# AI Citation Rates by Industry: Why Your Vertical Determines Which Sources Answer Engines Trust

Three independent studies covering 14,400+ observations and 143,000+ citations reveal that AI answer engines cite dramatically different source types depending on the industry. Healthcare concentrates 71% of citations in ten domains. SaaS spreads citations across thousands. Your vertical dictates the Machine Relations playbook.

Canonical URL: https://machinerelations.ai/research/ai-citation-rates-by-industry-vertical-source-data-2026
Published: 2026-07-28
Research type: Research Synthesis
Tags: machine-relations, ai-search, citations, ai-visibility, industry-data, citation-rates, vertical-analysis, mri

## Source Body

AI answer engines do not cite equally across industries. Three independent measurement studies — covering 14,400 prompt-engine observations, 143,424 tracked citations, and 38,000 sector-tagged prompts — converge on the same finding: the source types AI engines rely on, the number of citations per answer, and the competitive concentration of those citations shift dramatically depending on the vertical ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026); [Orbator, 2026](https://dev.to/henrik_telle/we-tracked-143424-ai-citations-nearly-half-point-back-to-the-vendors-being-recommended-3g6f); [Presenc AI, 2026](https://presenc.ai/research/ai-citation-statistics-by-industry)).

A brand running the same AI visibility playbook across verticals is optimizing for an average that does not describe any actual industry. Separate research from Semrush and Kevin Indig across 3,981 domain appearances found that citation behavior itself differs by query type — informational queries produce 89.3% citation rates but only 18% brand mention rates, while comparative queries produce 43.3% mention rates ([Semrush, 2026](https://semrush.com/blog/the-ghost-citations-study)). The vertical determines not just whether you get cited, but how.

## Citation Density Varies 2x Across Verticals

Not every industry gets the same number of citations per AI answer. Across four engines (ChatGPT, Claude, Gemini, Perplexity), the blended citation density ranges from 2.4 to 5.1 citations per answer depending on the vertical ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

| Vertical | Perplexity | Claude | ChatGPT | Gemini | Blended Average |
|---|---|---|---|---|---|
| SaaS | 7.4 | 4.6 | 4.1 | 3.2 | 5.1 |
| Legal | 6.8 | 4.3 | 3.7 | 3.0 | 4.7 |
| Fintech | 6.7 | 4.2 | 3.6 | 2.9 | 4.6 |
| Consumer Electronics | 6.5 | 4.1 | 3.5 | 2.8 | 4.5 |
| B2B Services | 6.4 | 3.9 | 3.3 | 2.6 | 4.3 |
| Insurance | 6.2 | 3.7 | 3.1 | 2.4 | 4.0 |
| Education | 6.1 | 3.6 | 2.9 | 2.3 | 3.9 |
| Real Estate | 5.8 | 3.4 | 2.8 | 2.2 | 3.6 |
| Travel | 5.6 | 3.3 | 2.7 | 2.1 | 3.5 |
| DTC Apparel | 5.4 | 3.1 | 2.6 | 2.0 | 3.3 |
| Food & Beverage | 5.1 | 2.9 | 2.4 | 1.8 | 3.1 |
| Healthcare | 4.6 | 2.5 | 2.0 | 1.5 | 2.4 |

SaaS gets more than twice the citation slots per answer compared to healthcare. That gap creates a fundamentally different competitive dynamic: SaaS brands compete for more available slots against more distributed sources, while healthcare brands compete for fewer slots that are already locked by institutional incumbents.

Perplexity consistently cites 2.1x more sources than ChatGPT and 2.7x more than Gemini across all verticals, meaning the engine a buyer uses changes the citation opportunity as much as the industry itself ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

## Healthcare: 71% of Citations Concentrate in Ten Domains

Healthcare is the most concentrated vertical in AI citation data. The top ten cited domains capture 71.2% of all healthcare citation slots, compared to 28.4% in SaaS ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). The gatekeepers are institutional: NIH commands 14.7% of all healthcare citations, Mayo Clinic takes 12.3%, and the CDC holds 9.8% ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

Academic and government sources account for 24.1% of healthcare citations — the highest of any vertical measured. Vendor sites, by contrast, hold only 5.2% ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). A separate study by Presenc AI found similar patterns: healthcare leads all 14 measured industries in average citations per answer (6.1) and draws the highest share from news and editorial sources (27%) ([Presenc AI, 2026](https://presenc.ai/research/ai-citation-statistics-by-industry)).

Separately, academic research presented at ICML found that in Google AI Overviews, AI-generated documents are cited more frequently than human-authored documents even after controlling for retrieval rank, with the effect most pronounced at highly ranked positions ([Kakimov et al., 2026](https://proceedings.mlr.press/v318/kakimov26a.html)). In a vertical like healthcare, where institutional credibility determines citation selection, this retrieval bias compounds the incumbent advantage.

