Two years ago airline media networks were just emerging. Now it's a category. United is in its third year, SAS launched in June with Kevel, Virgin Australia in September, and Lufthansa folded three carriers' inventory into one media business the same month. With the global market for in-flight retail and advertising projected to reach $5.9 Billion by 2032, the question has moved from whether to enter the market to how.
The how is where it gets interesting. Every airline spinning up their own media network is being offered a playbook. The retail media side brings the paybook that built grocery and marketplace networks. The travel media folks bring the playbook that built booking sites. And while both work for those specific verticals, neither goes far enough to make it the most compelling for an airline.
Airline media is almost an entirely different business. The easiest way to see it is to follow a trip: a family of four, booked in March, flying to Orlando for nine days in July.
Four things few other networks have
Airlines see the entire itinerary. A grocer sees a basket. A booking site sees a search. A card sees a transaction. The airline has the whole record in March (where, when, how long, which cabin, who else) and is still there in July, at the lounge and in the seatback. The record and the presence, in one place.
They know their travelers at every phase. Retail infers intent from the cart. Travel infers it from the search. The airline meets its customer at the first search and reads the rest off the reservation: corporate code or leisure fare, one seat or four, first or basic economy, three days or three weeks, and so much more. That is a persona and a plan, weeks before anyone has searched or booked a hotel.
One identity, a journey of surfaces. A retailer owns its shelf. A booking site owns a website. An airline serves the app, the site, email, lounge screens, the gate, the seatback and the Wi-Fi portal, and most of the hardware on board is operated by a partner with its own constraints on content and tracking. When United launched, its loyalty chief said the quiet part out loud: anything that goes onto an aircraft is security-screened, media files included, so they had to build a bridge. Connectivity is arriving fleet by fleet through the end of the decade; one large European group has scheduled completion for 2029, and one of Europe's biggest carriers has declined it. Identity, meanwhile, is solved. Free Wi-Fi for a loyalty login is now the standard move, and it turns an anonymous seat into an authenticated one. Identity is the easy part. The hard part is everything after.
Rich endemic & non-endemic opportunities. Every other network sells one slice of the trip and competes with the rest of it. A hotel won't sell a rival. A car rental company won't promote the train or ride share service. A retailer sells the brands on its own shelves. The airline is the one network in the trip that has an opportunity across a broader ecosystem. No retail or travel network has the same endemic and non-endemic opportunity. SAS is selling both today: hotels, tourism boards and car rental companies on one side, lifestyle brands on the other.
The two playbooks don't go far enough
Retail media was built around a SKU, a shelf and a checkout: sponsored products, onsite search, in-store screens, and a closed loop with one question: did the exposure become a sale? It's the most successful ad model of the decade, and airlines share only some of it. An airline has a menu and a cart on board, and a sponsored spot at the top of either is real money. But there's a lot more to it than sponsored products and baskets. The cabin isn't just a checkout.
Travel media was built around a moment. Booking sites own the search, then lose the traveler, and increasingly sell the intent signal through someone else's DSP; one announcement this June bundled four booking brands, a hotel group, a rideshare company and an airline into a single programmatic buy. Rideshare sells a short session. Hotels sell the property. The inflight media business airlines already run sells reach by the month, with insertion orders due ten to twelve weeks out and results reported as contacts and share of voice. Followed literally, the cabin becomes an awareness buy at a rideshare CPM, and the richest first-party data in travel sits next to a rate card that never touches it. A nine-day July trip booked in March appears nowhere on that card.
Both playbooks get you to a sponsored placement and a price list. Neither gets you to one business across the whole trip, with one identity, two advertiser classes and several surfaces that are difficult to serve into, let alone run from one decision engine. And both were built for the auction: sponsored products and programmatic, priced by the bid. An airline's endemic and non-endemic buyers want sponsorships, flat fees, guaranteed share of voice and fixed CPMs as often as they want an auction, and a network that can sell only one of those leaves the rest on the table. That gap is the actual work.
Then what does?
Selling both sides of a trip under one identity is an engineering problem before it's a sales problem. Five things have to be true of the stack, and each comes with a question for any vendor. The test for all of them is the same: does it hold from March to July, end to end?
One unified decision engine for every surface. App, Wi-Fi portal, lounge screen and the partner's seatback all call the same decisioning layer, or take a cached manifest from it. Not one ad server per environment, reconciled afterward in a report. Ask: does the seatback call the same endpoint as the app?
Built for every aircraft in the fleet. The network is sold against the whole fleet, so it has to work where there's no signal for four hours: decisions cached before pushback, impressions logged when the link comes back, nothing that needs a live call to show an ad. Ask: what happens to the campaign on a plane with no connectivity?
Endemic and non-endemic. Hotels, cars and tourism boards belong pre-trip, when the trip is ticketed and nothing else is booked. Luxury, financial services and auto belong in the lounge and the cabin. Two inventories, two price books, and every pricing and bidding type in each (sponsorship, flat fee, fixed CPM, auction), not one pool at a blended rate. Ask: can you customize your underlying platform to waterfall across multiple pricing, buying, selection, and advertiser types for an entire travel journey?
Waterfalls and relevancy across three kinds of demand. The airline's own promotions, its marketing partners (the co-brand card, alliance carriers, the sponsor on the Wi-Fi login) and its paid advertisers all want the same slot. Priority tiers and relevancy scoring decide who wins, and house campaigns carry a notional price so paid demand can outbid them when it should. Ask: can an upgrade, a card partner and a paid advertiser compete for one impression, and can I see why an ad won?
Forecasting, targeting, reporting & measurement in one place. Forecast by route, cabin and season before the package is sold. Report across the journey, not per surface. Cap frequency everywhere, so nobody sees the same rental car ad at booking, at the gate and at cruising altitude. Measure against the airline's own outcomes (ancillary attach, partner bookings, loyalty engagement), not metrics borrowed from a grocery aisle. Ask: can you forecast a route-and-cabin package before I sell it, and cap frequency across every surface?
How Kevel’s rich history helps us solve the toughest challenges in airline media
Kevel has spent sixteen years solving for difficult ad serving environments one at a time: cinema screens, airport screens, other digital-out-of-home, browsers where latency was measured in revenue, inflight entertainment providers, and the retailers, marketplaces, delivery apps and payments companies that wanted to own a media business rather than rent one. Airlines are all of them at once.
With more than 25 million passengers a year and 135 destinations, SAS Airlines is able to sell endemic and non-endemic advertising, across their variety of surfaces, with full control of its own data, all built on Kevel.
Another great example is TouchTunes, which runs digital jukeboxes in tens of thousands of bars. Its hardware calls a proxy, the proxy pulls decisions from Kevel, caches them, and sends the right ad to the right jukebox when it's needed, direct-sold and programmatic, from one ad server. A jukebox in a bar with bad Wi-Fi is just like a seatback with a different logo.
A class of its own needs a stack built for every class at once. Airline media starts before the trip and knows where it ends. The offer that found the family in March is the one that knows they're landing in July. From app to altitude, and back.
