Research Analysis · AI Search Research
What the AirOps 2026 State of AI Search report means for B2B SaaS
An analysis of AirOps' 2026 State of AI Search report: freshness, page structure, community sources and unstable visibility, and what software companies should do about each.
- Original research
- AirOps
- Published
- Last updated
- , sources checked 5 November 2026
- Analysis by
- Asad Shahzad · 5 min read
This is AEO Operators' analysis of research published by AirOps. It is not original research by AEO Operators.
Executive summary
AirOps found that AI search visibility is unstable, that recently updated and well-structured pages are cited more often, and that community platforms supply about 48% of citations.
For a growing software company, this means AI visibility has to be managed as a continuous programme rather than a single project, with regular content refreshes and a deliberate presence on third-party sites.
Research background
AirOps, which sells content software for AI search, published The 2026 State of AI Search on 2 December 2025.
The report combines AirOps' own studies with research from the analyst Kevin Indig, and it looks at which signals were linked to brand visibility in AI answers during 2025.
It covers ChatGPT, Perplexity, Gemini, Google AI Mode and Google AI Overviews. The report is a summary of several studies, so sample sizes and exact collection dates differ between sections and are not all stated on the page.
How the research was done
The figures come from several AirOps analyses of cited pages and repeated AI answers, plus Kevin Indig's analysis of Google AI Overviews. Freshness figures describe how recently cited pages were updated.
Structure figures describe the share of pages cited in ChatGPT that have a given feature. Stability figures come from running the same prompts several times and recording which brands appeared.
- Report
- The 2026 State of AI Search: How Modern Brands Stay Visible
- Research period
- 2025 (report published 2 December 2025)
- AI platforms
- ChatGPT, Perplexity, Gemini, Google AI Mode, Google AI Overviews
- Original author
- AirOps, with research from Kevin Indig
Key research statistics
30%
of brands stay visible from one answer to the next.
Repeated AI answers to the same prompt. Source: AirOps
>3×
more likely to lose citations if a page is not updated each quarter.
Compared with recently refreshed pages; an observational association.. Source: AirOps
~48%
of AI search citations come from community and user-generated sources.
About 48%, across platforms studied. Source: AirOps
~59.6%
of AI Overview citations come from pages outside the top 20 organic results.
Google AI Overviews. Source: AirOps
The numbers in charts
Each chart uses the values published in the original report.
How recently cited pages were updated
Share of AI-cited pages updated within each window, all citations compared with commercial citations
Grouped bar chart. Over 70% of all cited pages and about 83% of commercial citations were updated within 12 months. Over 50% of all cited pages and over 60% of commercial citations were updated within 6 months.
Show the data table
| Update window | All cited pages | Commercial citations |
|---|---|---|
| Within 12 months | ~70% | ~83% |
| Within 6 months | ~50% | ~60% |
Method and limits: AirOps reports these as 'more than 70%', 'more than 50%', 'about 83%' and 'more than 60%'. The bars show the reported figure, so the real values for 'more than' figures are somewhat higher.
Research period: 2025. Sample: Not stated on the report page.
Values marked with ~ are approximate as published or read from the original chart.
Data source: AirOps, The 2026 State of AI Search, 2 December 2025. Visualization recreated by AEO Operators. Values are approximate reconstructions from the published chart. View the original report
Grouped bar chart. Over 70% of all cited pages and about 83% of commercial citations were updated within 12 months. Over 50% of all cited pages and over 60% of commercial citations were updated within 6 months.
Structural features reported for cited pages
Share of cited pages that have each feature
Horizontal bar chart. 87% of cited pages use a single H1, nearly 80% use lists, 68.7% follow sequential headings, about 61% use three or more schema types and 10.5% use FAQ schema.
Show the data table
| Feature | Share of cited pages |
|---|---|
| Single H1 | 87% |
| Lists for key information | ~80% |
| Sequential heading hierarchy | 68.7% |
| Three or more schema types | ~61% |
| FAQ schema | 10.5% |
Method and limits: AirOps gives lists as 'nearly 80%' and three or more schema types as 'about 61%'. These are correlations, not proof that the feature causes citations. The report explicitly identifies ChatGPT for the heading-hierarchy and list findings, but does not explicitly establish that platform scope for every other feature.
Research period: 2025. Sample: Not stated on the report page.
Values marked with ~ are approximate as published or read from the original chart.
Data source: AirOps, The 2026 State of AI Search, 2 December 2025. Visualization recreated by AEO Operators. Values are approximate reconstructions from the published chart. View the original report
Horizontal bar chart. 87% of cited pages use a single H1, nearly 80% use lists, 68.7% follow sequential headings, about 61% use three or more schema types and 10.5% use FAQ schema.
