EWR Ancillary Revenue Soars 3.8x ROAS in 2026

Airlines are always on the hunt for revenue beyond the ticket price, and as Statista reports show, ancillary revenue is a huge and growing pot. For us, the focus zeroed in on Newark Liberty International Airport (EWR) in late 2025, where maximizing that non-ticket income became a top priority. Our goal was to run a predictive modeling campaign to sell more ancillary services, think seat upgrades, extra bags, and in-flight perks for EWR departures. We wanted to use our data to get ahead of what passengers might need and hit them with a tailored offer at the right time, turning what’s often an afterthought into a real profit machine.

Campaign Strategy: Anticipating Passenger Needs

Our whole strategy was built around a predictive model that did one thing: find the passengers most likely to buy ancillary services. We went straight for granular behavioral analysis instead of just relying on broad demographic targeting. The hypothesis was that we could predict who would buy what based on their booking patterns, flight details, and even what they were doing at the airport in real time. For example, we flagged a traveler on a basic economy ticket with a tight connection as a perfect candidate for an offer on an expedited boarding pass or an extra bag allowance.

We ran the campaign for 12 weeks straight, from October 2025 to January 2026, to capture the peak holiday travel season. With a total budget of $250,000 spread across our digital channels, the core of the work was integrating our predictive model directly with real-time ad platforms. This let us put super-relevant ads right in front of people at the perfect moment, whether that was in a post-booking email or on the Wi-Fi login page at EWR’s Terminals A, B, and C.

Data Integration and Predictive Modeling

The model was hungry, so we fed it a ton of data. We pulled in historical booking info (fare class, destination, how far in advance they booked), loyalty status, past ancillary purchase history, and even anonymized data on travel patterns from specific EWR gates. We found, for instance, that flights out of Gate 62 in Terminal C, which are often international, had a high historical purchase rate for premium lounge access. We also piped in real-time operational data like flight delays and gate changes to trigger offers. A two-hour delay is the perfect time to push a lounge access or meal voucher offer.

Under the hood, the model used a mix of logistic regression and gradient boosting algorithms to calculate a propensity score for each product for every passenger. A high score meant a high probability of purchase, which let us create audience segments on the fly and tailor the ad delivery. Every decision was data-backed. In this market, you can’t afford to operate any other way.

Creative Approach: Contextual Relevance is King

Our creative had to be just as smart as the model. Generic ads were a non-starter. We built a set of dynamic creative templates that our system could populate with specific offers based on what the model predicted a passenger wanted. If someone was flagged as needing extra baggage, the ad would highlight the convenience of pre-purchasing a bag and show the exact price for their specific flight. If they were a candidate for a seat upgrade, they’d see images of extra legroom and priority boarding.

We tested a bunch of different calls-to-action (CTAs) and value props. At first, we led with price discounts, but we quickly found that messaging around convenience and comfort resonated much better, especially for offers that were time-sensitive. “Skip the line, relax before your flight” absolutely crushed “20% off lounge access.” All the creative was built mobile-first, clean, and easy to read. We obsessed over the user experience because it has a direct, measurable impact on conversion rates.

Targeting and Channel Mix

Our targeting was layered. The propensity score from the model was the primary filter, but we added contextual and behavioral signals on top of that. Our channel mix was pretty standard at first:

• Email Marketing: Post-booking and pre-departure emails triggered by the model. These got a solid 28% open rate and a 5.5% CTR.
• In-App Notifications: We used push notifications for app users, triggered by their location (like getting close to the EWR security line) and predicted need.
• Paid Social Media (Meta & LinkedIn): We retargeted people who had booked EWR flights but hadn’t bought ancillaries yet, using custom audiences from our booking data. The CTR here averaged 1.8%.
• Programmatic Display Advertising: Banner ads on travel sites, served to our model-defined segments. We ran on Google Display Network and some private exchanges, hitting about 15 million impressions.
• Airport Wi-Fi Portals: This was a great channel. We ran targeted banner ads on the EWR Wi-Fi splash pages, which was perfect for last-minute upgrades and lounge access.

Our initial budget split was 40% to email, 25% to social, 20% to display, and 15% to in-app/Wi-Fi. That didn’t last long. We saw how well the in-app and Wi-Fi channels were converting and quickly started shifting more budget there.

What Worked and What Didn’t

What Worked:

• Real-time Triggering: The stuff that hit people at the airport, especially when tied to flight status, had way higher conversion rates. A passenger who just found out their flight is delayed two hours is much more likely to buy lounge access than someone getting the same offer three days earlier. Once we saw that, we immediately shifted more of the budget into real-time advertising at the airport.
• Personalized Pricing: On some products, like premium seat upgrades, we could dynamically adjust the price based on demand and passenger segment (like their loyalty status). It didn’t work for everything, but it was a nice lift where we used it.
• Bundling: Offering small bundles like “priority boarding + extra legroom” usually beat single-item offers and increased our average transaction value by 7%.
• Benefit-Focused Creative: We quickly learned that creative focusing on benefits trounced creative focusing on features. “Travel stress-free with guaranteed space for your souvenirs” worked way better than “Buy 15kg extra baggage.”

