How Expedia Uses AI to Modernize Travel Marketing

Most companies get AI in marketing backwards: they start with content generation and end up with more noise. Expedia’s marketing evolution points in the opposite direction—start with the traveler’s intent, then use AI to make every message more relevant, more timely, and easier to measure.

That’s why a recent conversation with Expedia Group CMO Jochen Koedijk is worth treating as a case study for the broader U.S. digital services economy. Travel is one of the hardest categories to market: customers research across dozens of sites, prices change constantly, and “the product” is a bundle of airlines, hotels, rentals, and experiences. If AI can make marketing work here, it can make marketing work almost anywhere.

This post breaks down what “AI-powered marketing” actually looks like in a large digital platform, what you can borrow from Expedia’s approach, and where teams tend to slip—especially as we head into 2026 planning (and the post-holiday booking surge that follows late December).

Why AI matters more in travel than in most industries

AI matters in travel marketing because traveler intent changes fast and the inventory is volatile. The best marketing system isn’t the one that shouts the loudest—it’s the one that updates its understanding of what a specific customer is trying to do and responds quickly.

Travel platforms operate in a high-frequency decision loop:

• A traveler searches, compares, abandons, returns, and books across days (sometimes weeks).
• Prices and availability shift by the hour.
• A “trip” has multiple components, each with its own constraints.

In practice, this makes AI marketing automation less about writing copy and more about decisioning: who should see what, when, and why.

The real shift: from campaigns to systems

Traditional marketing planning is campaign-centric: pick a destination, pick a promo, run it for a few weeks. Expedia’s evolution reflects a more modern approach: build an always-on marketing system that can adapt to demand.

That system typically includes:

• Predictive models (propensity to book, cancel, upgrade, or churn)
• Personalization engines (ranked recommendations, next-best action)
• Creative testing pipelines (rapid experimentation at scale)
• Measurement frameworks (incrementality, not just last-click)

This is the model many U.S. digital services companies are moving toward: AI isn’t a side project; it’s the operating layer underneath growth.

What Expedia’s “AI-powered marketing evolution” looks like in practice

AI-powered marketing at Expedia is best understood as three connected moves: better signals, better decisions, better experiences.

1) Better signals: turning travel intent into usable data

The first step is collecting and interpreting intent signals—search queries, dates, party size, destination flexibility, device type, loyalty status, and prior trips.

In travel, “intent” is rarely a single action. It’s a sequence. AI helps by:

• Identifying patterns that correlate with booking (for example, repeated searches for similar dates)
• Detecting when someone is still exploring vs. ready to buy
• Distinguishing leisure trips from business trips based on behavior

If you’re running a SaaS or digital service, the equivalent is product telemetry: the in-app behaviors that predict conversion and retention. Expedia’s lesson is simple: AI is only as good as the signals you prioritize and clean.

2) Better decisions: next-best message, channel, and timing

Once you’ve got signals, the marketing problem becomes decisioning. For a travel platform, that can mean:

• Which destination or property to feature
• Whether to emphasize price, flexibility, loyalty points, or convenience
• Whether to message via email, app push, paid search, social, or onsite personalization

This is where teams often over-rotate into “AI writes the ad.” The higher-value move is AI decides the context for the ad.

A practical decisioning stack usually includes:

• Propensity scoring (likelihood to book within a window)
• Dynamic audience building (real-time segments, not static lists)
• Frequency and fatigue controls (protecting the customer experience)
• Suppression logic (don’t market a hotel to someone who already booked it)

Snippet-worthy truth: Personalization isn’t inserting a first name—it’s choosing the right problem to solve for that person right now.

3) Better experiences: marketing that feels like service

For large U.S. digital platforms, the best marketing increasingly looks like customer service:

• “Your price dropped.”
• “Rooms are filling up for those dates.”
• “Free cancellation options match your preferences.”

That’s not just messaging polish. It’s a product-and-marketing collaboration where AI helps translate complex inventory into a simple choice.

