How AI Search Is Recommending Flight Bookings: Monthly Trends

Executive Summary

The Flight Bookings category was stable but repositioning in October 2026. AirHelp remains the coverage leader at 2.3 percent, unchanged from September 2026 and up 0.4 points from 1.9 percent in July 2026, and the benchmark classifies the month as quiet with no brand recording a significant movement in valid recommendation coverage. The gap to the rest of the field remains wide: AirHelp is the only brand holding any valid recommendation coverage this month, after Flightright's 2.3 percent in September fell back to zero.

The sharpest supporting movement was in raw mention presence, not recommendation outcomes. AirHelp's presence rate climbed to 86.1 percent in October 2026 from 40.2 percent in July 2026, a 45.9 point gain, and the brand now appears in 37 of 43 qualified observations. AirAdvisor moved the other way, down to 23.3 percent in October from 57.9 percent in July, a 34.6 point decline.

The sharpest single-month decliner on coverage was Flightright, which moved from 2.3 percent in September 2026 to 0.0 percent in October, ending the two-month streak it had built on single qualified recommendations. Flightright's presence, however, edged up to 4.7 percent from 1.9 percent in July, so it remains visible without being recommended.

Against the prior month, the category produced no coverage change at the leader and no new entrant into recommendation outcomes. The qualified base continued its long contraction, from 107 observations in July 2026 to 55 in August, 44 in September, and 43 in October, which means each percentage now rests on a thinner set of prompts than at the start of the series.

Each monthly run begins with 620 prompt-surface observations (476 unique questions) in July 2026 and 228 prompt-surface observations (177 unique questions) in October 2026 across the benchmark's defined AI/search surface universe. Of those, 414 in July and 228 in October mentioned a tracked brand or competitor; 157 in July and 84 in October were relevant, while 257 in July and 144 in October were irrelevant. The public metrics use the 107 qualified observations in July and the 43 in October that survive both qualification stages. Intermediate months followed the same funnel: August opened at 706 prompt-surface observations (531 unique questions) and September at 261 (198 unique questions).

Key Findings

AI Response Inconsistency Alerts

Two critical or high-severity factual inconsistencies were detected this month, both involving AirHelp and both spanning three AI platforms: Copilot, AI Mode, and AI Overviews.

AirHelp

AI platforms provided conflicting information about the price of an AirHelp+ annual membership, on a high-severity, high-confidence pricing conflict. When asked "Is AirHelp a free service?", Copilot stated that the membership starts at $179.99 per year, citing the company's Our Fees and price-list pages, while AI Mode stated that it starts at around €39.99 per year, citing an AirHelp blog explainer and the brand's international homepage. The flagged source pages carry the same excerpt on both sides: "Purchase before flight disruption From $179.99 per year" on the US fee page and "From €39.99 per year" on the international fee page. The two price points are incompatible as a starting figure for the same membership.

A second high-severity, high-confidence pricing conflict surfaced on the same question. Copilot again stated the membership starts at $179.99 per year, while AI Overviews stated it falls in a range of around €29.99 to €39.99 per year, citing AirHelp pages and a third-party review titled "AirHelp Review: Is This Flight Compensation Service Worth It?". The flagged sources again point to the two regional fee pages, one showing "From $179.99 per year" and the other "From €39.99 per year", which cannot both describe the same entry price.

The commercial consequence is that a prospective customer asking whether the service is free can receive materially different membership pricing depending on which AI surface answers the question. This is a currency and regional pricing mismatch exposed by inconsistent source pages, not an endorsement signal, and it sits alongside the recommendation benchmark rather than within it.

Benchmark Context

Questions This Section Answers

• How much did the qualified flight compensation observation set shrink between July and October 2026?
• Which prompt-surface stages are filtered out before a brand's flight compensation recommendation percentage is calculated?

The report separates the raw collection universe from the qualified analysis set. Brand-level recommendation percentages are calculated within the qualified benchmark set.

The qualified set contracted again in October, from 107 observations in July 2026 to 43, its smallest base in the series. The raw collection also shrank, from 620 to 228 prompt-surface observations. August 2026 recorded 55 qualified observations and September 2026 recorded 44. All six canonical surface families, ChatGPT, Copilot, Gemini, Perplexity, AI Overviews, and AI Mode, remained active across the series.

