Travel Industry Faces Two-Sided AI Cost Squeeze
Skift reports the travel industry is being squeezed by AI on both the demand and supply sides. On the demand side, Skift writes that the look-to-book ratio is deteriorating as AI agents run far more searches than human browsers, often without converting; on the supply side, every agent search adds backend cost regardless of whether it ends in a booking. At the Skift Data and AI Summit, Hilton CIO Michael Leidinger described the rising backend-cost problem as "tokenomics," a term Skift says has circulated in enterprise tech. Skift notes Hilton's conversational AI trip planner, built on Anthropic's Claude and live since March, is seeing slightly higher conversion, even as the company closely watches AI token costs.
What happened
Skift reports that the travel industry faces simultaneous AI-driven pressure on conversion and on backend cost. On the demand side, Skift writes that the look-to-book ratio, the number of searches per actual booking, is deteriorating because AI agents generate far more searches than human browsers and often do not convert. On the supply side, Skift reports that each search imposes cost on airlines and hotels regardless of conversion, with search volumes multiplying by orders of magnitude.
The 'tokenomics' problem
At the Skift Data and AI Summit, Hilton CIO Michael Leidinger described the rising backend-cost phenomenon as "tokenomics." Skift reports Hilton's AI trip planner, a conversational tool built on Anthropic's Claude that lets travelers describe a trip in natural language, has been live since March and is delivering slightly higher conversion, even as the company emphasizes careful cost management and AI education across the organization. Skift coverage by Adriana Lee frames the combined effect as a two-sided squeeze on travel distribution.
Why it matters
Agent-driven workflows typically raise request volume per session, broaden context-window usage, and trigger repeated retrievals, which together increase per-customer token consumption and downstream costs. Higher interaction volume does not translate into proportionate bookings, so the economics shift toward margins and distribution budgets rather than incremental revenue.
What to watch
- •Metrics: look-to-book ratio, average searches per session, and token consumption per session reported by distribution partners.
- •Cost levers: adoption of smaller or local models, caching and aggregation layers, and server-side filtering to cut downstream API calls.
- •Commercial terms: any changes in AI provider pricing aimed at high-volume enterprise search.
Bottom line
Skift describes a distinct economic stressor for travel, escalating agent-driven demand that lowers conversion efficiency paired with rising token costs that raise operating expense, that data and platform teams should treat as a structural issue, not an implementation detail.
Key Points
- 1Skift reports AI agents are multiplying travel searches, degrading the look-to-book ratio while adding backend "tokenomics" cost per query.
- 2Travel distribution long relied on human attention limits; removing them shifts unit-economics pressure onto airlines' and hotels' margins.
- 3Mitigations like caching, aggregation, smaller models, and server-side filtering become central to controlling per-search AI spend.
Scoring Rationale
A well-reported, sector-specific economic problem, AI agents degrading travel's look-to-book ratio while raising backend "tokenomics" costs, grounded in named executives and a major industry summit, and relevant to data and platform teams in travel. It is a single-sector trade story rather than a cross-industry paradigm shift, so it rates as a solid notable item.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
