Featured AI Citation Report, June-August 2026

Findings

• Featured Research · August 2026
• We analysed 405 GEO audits run on Featured between June and August 2026 — 22,881 real citations that AI search engines returned across 11,499 different domains, matched against Moz Domain Authority. Here is what those sources actually look like.
• 34.5% of the citations AI search returned pointed to a domain with Domain Authority under 40.
• More than a third of the citations AI search returned when answering buying questions pointed to domains with a Domain Authority under 40 — domains most brands have never thought to pitch.
• Key Definitions
• Domain — A website, identified by its address (for example, forbes.com or reddit.com). We count domains when we talk about how many websites appear in the study.
• Domain Authority (DA) — Moz’s 0–100 score for how much web authority a domain has. A third-party SEO metric, not something AI engines use themselves.
• Source — A domain that an AI engine draws information from when answering a question. The source is where the information comes from, not the answer itself.
• Citation — Each instance of a source appearing in an AI answer. A citation points to where information came from, not what the answer is. If forbes.com appears in 40 answers, that is 1 domain, 1 source, and 40 citations.
• Six findings
• 34.5% — A third of citations come from low-authority domains
• 34.5% of all 22,881 citations pointed to a domain scoring under DA 40, and 13.2% pointed to domains under DA 20.
• 53.9% — Most cited domains are small
• Counting each domain once, 53.9% of the 11,499 domains AI search drew from score under DA 40, and the median domain scores just 37. But counting each citation instead, the median rises to DA 56 because a small number of large domains are cited repeatedly. Most websites are small, but the websites that get cited repeatedly tend to be larger.
• 76% — There is no shortlist to get on
• 76% of the domains cited appeared exactly once across the entire study. The 100 most-cited domains account for under a quarter of all citations, and even the top 1,000 account for less than half. A fixed list of twenty big-name domains does not describe where AI search draws its answers from.
• 78.5% — Most brands do not appear in AI answers
• Of the 378 brands audited across 344 industries, 78.5% did not appear in the AI-generated answer to a single buying question in their category. Only 6.7% of the 1,983 questions tested included the brand in the answer at all.
• 92.6% — Your own domain is rarely a source
• 92.6% of the brands audited never had their own domain appear as a citation in an AI answer about their category. Content from the brands’ own websites accounted for just 1.7% of classified citations — competitor domains were cited 13× more often.
• 9.2% — YouTube and Reddit are the two biggest single sources
• YouTube and Reddit together account for 9.2% of all citations in the study (YouTube 5.5%, Reddit 3.7%). The ten most-cited user-generated platforms collectively account for 11.7% of all citations, with YouTube and Reddit alone making up 9.2%.
• The authority of a cited source
• The distribution is bimodal. There is a hard spike at the very top — 31.2% of citations go to DA 80+ giants — and a long, dense floor below DA 40 worth another 34.5%. The middle is thin, which is exactly where most brands’ owned content and mid-tier press coverage sits.
• Share of all 22,881 citations by the Domain Authority of the cited domain. Bars are scaled to the largest band.
• 34.5% of citations pointed to a domain under DA 40
• 13.2% went to a domain under DA 20
• 53.9% of the 11,499 distinct domains cited are under DA 40
• 37 median DA of a distinct cited domain
• There is no shortlist
• AI citation is extraordinarily long-tailed. 22,881 citations were spread across 11,499 different domains, and 76% of those domains were cited exactly once. Coverage climbs slowly, not sharply — there is no small set of domains that unlocks visibility.
• Top 1,000 domains (8.7%) → 45.1% of citations
• Top 100 → 24.8%
• Cumulative share of citations as domains are added in rank order. Even the top 1,000 domains — out of 11,499 — account for less than half of all citations.
• Where the citations came from
• Editorial coverage dominates, and brand-owned content is almost invisible at 1.7%. A brand’s own domain is the least likely thing to be cited about that brand — competitors are cited 13× more often.
• Based on the 10,176 citations our analysis classified by source type. This subset skews higher-authority than the full corpus, because the classification step selects notable sources.
• Method. 405 GEO audits completed on Featured between 2 June and 21 August 2026, covering 378 brands across 344 industries. Each audit tested roughly five real buying questions — 1,983 in total — and recorded the sources returned, producing 22,881 citations across 11,499 distinct domains. All audits were run by real users; no synthetic runs are included.
• Domain Authority. Moz Links API, queried at domain scope on 21 August 2026. The 1,000 most-cited domains were measured individually, covering 45.1% of all citations. The remaining 10,499 domains were estimated from a uniform random sample of 1,500. The estimate reproduces the known citation total to within 0.7%.
• Scope. Citation data reflects AI search engines that perform live web retrieval and return their sources; figures here are drawn from Perplexity. Models answering without web retrieval do not expose sources and are excluded from all citation figures. Domain Authority is a third-party Moz metric and is not used by AI engines themselves — it is applied here only to characterise the domains being cited.