{"id":9867,"date":"2026-08-14T16:19:28","date_gmt":"2026-08-14T16:19:28","guid":{"rendered":"https:\/\/pros.com\/learn\/?p=9867"},"modified":"2026-08-14T16:19:28","modified_gmt":"2026-08-14T16:19:28","slug":"mentioned-is-not-cited-the-ai-search-kpis-airlines-should-track","status":"publish","type":"post","link":"https:\/\/pros.com\/learn\/blog\/mentioned-is-not-cited-the-ai-search-kpis-airlines-should-track\/","title":{"rendered":"Mentioned Is Not Cited: The AI Search KPIs Airlines Should Track\u00a0"},"content":{"rendered":"
<\/div>\n

Airline Mentions Are a Baseline, Not the Outcome<\/strong> <\/h2>\n\n\n\n

In AI search, airline mentions are often table stakes. That may sound counterintuitive, but consider the structure of the market: for a given city pair, only a limited number of airlines usually offer relevant service. If a traveler asks which airlines fly the route, a carrier that operates it has a reasonable chance of being named. <\/p>\n\n\n\n

More importantly, PROS research<\/a> across 376 city-to-city flight prompts found that 74.6% of airline mentions in ChatGPT were not linked to the airline’s own website. An airline can be named while the supporting links point to an online travel agency (OTA) or flight aggregator. The brand receives exposure, but the click goes to another domain. <\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Mentions are not worthless. They can reveal route association, brand awareness, and competitive presence. However, they are better treated as baseline or low-discrimination metrics. The more selective\u2014and more consequential\u2014outcome is an airline-owned citation that gives AI users a direct path to the airline\u2019s website. <\/p>\n\n\n\n

This article defines the citation and mention-to-link KPIs airlines should prioritize, explains how to build a stable route-prompt panel, and outlines the controls, segmentation, tools, and reporting cadence needed to interpret the results. It also shows how citation gaps can indicate specific improvements to airline flight pages. <\/p>\n\n\n\n

Mentions vs. Citations<\/strong> <\/h2>\n\n\n\n

For this framework, a mention means that the airline’s brand name appears in the generated answer. <\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

A citation means that the answer contains a clickable airline-owned URL. Depending on the platform, that link may appear: <\/p>\n\n\n\n

Inline, within the generated response.<\/strong> <\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

In a source or citation panel associated with the response.<\/strong>\u00a0<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Mentions and citations can occur independently. An answer can name an airline without linking to it. It can also cite an airline-owned route page without prominently naming the carrier in the prose. <\/p>\n\n\n\n

Neither citation placement guarantees a click. It does create the possibility of a click, credit the airline as a source, and give travelers a direct path to first-party information. That makes citation performance the stronger leading indicator of airline-owned visibility. <\/p>\n\n\n\n

The Primary KPI Family: Citation Performance<\/strong> <\/h2>\n\n\n\n

The primary scorecard should measure airline-owned links across all tracked prompts that returned an AI answer, regardless of whether the airline was named. <\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Overall Citation Rate<\/strong> <\/h3>\n\n\n\n

Overall Citation Rate is the best headline KPI in the framework. It answers a straightforward question: among the tracked prompts that produced an AI answer, how often did the airline earn at least one link to its own domain? <\/p>\n\n\n\n

Counting prompts rather than raw links prevents responses that include multiple citations to the same domain from overwhelming the metric. Deduplicate source and inline links at the prompt level so a response containing both still counts once toward the overall rate. <\/p>\n\n\n\n

Source Citation Rate<\/strong> <\/h3>\n\n\n\n

Source Citation Rate indicates whether the airline’s pages are selected as supporting material. It is especially useful for evaluating whether route pages contain sufficient reliable, retrievable evidence about fares, schedules, connections, and the route itself. <\/p>\n\n\n\n

Inspect source citations at the URL level. A link to the airline’s homepage is not equivalent to a link to the relevant city-to-city page. The route page gives the traveler continuity and shows the airline which content the AI system actually retrieved. <\/p>\n\n\n\n

Inline Citation Rate<\/strong> <\/h3>\n\n\n\n

Inline Citation Rate captures airline-owned links embedded directly in the generated answer. These links may be especially visible in some interfaces, but their presentation varies by platform and can change over time. <\/p>\n\n\n\n