Benchmark-Level Metrics

The valid recommendation shortlist share moved from 1.9 percent in July 2026 to 2.3 percent in October 2026, and the recommendation-shaped answer share from 0.9 percent to 2.3 percent over the same period. In August 2026 those figures were 3.6 percent and 1.8 percent, and in September they were 2.3 percent and 4.5 percent, so both measures have moved within a narrow band on a shrinking base rather than trending decisively.

AI Recommendation Trend

Questions This Section Answers

• Which flight compensation brand holds valid recommendation coverage in October 2026, and how many qualified recommendations does it rest on?
• Did any brand record a top-three or rank-one recommendation this month?

AirHelp Holds the Only Coverage Position in a Category Where Recommendations Remain Scarce

AirHelp is the sole brand with valid recommendation coverage in October 2026, at 2.3 percent on a single qualified recommendation within 43 observations. No other brand holds coverage, and no brand recorded a top-three placement or a rank-one recommendation this month.

No brand exceeded normal month-to-month variation on coverage this month, and the benchmark classifies every brand as stable. The category-level result comes from AirHelp's single qualified recommendation persisting while Flightright's September recommendation did not repeat, not from any brand making a decisive move.

What Changed This Month

Questions This Section Answers

• How did AirHelp's presence change versus its recommendation coverage in October 2026?
• Why did Flightright's flight compensation coverage fall to zero despite stable presence?
• What happened to AirAdvisor's and Skycop's presence across the series?

AirHelp

AirHelp's valid recommendation coverage held at 2.3 percent in October 2026, unchanged from September 2026 and up 0.4 points from 1.9 percent in July 2026. The figure rests on a single qualified recommendation out of 43 observations. The benchmark classifies the movement as stable.

The real movement was visibility. AirHelp's raw mention presence rate moved from 40.2 percent in July 2026 to 86.1 percent in October 2026, up 45.9 points. The brand appeared in 37 of 43 qualified observations in October, up from 43 of 107 in July. On sentiment, AirHelp recorded four positive mentions and zero negative in October, a net sentiment score of 0.11, up from a net sentiment score of 0.07 in July.

The distinction the reader should notice: AirHelp now appears in more than four of every five qualified observations while holding recommendation coverage of just 2.3 percent and no top-three or rank-one placements. Presence and recommendation remain separate signals in this category, and the gap between them is the widest it has been in the series.

Highest-priority diagnostic: Which prompts mention AirHelp across 37 of 43 observations but stop short of recommending it, and which competitor is chosen when AirHelp is present but not selected?

Flightright

Flightright's valid recommendation coverage moved from 2.3 percent in September 2026 to 0.0 percent in October 2026, a decline of 2.3 points. The brand had recorded one qualified recommendation out of 44 observations in September, and one out of 55 in August. The benchmark classifies the change as stable and notes a one-month downward streak on coverage.

Presence moved the other way in the full series. Flightright's raw mention presence rate moved from 1.9 percent in July 2026 to 4.7 percent in October 2026, appearing in 2 of 43 qualified observations. The brand also recorded one positive and one neutral mention in October, a net sentiment score of 0.5, its strongest sentiment reading in the series.

The distinction the reader should notice: Flightright has now shown it can appear without converting to a recommendation, and its recommendation history rests entirely on single observations in two prior months. On this base the signal is directional, not established.

Highest-priority diagnostic: Which prompts produced the August and September recommendations, and why did that conversion not repeat in October despite stable presence?

AirAdvisor

AirAdvisor's valid recommendation coverage held at 0.0 percent, unchanged from July 2026. The brand has recorded no qualified recommendations in any month of the series.

The movement was in presence. AirAdvisor's raw mention presence rate moved from 57.9 percent in July 2026 to 23.3 percent in October 2026, down 34.6 points. The brand appeared in 10 of 43 qualified observations in October, down from 62 of 107 in July. August 2026 recorded 29.1 percent and September 29.5 percent, so the decline occurred mainly between July and August and has continued to drift down into October.

The distinction the reader should notice: AirAdvisor was the most visible brand in the category in July 2026 and has since fallen behind AirHelp on presence while never holding recommendation coverage. Visible does not mean recommended, and losing presence narrows the prompts on which any recommendation could occur.

