Highlights
- The dataset: This study is an analysis of over 200,000 AI source citations using prompts specific to the airline industry. It analyzes AI responses from four AI models run in 12 countries over the course of Q2 2026.
- Facebook fell from #1 to #6 as Reddit climbed to the top: Text-first sources displaced image and video social platforms.
- Reddit alone makes up 12% of ChatGPT's citations: That makes ChatGPT the most concentrated of the four models.
- ~82% of top sources differ between any two models: Any two of the four share only about a fifth of their top-20 cited airline domains (range 8% to 38%).
- 55% of Microsoft Copilot's citations are from editorial domains: The most editorial model.
Which sources should you target to get cited as an airline brand?
Each model has a signature set of distinctive top sources.
| Model | Top distinctive sources (citation share) | Citation character |
|---|---|---|
| Grok | Facebook (5.5%), worldairlineawards.com (4.2%), Reddit (4.0%), YouTube (3.2%) | Social + awards/ratings |
| Google AI Overview | Reddit (4.7%), YouTube (3.5%), alternativeairlines.com (3.2%), Facebook (2.9%) | Social + booking aggregator |
| Microsoft Copilot | airadvisor.com (2.5%), The Points Guy (2.3%), lefigaro.fr (2.0%), Time Out (1.9%) | Passenger-rights + points media |
| ChatGPT | Reddit (12.2%), Time Out (2.1%), worldairlineawards.com (1.7%), Reuters (1.4%) | Forum-led + newswire |
An airline tuning its content for the way one assistant reads the web is absent from roughly 82% of the domains the others prefer.
How have AI source rankings changed over time in the airline industry?
Reddit overtook Facebook as the single most-cited airline domain. We split the 15.8-week window at its midpoint and ranked domains by citation share within each half. Reddit's share climbed from 3.6% to 6.0% (rank #2 → #1) while Facebook's fell from 4.3% to 1.4% (rank #1 → #6).
We observe a rotation away from image-and-video social platforms (Facebook, Instagram, TikTok, YouTube) toward text-heavy sources the models can quote directly, Reddit threads, booking aggregators like alternativeairlines.com, and editorial outlets like Forbes and The Points Guy. We report this as an observation over a single 15.8-week window, not a confirmed long-term trend, track it in the live AI visibility rankings.
What type of content do AI models cite for airlines?
The models differ not just on which domains but on what type of content they trust. We classified every cited domain into six categories, the headline split is editorial versus commercial versus user-generated content.
Microsoft Copilot is the most editorial model by a wide margin and the least reliant on user-generated content, its airline answers are built from travel journalism and consumer-rights publications, not forums. Google AI Overview is the most commercial, the only model where commercial domains (35.4%) outweigh editorial ones. Grok is the most forum-heavy: a quarter of its airline citations are UGC, reflecting its appetite for Reddit, social platforms, and discussion sites.
Do AI models cite local-language content for airlines?
Outside English-speaking markets, only about 38% of airline citations land on a local-country domain, the majority of sources answering German, French, Italian, Dutch, Swedish, and Spanish airline questions sit on international, largely English-defaulted domains. Using country-code top-level domains as a proxy for local-language sourcing, the local-domain rate ranges from 55% (Swedish) down to 27% (Spanish). English prompts are excluded.
Spanish stands out for how rarely models cite its local sources, barely a quarter of citations sit on a Spanish-speaking ccTLD, likely because the Spanish-language airline web is split across many generic .com properties and several countries. Because the proxy treats a German-language article on a .com domain as non-local, these figures are a conservative floor on true localization.
Which AI model relies most on local sources for airlines?
Microsoft Copilot and Google AI Overview rely most on local sources; Grok relies on local sources least. The gap between models is large and consistent, Microsoft Copilot and Google AI Overview reach for local-country domains roughly half the time in non-English markets, while Grok does so only about a quarter of the time.
The pattern is consistent: Grok is the most English-defaulting model in every single language, while Google AI Overview relies on local sources the most in northern-European markets (82–84% in Dutch and Swedish). For an airline's local-market content team, a Dutch or Swedish landing page is far likelier to surface in Google AI Overview than in Grok.
Should airlines optimize for each AI model separately?
Yes, almost entirely. When the four models answer the same airline questions, their top-cited domains barely converge. We compared the 20 most-cited domains for each model and measured how much they share.