Keep source and inline rates separate because they describe different placements. Overall Citation Rate provides the deduplicated summary. <\/p>\n\n\n\n

The Diagnostic KPI Family: Mention-to-Link Performance<\/strong> <\/h2>\n\n\n\n

Mention metrics become more useful when connected to the link outcome. This diagnostic family starts only with prompts that name the airline and ask whether an airline-owned citation accompanied the mention. <\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The separation between the two KPI families is intentional. Citation Performance measures link coverage across every AI answer. Mention-to-Link Performance diagnoses what happened specifically when the airline was named. This also avoids a blind spot: an airline can earn a citation without being named in the answer text. <\/p>\n\n\n\n

Linked Demand and Demand at Risk add traditional keyword volume as a prioritization overlay. They can help distinguish a citation gap on a major route from one on a lower-priority route, but they are not estimates of prompt volume or AI-search traffic. <\/p>\n\n\n\n

Track Core Route Prompts, Then Test Meaningful Qualifiers<\/strong> <\/h2>\n\n\n\n

Travelers can express the same underlying need for a flight in many ways. A prompt might include dates, passenger counts, cabin preferences, budgets, nonstop requirements, loyalty considerations, or several qualifiers at once. There is no reliable platform-wide dataset of prompt volumes that shows how frequently those combinations are used. <\/p>\n\n\n\n

Use Seed Route Prompts for AI Overviews and AI Mode<\/strong> <\/h3>\n\n\n\n

Attempting to track every variation would create a large and unstable prompt set without necessarily producing a better benchmark. For Google AI Overviews and AI Mode, start with a consistent panel of unbranded city-to-city route queries, such as: <\/p>\n\n\n\n

Flights from London to Miami<\/em> <\/p>\n\n\n\n

These seed prompts represent the route subject the airline’s flight page is intended to cover. They are especially defensible for Google’s AI search experiences because users access them through Google Search and may continue to start with familiar route queries from conventional keyword research. <\/p>\n\n\n\n

Traditional keyword tools cannot measure demand for conversational prompts, but they can identify routes with meaningful search demand and help prioritize city pairs for AI citation monitoring. <\/p>\n\n\n\n

Add Short Retrieval Modifiers for Conversational Interfaces<\/strong> <\/h3>\n\n\n\n

For conversational platforms such as ChatGPT, retain the same route subject while adding a short retrieval-oriented modifier: <\/p>\n\n\n\n

Find me flights from London to Miami.<\/em> <\/p>\n\n\n\n

Wording such as “find me” signals that the user wants current external information and may make web retrieval more likely. It does not guarantee that a web search will occur. When possible, confirm that the tracking method measures a web-enabled response rather than an answer based only on model knowledge and conversational context. <\/p>\n\n\n\n

Use Qualifiers as a Controlled Test Layer<\/strong> <\/h3>\n\n\n\n

Use a smaller secondary panel to test how meaningful qualifiers affect citation performance. For example: <\/p>\n\n\n\n

Find me nonstop flights from London to Miami.<\/em> <\/p>\n\n\n\n

The goal is not to reproduce every traveler conversation. It is to determine whether the airline continues to earn citations when an additional requirement changes the information the system must retrieve. <\/p>\n\n\n\n

Begin with one modifier at a time, such as nonstop, round-trip, business class, a travel month, or a budget, before testing more complex combinations. This makes it easier to identify which added context changes citation ownership, cited URLs, or competing sources. <\/p>\n\n\n\n

Derive Qualifiers From Recurring Fan-Out Patterns<\/strong> <\/h3>\n\n\n\n

Google confirms that AI Mode and AI Overviews can break a query into subtopics and issue related searches. OpenAI<\/a> similarly explains that ChatGPT Search may rewrite a prompt into one or more targeted queries. See the Google AI search guide<\/a> for Google’s description of query fan-out. <\/p>\n\n\n\n

Sample core route prompts across the targeted surfaces and look for recurring fan-out themes. Record generated searches where they are observable; otherwise, use the subjects emphasized in the response and its cited pages as directional evidence. <\/p>\n\n\n\n

For city-to-city flight prompts, useful qualifier families to evaluate may include: <\/p>\n\n\n\n