How stable brand visibility is across repeated answers
Share of brands or answers meeting each condition
Bar chart. 30% of brands stay visible across back-to-back answers, 20% across five consecutive runs, and about 28% of answers include a brand that is both mentioned and cited.
Show the data table
| Measure | Share |
|---|---|
| Brands visible in back-to-back answers | 30% |
| Brands visible across five runs | 20% |
| Answers with a brand both mentioned and cited | ~28% |
Method and limits: The 28% figure is reported as about 28%. The first two figures are shares of brands, while the third is a share of answers, so they describe different things.
Research period: 2025. Sample: Not stated on the report page.
Values marked with ~ are approximate as published or read from the original chart.
Data source: AirOps, The 2026 State of AI Search, 2 December 2025. Visualization recreated by AEO Operators. Values are approximate reconstructions from the published chart. View the original report
Bar chart. 30% of brands stay visible across back-to-back answers, 20% across five consecutive runs, and about 28% of answers include a brand that is both mentioned and cited.
Growth in page views since the AI Overviews rollout
Change in page views for AI Overview queries compared with other searches since May 2024
Bar chart. Page views from AI Overview queries grew 21.5% since May 2024, compared with 1.3% for other searches.
Show the data table
| Query type | Page view growth |
|---|---|
| Queries with AI Overviews | 21.5% |
| Queries without AI Overviews | 1.3% |
Method and limits: Kevin Indig's analysis as reported by AirOps. The report also notes that queries without AI Overviews drive about twice as much traffic because they capture higher-intent users.
Research period: May 2024 onwards. Sample: Not stated on the report page.
Data source: AirOps and Kevin Indig, The 2026 State of AI Search, 2 December 2025. Visualization recreated by AEO Operators. View the original report
Bar chart. Page views from AI Overview queries grew 21.5% since May 2024, compared with 1.3% for other searches.
What the research actually reveals
The report's own conclusion is that freshness, off-site credibility and clear structure all matter.
Our reading is that the most important finding for software companies is the instability: if only 30% of brands stay visible from one answer to the next, a single screenshot of a ChatGPT answer tells you very little.
The freshness data is also more pointed for software than for many sectors.
AirOps says that in SaaS, finance and news, pages older than three months see steep drops in citation likelihood, which means a product comparison page written at launch and never touched again is probably losing ground.
The structural findings are correlations. A single H1 appearing on 87% of cited pages does not prove that a single H1 causes citations, because most well-maintained sites use one anyway.
We treat these findings as hygiene that removes reasons not to be cited, not as levers that guarantee results.
The community data shows large differences between platforms.
Perplexity references community platforms in more than 90% of answers, while Gemini does so in as few as 7%, so a brand that is strong on Reddit and weak elsewhere will look very different from one AI platform to the next.
Implications for Series A to Series C SaaS companies
For founders at an early stage, the main implication is to measure visibility as a trend over many runs rather than trust one answer.
For growth-stage marketing leaders, it means budgeting for content refreshes every quarter on the pages that buyers compare, such as pricing, alternatives and integration pages.
Product marketing leaders should own the structure and accuracy of comparison content, because that is where outdated claims do the most harm.
Technical leaders should make sure pages are crawlable and use clean heading hierarchies and relevant schema, since these are cheap to fix and appear on most cited pages.
Actionable recommendations
- Run your most important buyer prompts at least five times on each platform and record how often you appear, not just whether you appear.
- List every page that answers a commercial question about your product and give each one an owner and a quarterly review date.
- Check that each of those pages has one H1, headings in order, lists for key facts and schema that matches the visible content.
- Find the Reddit threads and YouTube videos that already rank for your category questions and make sure your product is described accurately there.
- Track mentions and citations separately, because AirOps reported that brands earning both were 40% more likely to resurface across consecutive runs than citation-only brands.
Research limitations
AirOps sells software for AI search content, so it has a commercial interest in showing that content work matters. The page summarises several studies without giving every sample size or collection date.
Most findings are correlations, and the report does not separate results for B2B software from other industries except in the freshness section.
Some values are given as approximate or lower bounds, such as 'more than 70%', and we show them that way.
Original sources and further reading
- The 2026 State of AI Search: How Modern Brands Stay Visible
AirOps, AirOps, with Kevin Indig, 2 December 2025. Section: Whole report. Vendor research.
Primary source for every figure in this article.
Want to understand how your SaaS brand appears across AI search platforms?
Get an AI Visibility Assessment from AEO Operators, and we will apply what this research shows to your own company and competitors.
Get an AI Visibility Assessment