How many sources does each AI model cite per answer?
Very different amounts, from 5.8 to 26. Grok is the firehose; ChatGPT is the sommelier.
Grok cites about 4.5 times as many sources per answer as ChatGPT, and about 4.2 times as many as Microsoft Copilot. That breadth is double-edged for brands: more slots to appear in, but also more competition diluting any single citation's weight.
Context
This analysis draws from Temso's AI visibility monitoring platform, which tracks how brands appear in AI model responses across ChatGPT, Microsoft Copilot, Grok, and Google AI Overview. The dataset covers airline-related prompts, "what's the best business-class carrier," "which airline handles delays best", across 12 markets and 7 languages over roughly 15.8 weeks (mid-March to late June 2026), drawing on more than 203,000 individual cited sources from nearly 21,700 model responses.
It measures what AI models cite, not the underlying quality of any airline. These findings reflect citation behavior for airline-related prompts in the monitored markets, and may not generalize to all carriers, prompt types, or other industry verticals.
Methodology
How we measured this
We tracked four AI models responding to airline prompts across 12 markets and 7 languages over roughly 15.8 weeks. Each response was parsed to extract its source citations, the URLs, domains, and metadata referenced. To measure how much the models agree, we took each model's 20 most-cited domains and calculated how many they share pair by pair; the headline overlap is the average across all six pairs. Domain categories (editorial, commercial, user-generated, reference, institutional) were assigned from a global domain registry. To track movement, we split the window in half and compared each domain's share of citations in the first half against the second. Model coverage varied across the observation window, so temporal movements are reported as relative ranks within each half.
The analysis covers over 200,000 cited sources from 21,689 AI responses. Sample sizes exceeded recommended thresholds for every reported finding. Localization is measured by a country-code-domain proxy, a German-language article on a .com domain counts as non-local, so those figures are a conservative floor and exclude English-language prompts, which sit overwhelmingly on generic domains. The Grok-versus-Microsoft Copilot localization gap was validated with a two-proportion test (p < 0.001). All findings are observational and describe citation behavior, not causal mechanisms inside the models.
Frequently asked questions
Do AI models cite different sources for the same airline question?
Yes, dramatically so. Any two of the four models share only about 18% of their top-20 cited domains on average, roughly 82% of the sources one model relies on never appear in another's top tier.
Which AI model cites the most sources per airline answer?
Grok, by a wide margin, about 26 sources per response, versus 5.8 for ChatGPT, a roughly 4.5-fold gap. More sources means more chances to appear, but also more dilution per citation.
Which AI model is best for reaching non-English airline markets with local content?
Microsoft Copilot and Google AI Overview. Both cite local-country domains roughly half the time in non-English markets (54.7% and 48.2%), while Grok defaults to English and cites local sources for only about 27% of the time. Google AI Overview is especially strong in Dutch and Swedish (82–84%).
What type of content should an airline create to get cited by AI?
It depends on the model you're targeting. Microsoft Copilot favors editorial travel and consumer-rights media (54.7% editorial). Google AI Overview leans commercial, airline-owned and booking domains (35.4% commercial). Grok rewards forum and social presence (24.8% UGC). There is no single content type that wins across all four.
Why does Reddit matter so much for airlines?
It's become the single most-cited airline domain, and it dominates ChatGPT specifically, 12.2% of all ChatGPT airline citations point to Reddit. Active, helpful participation in airline and travel subreddits moves visibility across multiple models at once.
Did the airline source ranking change over the study period?
Yes, visibly. Reddit overtook Facebook for the top spot, and image/video social platforms (Facebook, Instagram, TikTok) lost ground to text-first editorial and aggregator sites the models can quote directly. We frame this as an observation over a single 15.8-week window, not a confirmed long-term trend.
Is "local-language citation rate" the same as local-language content?
Not exactly. We use country-code domains as a proxy, so a German-language article on a .com domain is counted as non-local. The real local-language share is therefore somewhat higher than the reported figures, treat them as a conservative floor.
Can an airline optimize for one AI model and expect to show up in the others?
Largely no. With ~82% of top sources differing between any two models, a strategy tuned to one assistant's preferred domains leaves a brand mostly invisible in the rest. Visibility across the AI search requires a portfolio approach spanning editorial, commercial, and community sources.
