AEOForgedAEOForged
Get started

Original research · measured 2026-07-27

Editorial Domains Capture 92% of AEO/GEO Citation Share, But Leading Cited Domain Flips 100% of the Time on Same-Day Repeats

Original research · 92 engine checks across 24 questions, measured on 2026-07-27 · Methodology · Dataset

Executive summary

On 2026-07-27, this study recorded 92 answer-engine checks across 24 category questions and 3 engines (ChatGPT, Gemini, Perplexity), capturing every cited source domain with no failed calls. The single most striking finding: in all 10 same-day variance-panel repeat groups, the leading cited domain changed every time — a 100% flip rate. Any practitioner assuming a stable, predictable citation leader in this category is working from a false premise.

Citation Landscape

Editorial content dominates the source economy by a wide margin. Of all citations recorded, 813 — representing 92% of the total — came from editorial domains. Social sources accounted for 3.5% (31 citations), YouTube for 1.8% (16 citations), docs for 1% (9 citations), Reddit for 0.9% (8 citations), review sites for 0.7% (6 citations), and local directories for 0.1% (1 citation). The category rewards structured, text-forward editorial publishing far more than any other format.

Domain Leaderboard

At the top of the leaderboard, linkedin.com and youtube.com tied at 16 citations each out of 92 checks (17.4%, 95% CI 11%–26.4%). Two editorial domains — aiclicks.io and stackmatix.com — also tied at 14 citations each (15.2%, 95% CI 9.3%–23.9%). medium.com followed at 12 citations (13%, 95% CI 7.6%–21.4%). The wide confidence intervals throughout this leaderboard reflect the genuine instability in citation behavior, not a measurement limitation.

Per-Engine Divergence

The three engines behaved in structurally different ways. ChatGPT returned sources in only 9 of its 24 checks, averaging 2 sources per answer — a low-density, highly selective citation posture. Its most-cited domain, citescopeai.com, appeared just twice. Gemini returned sources on all 24 checks, averaging 9.8 sources per answer; its top citation, tryprofound.com, appeared 8 times. Perplexity returned sources on all 44 checks, averaging 13.6 sources per answer and driving the bulk of volume; linkedin.com led there with 16 appearances, and aiclicks.io and stackmatix.com each appeared 13 times alongside youtube.com.

Because citation behavior differs so sharply across engines — in both volume and domain selection — aggregate leaderboard rankings reflect a blend of very different underlying distributions. No single domain appeared consistently dominant across all three engines.

Practical Implication

The 100% same-day flip rate on repeat queries means citation position in this category cannot be treated as a durable asset. Every question in the 24-question inventory produced at least one source on at least one engine, confirming the category is indexable, but the identity of the cited source is volatile. Editorial publishing appears to be the most reliable structural bet, given its 92% share of citations, though which specific editorial domains surface on any given query remains unpredictable under current engine behavior.

Key findings

  • The leading cited domain changed in all 10 of 10 same-day repeat groups, producing a 100% flip rate across the variance panel measured on 2026-07-27.
  • Editorial domains captured 813 of total citations, representing 92% of the source economy across 92 checks and 24 category questions.
  • linkedin.com and youtube.com tied for the most-cited domains at 16 citations each out of 92 checks (17.4%, 95% CI 11%–26.4%).
  • ChatGPT returned sources in only 9 of 24 checks averaging 2 sources per answer, while Perplexity returned sources in all 44 checks averaging 13.6 sources per answer.
  • aiclicks.io and stackmatix.com each appeared in 14 of 92 checks (15.2%, 95% CI 9.3%–23.9%), ranking third on the domain leaderboard.
  • Every one of the 24 category questions produced at least one cited source on at least one engine, confirming full category indexability despite high citation volatility.

Findings

Citation Leaderboard

Across 92 recorded engine checks conducted on 2026-07-27 — spanning 24 category questions and 3 engines (ChatGPT, Gemini, Perplexity) — two domains tied at the top of the citation leaderboard: linkedin.com and youtube.com, each cited in 16 of 92 checks (17.4%, 95% CI 11%–26.4%). Their identical citation counts and overlapping confidence intervals mean the data cannot distinguish between them on frequency alone.