In Expedia’s world, that means marketing isn’t merely acquisition—it’s trip enablement across the full journey:

• Inspiration (where to go)
• Planning (what to book, in what order)
• Booking (confidence and conversion)
• Post-booking (changes, upgrades, support)

For the broader “How AI Is Powering Technology and Digital Services in the United States” series, this is a key theme: AI-powered customer communication scales when it’s anchored to real customer tasks.

The marketing KPIs that AI actually improves (and the ones it can distort)

AI tends to improve metrics tied to relevance and efficiency—but it can also inflate the wrong numbers if you measure poorly.

Metrics AI can improve when implemented well

• Conversion rate (CVR): Better matching between intent and offer
• Cost per acquisition (CPA): Less wasted spend through smarter targeting
• Return on ad spend (ROAS): Higher yield from the same budget
• Repeat purchase rate / loyalty engagement: Better lifecycle messaging
• Customer satisfaction signals: Fewer irrelevant messages and smoother trips

Metrics that can mislead you

• Last-click attribution: AI can optimize to “easy wins” rather than incremental growth
• Click-through rate (CTR): High CTR can still mean low-quality traffic
• Volume of content produced: More assets doesn’t mean better performance

If you want a sturdier approach, aim for incrementality: how much did this AI-driven personalization actually cause, compared to a holdout group?

A stance worth taking: If you can’t run holdouts, you don’t really know if your AI marketing is working—you only know it’s busy.

What U.S. digital services teams can copy from Expedia’s approach

You don’t need Expedia’s scale to adopt Expedia-like principles. You need clarity on decisions, data, and governance.

Start with one “high-intent” moment

Pick a moment where the customer is already close to value. In travel that might be “abandoned search” or “abandoned cart.” In other digital services, it could be:

• Trial user hits a usage threshold but doesn’t convert
• Customer nears renewal but engagement is down
• User searches a help topic that predicts churn

Build an AI workflow around that moment:

1. Define the intent signal
2. Decide the best next action
3. Choose the channel
4. Test against a holdout

Treat creative as modular, not precious

Expedia has to mix and match destination, price, dates, imagery, and value props. That forces a modular creative mindset—perfect for AI-driven experimentation.

A practical framework:

• Core message modules: price, flexibility, loyalty, convenience
• Proof modules: ratings, limited availability, member perks
• CTA modules: book now, hold price, continue planning

AI can then help identify which combinations work for which segments.

Build governance before you scale automation

AI marketing at scale breaks when teams don’t set guardrails. Things I’ve found work:

• A clear policy on personalization sensitivity (health, finances, location granularity)
• Brand safety reviews for generated copy and images
• Frequency caps by channel and by lifecycle stage
• “Customer-first” suppressions (don’t upsell during a disruption)

For travel specifically, this matters because disruptions happen. A traveler dealing with a cancellation doesn’t want cheery promotional nudges.

People also ask: practical questions about AI in travel marketing

Does AI replace marketers at companies like Expedia?

No. It changes the work. Marketers spend less time manually segmenting lists and more time designing decision frameworks, creative systems, experiments, and measurement.

What kind of AI is most useful for digital marketing platforms?

Predictive models and ranking systems typically create more value than text generation alone. Generative AI becomes more useful when it’s connected to real-time decisioning and a strong testing pipeline.

What’s the fastest way to prove ROI from AI marketing?

Pick one high-intent workflow (like abandoned search), run an A/B test with a holdout, and measure incremental bookings or revenue—not clicks.

Where this is heading in 2026: AI as the operating system for growth

Late December is when many teams finalize Q1 plans, and travel marketing is already gearing up for the post-holiday booking wave. The direction is clear across U.S. digital services: AI becomes the operating system for how growth teams decide, not just a tool for producing assets.

Expedia’s marketing evolution highlights a practical path: focus on intent signals, build decisioning that respects the customer, and measure incrementality. Do those three things and you’ll avoid the most common failure mode—automation that scales irrelevance.

If you’re building AI-powered marketing in your own digital service, start small but design for scale: pick one intent moment, connect it to a next-best action, and put measurement guardrails in place. Then expand.

Where could your customer communication feel more like a helpful service and less like a campaign?