feed, the third-party comparison engine, the loyalty platform - and they have never been joined up to answer this specific question.

A commercial analyst working in Excel cannot stitch this together at the cadence the market requires. By the time the quarterly board paper is written, the switching pattern is three months old and the customer has habituated to the alternative. The AI shift here is not “AI sets the price.” It is “AI continuously joins the operational data and surfaces the patterns the commercial team needs to see this week, not next quarter.” Smart parking technologies and AI-powered dynamic pricing are now deployed at over 600 airports globally, improving revenue per bay by 12–18% - but the revenue gains are downstream of the analytical visibility, not separate from it.

This is the dataset airports have always had and have never used.

What good looks like when you can see the data

Three commercial outcomes follow once switching signatures are visible at the cadence operations require:

• You can retain.Targeted retention offers to customers showing early defection signals - a loyalty bundle to the mode-shifter, a price-matched product to the price-comparison defector - recover revenue at a fraction of acquisition cost. Retention economics in parking are not yet well-understood at most airports because the data has not been joined up to make them visible.

• You can grow. Identifying which competitor - off-airport aggregator, hotel partner, TNC, public transport - is winning which customer segment tells you where to compete on product, not just on price. The competitive landscape becomes a map rather than an anecdote.

• You can plan capacity. Capital decisions about parking capacity (new structure, conversion to drop-off zone, EV charging investment, premium product expansion) need a forward demand curve, not a rear-view one. The switching data is what makes the forward curve credible to the finance committee.

Closing

The conversation about airport parking revenue has been dominated by what AI can do for pricing. The more important conversation is what AI can do for visibility - surfacing the switching patterns that have been eroding the line for years but have never been measurable in time to act on. Airports that can see their customers leaving can do something about it. Airports that cannot, cannot.

If you are working out where parking revenue has gone at your airport and want to look at the data architecture that makes switching signatures visible, the VE3 team would welcome a 30-minute conversation.