Travel organizations are used to cross-functional tension. Marketing owns the story, revenue management owns the price, and digital experience owns the journey that ties them together. In a category where a single booking decision can take weeks and touch dozens of interactions, that tension can be productive. It forces teams to reconcile customer experience, demand, inventory, margin, and business priorities before something reaches the traveler.
Optimizely’s Agent Platform changes the volume of decisions that structure has to absorb. It can help teams generate and execute more ideas in less time, including:
- New pricing and packaging treatments
- Personalization rules based on traveler behavior or demand signals
- New ways to position an itinerary or destination
- Content variants for different audiences, channels, or trip types
- New experiments that previously would have taken too much time to build and test
For years, the constraint on experimentation was often how many ideas a team could develop, build, and launch. As Agent Platform reduces some of that friction, the constraint moves downstream. Teams can generate more viable ideas than their existing decision structures were designed to evaluate.
That matters especially in travel, where many of the highest-value opportunities cross organizational boundaries. A pricing treatment affects messaging. A personalization rule can influence demand for particular inventory. A new way of positioning an itinerary can change which departures or packages receive attention.
AI can generate the hypothesis quickly. The organization still has to decide whether to act on it, under what conditions, and who owns the result.
When informal ownership stops scaling
Most organizations already have ways of making these decisions. The problem is that many of them are informal: a Slack message, a mention during Monday’s standup, an email thread, or a conversation between team members.
That can work when the volume of ideas is relatively low. It becomes much harder when Agent Platform makes it possible to identify and develop viable experiments faster than teams can evaluate them.
The symptoms might look familiar:
- Prioritization depends on who happens to raise an idea at the right moment.
- Approval timelines vary based on which team encounters the idea first.
- Decisions get reopened every time an experiment crosses another team boundary.
- No one is entirely sure who has final authority when teams disagree.
- Accountability becomes just as ambiguous once an experiment is live.
The technology may be working exactly as intended. The operating model around it has simply reached its limit.
What Road Scholar changed
Road Scholar encountered this underlying ownership problem as it expanded its use of Optimizely’s Agent Platform. Requests for new agents and use cases were coming through scattered channels, including Slack, meetings, and email. Prioritization varied by team, and ownership was not always explicit.
Road Scholar established a weekly AI review, assigned named decision owners, and required end-user sign-off before an agent reached production. With that structure in place, the path from concept to a live agent could move in as little as two weeks.
The important lesson is not that every organization needs the same weekly meeting. It’s that AI moves faster when teams establish a consistent path from idea to decision to accountable owner. That path has three parts:
- AI generates the hypothesis. A personalization engine might identify an opportunity around a shoulder-season package, or a content workflow might propose new ways to position an underperforming itinerary.
- The right leaders evaluate it. The idea is assessed against business strategy, customer experience, operational realities, risk, and the metrics that matter across the teams it affects.
- A named owner makes the call and stays accountable. Once a decision is made, one person is responsible for moving it forward and understanding what happens after launch.
Without that last step, ownership tends to sit with a function rather than a person. Marketing assumes revenue will make the call. Revenue assumes marketing owns the experience. Digital gets pulled in once something needs to be built. The same decision gets revisited every time the idea crosses another team boundary.
Who actually owns the decision?
Clear ownership does not mean assigning one person to approve every AI experiment. The decision owner should change based on the type of decision being made.
A pricing decision may ultimately belong to revenue management. Messaging may sit with marketing. Technical deployment may require digital or technology ownership. Some experiments will require input from all three. The goal is to determine who needs to be consulted and who has final authority before the experiment reaches the approval stage.
That distinction matters because cross-functional input and shared decision ownership are not the same thing. Consider a personalization treatment that increases conversion but shifts demand toward inventory that does not need additional promotion. Marketing can be right about the conversion opportunity while revenue management is equally right about the inventory implications.
The team needs both perspectives. It also needs to know who makes the final call.
Where ownership gets complicated in travel
The overlap between marketing and revenue management makes some areas particularly vulnerable to unclear decision rights.
Pricing and packaging
Pricing and packaging tests can affect messaging, conversion, margin, inventory, and yield at the same time. Before an AI-generated test launches, teams need to know which metrics will determine success, who needs to provide input, and who has final authority when those metrics point in different directions.
That becomes even more important as Agent Platform makes it easier to identify and launch highly targeted experiences. Increasing the number of possible experiments also increases the number of tradeoffs someone has to evaluate.