The next tier — aiclicks.io and stackmatix.com — each appeared in 14 of 92 checks (15.2%, 95% CI 9.3%–23.9%), again indistinguishable from each other and only modestly separated from the top pair when intervals are considered. medium.com followed at 12 of 92 checks (13%, 95% CI 7.6%–21.4%).

Three domains clustered at 11 of 92 checks each (12%, 95% CI 6.8%–20.2%): blog.hubspot.com, discoveredlabs.com, and hubspot.com. The fact that two distinct HubSpot subdomains appear independently in this band is a structural observation about how engines resolve and attribute sources within a single publisher's property.

The next group — solcrys.com and tryprofound.com — each recorded 10 of 92 checks (10.9%, 95% CI 6%–18.9%). Four domains then clustered at 9 of 92 checks (9.8%, 95% CI 5.2%–17.6%): conductor.com, gen-optima.com, higoodie.com, and nicklafferty.com. Closing the top 15, aeoengine.ai appeared in 8 of 92 checks (8.7%, 95% CI 4.5%–16.2%).

The confidence intervals throughout the leaderboard are wide — a direct consequence of the 92-check sample size. Ranks 1 through 15 cannot be treated as a strict hierarchy; interval overlap is the rule, not the exception. What the leaderboard does establish is a clear divide: these 15 domains each cleared an 8.7% floor, while the remainder of the citation pool sits below it.

Source-Type Economy

The source-type breakdown reveals a heavily concentrated citation economy. Of 884 total citations recorded, editorial sources accounted for 813 citations (92%). No other source type comes close. Social sources — a category that includes linkedin.com and medium.com — contributed 31 citations (3.5%). YouTube contributed 16 citations (1.8%), matching the raw check-level appearance of youtube.com in the domain leaderboard. Docs sources added 9 citations (1%), Reddit 8 citations (0.9%), review sites 6 citations (0.7%), and local directory sources 1 citation (0.1%).

This 92% editorial concentration has a direct structural implication for the category being measured. Answer engines responding to questions about how to evaluate and trust AEO/GEO platforms, agencies, and citation-tracking tools are drawing almost exclusively from editorial content — articles, guides, and blog posts — rather than from review platforms, community discussion, or vendor documentation. Reddit, the one community-sourced type in the data, contributed less than 1% of citations overall, despite appearing as the most-cited single domain for the question "is AEO a scam" (reddit.com, 4 appearances in 5 checks).

Per-Question Concentration

At the individual question level, citation concentration around a single domain is frequently high. For questions where sources were returned, the leading domain often appeared in 3 or 4 of the 5 checks run: citeflow.io led "best tools to track ChatGPT citations" at 4 of 5 checks; airankchecker.net led "best tools to track Perplexity citations" at 4 of 5 checks; conductor.com led "AEO platforms for marketing agencies" at 4 of 5 checks; discoveredlabs.com led "how to vet an AEO or GEO vendor" at 4 of 5 checks; and capterra.com led "questions to ask before buying an AI visibility tool" at 4 of 5 checks. This pattern suggests that for specific, bounded questions in this category, engines frequently converge on a single leading source — while for broader or more contested questions, the leading domain appeared only once or twice across checks.

Engine behavior

Measurement date: 2026-07-27 | Sample: 92 recorded engine checks across 24 category questions and 3 engines (ChatGPT, Gemini, Perplexity); 0 of 92 planned calls failed.

Source Volume by Engine

The three engines differ substantially in how many sources they surface per answer. Perplexity was the most source-dense, completing all 44 checks with sources and averaging 13.6 sources per answer. Gemini completed all 24 checks with sources, averaging 9.8 sources per answer. ChatGPT returned sources in only 9 of its 24 checks, averaging 2 sources per answer across those that returned any. The remaining 15 ChatGPT checks returned zero sources — a measured negative, not a data gap. Every question produced at least one source on some engine, but ChatGPT's zero-source rate (15 of 24 checks) signals that it frequently answered without external citation in this category.

