Research Analysis · AI Search Research
Which commercial pages are more likely to earn AI citations?
AirOps found that citation-associated content signals differed by buying stage: evidence helped discovery, concise sentences helped shortlisting, tables helped comparison and organised lists helped validation. For B2B SaaS, the practical response is to match page structure to the decision being made, not apply a universal formatting formula.
- Original research
- AirOps
- Published
- Last updated
- , sources checked 5 November 2026
- Analysis by
- Asad Shahzad · 4 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 citation-associated content signals differed by buying stage: evidence helped discovery, concise sentences helped shortlisting, tables helped comparison and organised lists helped validation.
For B2B SaaS, the practical response is to match page structure to the decision being made, not apply a universal formatting formula.
Research background
This AirOps study examines the step between retrieval and citation in ChatGPT commercial answers.
Its dataset contains 217,508 retrieved pages across 7,500 commercial prompts, split into awareness, research, comparison and validation.
Unlike a study of cited pages alone, it compares cited pages with retrieved pages that were not selected.
How the research was done
AirOps reports 7,500 commercial queries and 217,508 retrieved pages, with 1,875 queries in each buying-stage group.
It extracted 25 on-page signals and used logistic regression, random forest models, sensitivity variants and domain-authority-stratified modelling. This article interprets the public results; it does not rerun the models.
The chart compares reported relative associations, not absolute citation rates.
- Report
- From Retrieved to Cited: How Commercial Content Earns Citations in AI Search
- Research period
- Report published April 2026
- AI platforms
- ChatGPT
- Original author
- Oshen Davidson
Key research statistics
20.3%
higher citation likelihood accompanied evidence-rich discovery content..
Pages with 5 to 7 statistics.. Source: AirOps
18.8%
more citations accompanied concise shortlist content..
Sentences averaging 10 words or less.. Source: AirOps
Up to 26.9%
more citations accompanied organised validation lists..
Validation pages with 8 list sections.. Source: AirOps
The numbers in charts
Each chart uses the values published in the original report.
Reported citation associations at different buying stages
Different buying stages and conditions; not absolute citation rates or additive effects.
Reported citation associations at different buying stages. Discovery: evidence: 20.3% ; Shortlisting: concise sentences: 18.8% ; Comparison: tables: 25.7%
Show the data table
| Reported measure | Reported relative association |
|---|---|
| Discovery: evidence | 20.3% |
| Shortlisting: concise sentences | 18.8% |
| Comparison: tables | 25.7% |
Method and limits: Values are reproduced from the public report page without deriving new estimates. AirOps reports 7,500 commercial queries and 217,508 retrieved pages, with 1,875 queries in each buying-stage group. It extracted 25 on-page signals and used logistic regression, random forest models, sensitivity variants and domain-authority-stratified modelling. This article interprets the public results; it does not rerun the models. The chart compares reported relative associations, not absolute citation rates.
Research period: Report published April 2026. Sample: See the methodology note for scope and denominator differences..
Data source: AirOps, From Retrieved to Cited: How Commercial Content Earns Citations in AI Search, 2026-04-01. Visualization recreated by AEO Operators. View the original report
Reported citation associations at different buying stages. Discovery: evidence: 20.3% ; Shortlisting: concise sentences: 18.8% ; Comparison: tables: 25.7%
What the research actually reveals
What helps early discovery? In the detailed findings, pages with 5 to 7 statistics had a 20.3% higher citation likelihood.
The likely content lesson is to support category explanations with relevant evidence, not to insert arbitrary numbers. The detailed figure is used here rather than the rounded headline.
What helps shortlisting? Pages averaging 10 words or less per sentence earned 18.8% more citations.
Short, clear explanations can make product differences easier to extract, but technical accuracy should not be sacrificed to a sentence-length target.
What helps comparison and validation? Comparison pages with 3 tables earned 25.7% more citations. Validation pages with 8 list sections earned up to 26.9% more citations.
These are different stage-specific associations, not competing estimates of a single treatment. The chart shows the first set of unqualified findings and omits the “up to” value to avoid making it look like a typical effect.
The common principle is decision usefulness. A comparison page needs explicit criteria and fair trade-offs; a compatibility page needs an unambiguous answer and supporting conditions.
The study does not establish that copying the observed counts will reproduce the reported uplift.
Implications for Series A to Series C SaaS companies
For Series A–C B2B SaaS companies, this supports auditing existing buying pages before commissioning more broad educational content.
Map each page to the buyer’s immediate job: understand a category, narrow options, compare products or confirm a detail. A single page can support several jobs, but its primary answer should remain obvious.
Product marketing should verify comparison criteria and commercial claims. Content owners should simplify presentation without removing necessary qualifications.
Success should include accurate answers and qualified commercial engagement, not only the number of citations.
Actionable recommendations
- Build a buying-stage map for the questions that matter commercially. Add traceable evidence to discovery explanations, concise differentiated descriptions to shortlist pages, accessible comparison tables to evaluation pages and direct factual answers to validation documentation. Only include statistics, images or pricing details when they serve the reader.
- Test substantive revisions on a defined group of pages and retain the previous versions for comparison. Record whether pages are retrieved as well as whether they are cited, where the available monitoring allows it. Avoid describing a citation association as a promised performance uplift. AEO Operators is a natural next step for translating these findings into a prioritised commercial-content audit and measurement plan.
Research limitations
AirOps is a software vendor with a commercial interest in content optimisation. Modelling and controls improve the analysis but do not make an observational study causal.
Stage-specific figures have different conditions and should not be added together. The public page does not supply every uncertainty estimate or collection date.
Some findings use “citation likelihood” and others “more citations”; this article preserves those distinctions rather than treating them as identical probabilities.
Original sources and further reading
- From Retrieved to Cited: How Commercial Content Earns Citations in AI Search
AirOps, Oshen Davidson, 1 April 2026. Section: Public report page. Vendor research.
The publisher’s public report page supplies every numerical finding used in this analysis.
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