Personalization and content reuse
AI can create increasingly specific experiences based on traveler behavior, destination interest, trip type, seasonality, audience signals, and other factors. Those decisions can influence more than the page experience. A recommendation may affect which itineraries receive demand, which departures fill faster, or which packages receive greater visibility.
Content creates a similar issue. AI dramatically lowers the effort required to adapt an approved message across destinations, itineraries, package tiers, audiences, and channels. Teams need to establish when that reuse can happen within existing guardrails and when a new context requires another decision.
Seasonal pressure
An approval model that works during a slower booking period can become a bottleneck when demand spikes. Teams are trying to respond quickly to changing inventory, pricing, traveler behavior, and competitive conditions at exactly the moment decision volume is highest.
Peak booking season is a bad time to discover that three teams believe they have approval authority over the same experiment. Decision rights need to be established before volume puts them under pressure.
Decision capacity becomes the constraint
Optimizely’s Agent Platform can reduce the effort required to generate ideas, create variants, identify opportunities, and execute parts of an experiment. But increasing the supply of ideas does not automatically increase the number of experiments an organization can put into market.
Once idea generation accelerates, decision capacity can become the constraint. That means the operating infrastructure around Agent Platform needs to establish:
- What agents can execute independently without defined guardrails
- What requires human review
- Which functions need to provide input
- Who has final decision authority
- What triggers escalation
- Who owns the outcome after launch
These are not just governance questions. They determine how much of the capacity created by AI an organization can actually use.
Establishing those decision rights also keeps senior leaders from becoming the default approval layer for every AI-generated idea. When routine decisions have clear owners and guardrails, leadership attention can stay focused on the decisions that genuinely require senior judgment.
For travel organizations scaling Agent Platform, decision architecture becomes part of the operating infrastructure around the technology. The teams that can generate more ideas also need a way to evaluate those ideas, make a clear call, put them into market, and remain accountable for what happens next.
Build ownership before the backlog grows
Travel teams do not need another approval layer for every AI-generated idea. They need clear decision rights that match the speed and volume Agent Platform introduces into the organization.
Road Scholar’s experience shows what that can look like in practice: a consistent review cadence, named decision owners, and clear sign-off before production. The exact model will vary by organization, but the underlying principle holds as Agent Platform adoption expands.
When the supply of ideas increases, the ability to make good decisions at the same pace becomes part of the advantage. Establishing ownership early gives marketing, revenue management, and digital teams a way to use the additional capacity created by AI without creating a new bottleneck downstream.
FAQ: AI decision ownership in travel marketing
Why does AI create more cross-functional decisions in travel? AI can generate more pricing, personalization, content, and experimentation opportunities than teams could reasonably develop manually. In travel, those ideas often affect multiple functions at once. A personalization decision can influence demand, a pricing test can affect messaging, and a content change can shift attention toward particular destinations or departures. As the volume of ideas increases, so does the number of decisions that cross team boundaries.
Who should own AI experimentation in a travel company? There does not need to be one universal owner for every AI experiment. Decision authority should follow the type of decision being made. Revenue management may own pricing decisions, marketing may own messaging, and digital or technology teams may own aspects of deployment. Other teams can contribute to the decision while one named owner retains final authority.
How did Road Scholar improve its AI decision process? As Road Scholar expanded its use of Optimizely’s Agent Platform, it established a weekly AI review, named decision owners, and required end-user sign-off before agents reached production. With that structure in place, the path from an agent concept to production could move in as little as two weeks.
Does AI governance require adding more meetings? Not necessarily. Effective governance establishes decision rights, guardrails, accountability, and escalation paths so teams do not have to renegotiate ownership every time an idea crosses functions. A consistent review cadence can replace scattered Slack messages, email threads, and one-off approval conversations.
Where should travel teams define decision ownership first? Start with areas where AI-generated ideas frequently cross functional boundaries or carry meaningful business implications. Pricing and packaging, personalization, content reuse, inventory-sensitive promotions, and high-volume seasonal campaigns are common places where unclear ownership can slow experimentation.
What should an AI decision framework include? Teams should define what agents can execute within established guardrails, what requires human review, which functions need to be consulted, who has final authority, what triggers escalation, and who remains accountable for the outcome after launch.
Nansen helps travel and hospitality teams build the decision structures, governance, and workflows required to put Optimizely’s Agent Platform into production and scale it responsibly. If AI-generated ideas are piling up faster than your organization can act on them, let’s talk about where decision ownership is getting stuck.