What Didn’t Work:

• Aggressive Retargeting: In the beginning, we hit people a bit too hard and saw ad fatigue set in. We had to implement frequency caps fast (max 3 ads per user per day for any single product).
• Generic Offers: Our early tests with broad discounts that weren’t tied to the predictive model were a complete waste of money. It just confirmed we needed the model.
• Long Purchase Funnels: If an offer took more than two clicks to buy, the drop-off was huge. We had to make the purchase process dead simple, ideally right from the ad or landing page.

Optimization Steps and Results

During the 12-week campaign, we were constantly running A/B tests on creative, targeting, and channel spend. Our main optimization efforts were:

1. Refining the Model: We kept feeding the model new data, like historical weather at destinations and average connection times on certain EWR routes. That one change pushed our prediction accuracy up by 12%.
2. Dynamic Creative Optimization (DCO): We used DCO platforms to automatically serve the best-performing creative to the right audience segments. The result was an 18% lift in CTR across our display and social ads.
3. Shifting Budget: Based on daily ROAS numbers, we moved money to where it was working best. In the last month, we moved 20% of the budget out of programmatic display and into in-app and airport Wi-Fi ads because they were so much more efficient.
4. Landing Page Tweaks: We tested different landing page layouts to reduce friction. Just by simplifying the checkout to be mobile-first, we got a 9% bump in the conversion rate.

Performance Metrics

The final numbers showed the campaign was a major success:

• Total Ancillary Revenue Generated: $950,000
• Total Ad Spend: $250,000
• Return on Ad Spend (ROAS): 3.8x (For every $1 we spent, we made $3.80 in ancillary revenue.)
• Cost Per Lead (CPL): $1.20 (A lead was a click on an offer that went to the purchase page.)
• Click-Through Rate (CTR): 2.1% average across all channels.
• Impressions: Over 50 million total.
• Conversions (Ancillary Product Purchases): 7.2% of the people we targeted bought something.
• Cost Per Conversion: $3.47

Numbers like these just show how effective data-driven marketing can be, even in a crazy environment like air travel. The predictive model took the guesswork out of the equation and added directly to the bottom line. It just goes to show that putting money into real analytics for customer engagement actually pays off.

One of the big lessons was that even the best models need human oversight. The tech did the heavy lifting, but our team’s daily performance reviews were essential for spotting weird trends and making strategic calls that an algorithm would miss. For example, the model didn’t initially catch a sudden spike in baggage purchases for Caribbean flights during a cold snap at EWR, but a human analyst saw it and we jumped on the opportunity.

This campaign’s success was also a direct result of getting our marketing tech stack talking to the airline’s operational systems. Without real-time access to flight data and passenger profiles, the model would have been a shadow of itself. The airlines that actually break down their data silos are the ones who will win in this space. This kind of smart, integrated work is the only way forward for ancillary revenue growth at a hub like EWR.

Going forward, there’s still a ton of growth potential here, but we’ll only get it by constantly refining our predictive models. We’re planning to feed it more granular data, like a user’s browsing history on the site, and test new ancillary products like carbon offsets or expedited security access within the same predictive framework.

The success at EWR wasn’t a fluke. It’s a scalable playbook. We’re confident we can get similar results at other major hubs by applying the same predictive methods, just tailored to that airport’s specific passenger flows. When you can anticipate what a passenger needs and offer them the right thing at the right time, it completely changes the game for airline commercial strategy.

Our goal was selling smarter, not just selling more. Using passenger journey data to predict what someone wants turns a potentially annoying upsell into a genuinely helpful service. It builds goodwill and drives revenue at the same time. That’s a win-win, and it’s what modern marketing should be doing.

The data makes a clear case: investing in predictive analytics for ancillary revenue pays for itself. Airlines at a busy hub like EWR are sitting on a goldmine of data and can’t afford to let it go to waste. Turning that raw data into concrete insights that generate revenue is what separates you from the competition.

We learned that the tech is a powerful tool, but a person has to be at the wheel. Getting that balance right is how you win in a complicated industry like this.

If an airline wants to maximize ancillary revenue, it has to start with solid data infrastructure and a real commitment to learning and adapting on the fly. This EWR campaign shows exactly what happens when you get those pieces to work together.

Running a predictive modeling campaign for ancillary revenue at a hub like EWR means you have to get data science, creative, and fast optimization all working in concert to make the most of every single passenger touchpoint.