Domain Citation Patterns

Each engine drew from a distinct citation ecosystem. ChatGPT's top cited domain was citescopeai.com at 2 citations; aeo-platform.com, aeoengine.ai, amplefound.com, and any.run each appeared once. Given the low source volume, citation concentration here reflects scarcity as much as preference.

Gemini cited more broadly. Its leading domain, tryprofound.com, appeared 8 times across 24 checks. hubspot.com followed at 7 citations, medium.com at 6, discoveredlabs.com at 5, and frizerly.com at 4. The distribution suggests Gemini draws from a mix of category-specific properties and general marketing publishers.

Perplexity showed the widest citation footprint. linkedin.com led at 16 citations across 44 checks, followed by aiclicks.io and stackmatix.com at 13 each, youtube.com at 13, and blog.hubspot.com at 11. The presence of linkedin.com and youtube.com alongside category-specific domains indicates Perplexity pulls from platform-hosted content at meaningful scale in this topic area.

No domain appeared as a leading citation across all three engines simultaneously, which means there is no single universally favored source in the category as measured.

Test-Retest Flip Rate

The variance panel ran 10 same-day repeat groups to test answer stability. All 10 of 10 repeat groups changed their cited-domain set, producing a 100% flip rate. This means that for this category, no engine reproduced the identical citation set when the same question was asked again on the same day. Citation presence at any single moment should not be interpreted as stable placement. Measurement programs relying on single-point checks would systematically misrepresent citation frequency in this environment.

What this means

The citation economy measured on 2026-07-27 across 92 engine checks tells a clear structural story: editorial content owns this category almost entirely. Editorial sources account for 813 citations, or 92% of the total citation share. Every other source type — social (3.5%), YouTube (1.8%), docs (1%), Reddit (0.9%), review sites (0.7%), and local directories (0.1%) — together account for the remaining 8%. For teams working in the AEO/GEO evaluation and trust category, that distribution is the single most important finding to internalize.

The domain leaderboard reinforces this pattern. The two most-cited domains, linkedin.com and youtube.com, each appeared in 16 of 92 checks (17.4%), but neither is a specialist source in this category — they are general-purpose platforms that happen to host relevant content. The immediately trailing positions are held by editorial domains: aiclicks.io and stackmatix.com, each cited in 14 of 92 checks (15.2%). The top 15 list contains 12 editorial domains, 1 social platform, 1 video platform, and no review sites, no documentation hubs, and no vendor knowledge bases.

One structural observation worth noting: the leading cited domain sits at only 17.4% citation share. No single source has achieved anything close to dominant coverage across this 24-question category. The confidence intervals throughout the leaderboard are wide — the top domains carry 95% CIs spanning roughly 11% to 26% — indicating that citation leadership here is not yet consolidated. That is open ground.

For teams producing content in this category, the data suggests several directional priorities. Editorial formats — articles, guides, and analysis pieces hosted on indexed editorial domains — represent the source type answer engines actually draw from at scale. Social content and video content are cited, but at a fraction of the rate. Review sites, despite being intuitively relevant to a category about evaluating and trusting platforms and agencies, accounted for only 6 citations total (0.7%) across the full sample.

The absence of any dominant authority across all 24 questions suggests that comprehensive, question-level editorial coverage may be underserved. Teams that produce systematic editorial content mapped to the specific evaluation questions buyers ask in this category may improve their surface area for citation, though no outcome is guaranteed. The measurement captures a single day; citation patterns in emerging categories can shift as new content enters the index.

Methodology and data

This study was produced with a deterministic measurement instrument (92 planned engine calls, 92 recorded, 0 failed — recorded as negatives). The complete instrument description, sample, refusal conditions, and limitations are in the methodology; every underlying measurement is in the dataset.