Highest-priority diagnostic: Which surfaces and prompt types account for the presence decline, and does the smaller presence set overlap with the prompts that generate recommendations?

Skycop

Skycop's valid recommendation coverage held at 0.0 percent across all four months of the series, with zero qualified recommendations out of 107 observations in July, 55 in August, 44 in September, and 43 in October. Skycop's presence also fell to zero in October, down from 3.7 percent in July 2026, when it appeared in 4 of 107 qualified observations, a 3.7 point decline. In October the brand appeared in none of the 43 qualified observations.

AirHelp records 2.3 percent valid recommendation coverage in October 2026, while Skycop stands at 0.0 percent, a gap that has held across every month of the series. The more important issue is what that gap means for Skycop's recommendation position: AirHelp's presence rate climbed to 86.1 percent over the same period, meaning the one brand holding any recommendation coverage this month is also the brand most consistently mentioned, while Skycop has lost the presence that would even create the opportunity to be considered. This is a measurable recommendation gap, not a difference in overall category attention, and it leaves Skycop as the only tracked brand in October 2026 with zero presence and zero recommendation coverage at the same time. Left unresolved, the question is not whether Skycop can close a narrow gap to the leader but whether it can re-establish any presence at all on the prompts that generate recommendations.

Highest-priority diagnostic: Which prompts mentioned Skycop in July, which surfaces produced them, and why did that presence disappear entirely by October?

Buyer-Intent Interpretation

Questions This Section Answers

• Which buyer-intent clusters produced qualified flight compensation observations, and which ones did not?
• What can the benchmark not yet answer about pricing and multi-brand comparisons in this category?

In all four months of the series, the qualified observations fell entirely into the Brand Recommendation cluster. The Pricing & Value and Multi-Brand Comparison clusters recorded no qualified observations in any month. The public benchmark therefore cannot yet answer how AI systems describe pricing trade-offs, fee structures, or direct brand comparisons in this category, even though the inconsistency data shows pricing questions are being asked and answered inconsistently across surfaces.

Brand Opportunity Summary

Questions This Section Answers

• For each tracked flight compensation brand, what signal does the benchmark surface in October 2026?
• Which brands show a gap between presence and recommendation coverage that warrants further investigation?

The benchmark identifies where attention is warranted; a company-level analysis is needed to explain why.

Evidence Behind the Benchmark

The aggregate metrics are built from prompt-level observations (query, surface, recommendation outcome, rank, sentiment, and citations where exposed). Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns. Source presence is not automatically treated as proof of causation.

About This Benchmark

This report is part of the LLM Authority Index AI Visibility Market Discovery research program.

• AI Visibility Industry Market Methodology
• AI Visibility Industry Market Metrics
• AI Visibility Industry Market Standards

Report-Specific Interpretation Notes

• Small-count movement: AirHelp's 2.3 percent coverage rests on a single qualified recommendation out of 43 observations, and Flightright's prior coverage rested on single recommendations in August and September. These indicate direction, not established position.
• Qualified denominator vs raw collection: the qualified set contracted from 107 observations in July 2026 to 43 in October 2026, which concentrates percentages on a smaller base in the current month.
• Presence versus coverage: AirHelp's presence rise and AirAdvisor's presence decline did not produce corresponding recommendation coverage movements, keeping visibility and recommendation outcomes as distinct signals.
• Platform pricing inconsistency: two high-severity pricing conflicts on the AirHelp+ membership appeared across Copilot, AI Mode, and AI Overviews, reflecting conflicting regional source pages rather than a benchmark recommendation signal.
• Directional analysis: month-over-month movement identifies changes worth investigating; it does not by itself establish the cause of those changes.

Next Step

The Public Benchmark Shows Where a Brand Is Winning or Losing. A Company-Level Audit Shows Why.

The aggregate percentages raise questions that sit beneath the surface: which high-intent prompts are won, which competitor takes the recommendation when a brand loses, what attributes AI systems associate with each option, and which external sources shape those answers. In a category where only one brand holds any recommendation coverage, the question is not just who leads on a single observation, but why an entire field with high presence is converting so little of it into recommendations.

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy.

Request an AI visibility audit