For healthcare brands, the implication is structural: AI engines defer to institutional authority. Earned placements in NIH-indexed journals, CDC-cited research, or Mayo Clinic-level editorial carry disproportionate weight. Vendor content strategies that work in SaaS have almost no traction here.

## SaaS and B2B: The Most Open Citation Markets

SaaS citation patterns are the inverse of healthcare. Top-ten domain concentration is 28.4% — the lowest of any vertical — meaning citations spread across thousands of smaller domains ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). The top cited domain in SaaS is G2 at 9.4%, followed by Reddit at 6.8%, Capterra at 5.1%, HubSpot at 3.9%, and TrustRadius at 3.4% ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

Editorial reviews account for 24.1% of SaaS citations, the highest editorial share of any vertical. Vendor sites take 21.3% — also the highest vendor share ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). This confirms what Machine Relations Index data shows about G2's citation authority: G2 is cited across six AI engines, 35 distinct queries, and 10 verticals, with 145 total citations in the last 30 days alone.

Presenc AI's independent data confirms the pattern: SaaS and B2B lead in average citations per answer (5.8) and carry the highest review/comparison source share (34%) across 38,000 sector-tagged prompts ([Presenc AI, 2026](https://presenc.ai/research/ai-citation-statistics-by-industry)).

For SaaS companies, the Machine Relations strategy is clear: third-party review platforms and comparison content are the primary citation surface. Brand-owned comparison pages, customer reviews on G2 and Capterra, and Reddit presence are measurably more valuable than institutional or academic sources that dominate healthcare.

## Legal and Fintech: Where Academic and Regulatory Sources Win

Legal citations are structurally distinct. Academic and government sources command 18.4% of legal citation slots — second only to healthcare — with Cornell Law (7.4%), Nolo (6.3%), FindLaw (5.8%), the SEC (4.9%), and Wikipedia (4.6%) forming the top five ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). Reddit, which dominates consumer electronics and food, accounts for only 1.4% of legal citations.

Fintech follows a related pattern. NerdWallet (11.2%) and Investopedia (9.7%) together capture more than 20% of all fintech citation slots. Academic and government sources hold 5.9%, and editorial reviews hold 31.7% — the highest editorial share of any vertical ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). Presenc AI confirms that fintech draws heavily from news and editorial (22% each) rather than UGC (5%) ([Presenc AI, 2026](https://presenc.ai/research/ai-citation-statistics-by-industry)).

For companies in regulated industries, AI engines reward authoritative, compliance-adjacent content. Placements in NerdWallet, Investopedia, or legal reference publishers carry citation weight that vendor blog posts cannot match.

## Travel and Consumer Electronics: Reddit and UGC Shape AI Answers

ChatGPT includes citations in 22.6% of travel answers — 3.3 times its overall average of 6.8%. Travel is the most citation-dense vertical for ChatGPT specifically, according to Similarweb's 2026 Generative AI Landscape Report covering U.S. desktop usage ([Search Engine Journal, 2026](https://searchenginejournal.com/chatgpt-links-out-most-on-travel-questions/583824)). BrightEdge's research found that 57.1% of Google AI Overview sources already originate outside Google's top 10 organic results ([BrightEdge, 2025](https://www.brightedge.com/resources/research-reports/how-ai-search-is-reshaping-visibility)), which means the citation surface extends well beyond traditional SEO winners.

The source mix explains why. Travel citations are dominated by reviews and user-generated content: 54.1% of ChatGPT's top 10,000 travel citations come from reviews and UGC sources ([Search Engine Journal, 2026](https://searchenginejournal.com/chatgpt-links-out-most-on-travel-questions/583824)). In the Attrifast study, Reddit alone holds 9.6% of travel citation slots, followed by TripAdvisor (6.4%), NomadList (4.7%), and YouTube (4.1%) ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

Consumer electronics is even more Reddit-dependent: 23.1% of citation slots go to Reddit, with Rtings (7.8%), Wirecutter (6.9%), YouTube (5.4%), and Tom's Hardware (3.6%) rounding out the top five ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

Travel and consumer electronics brands face a Machine Relations environment shaped by community trust. Reddit threads and YouTube reviews are not peripheral content — they are the primary citation surface AI engines rely on. A brand absent from these platforms is absent from the answers.

## Perplexity Cites 2x More Than ChatGPT — Across Every Vertical

Engine-level differences compound the vertical effect. Perplexity's median citation count is 6.4 per answer, compared to 3.6 for Claude, 3.1 for ChatGPT, and 2.4 for Gemini ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

Year-over-year citation growth is also uneven across engines. Claude's citation density grew 50% from May 2025 to May 2026 (2.4 to 3.6), while ChatGPT grew only 14.8% (2.7 to 3.1). Perplexity grew 30.6% and Gemini grew 33.3% ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

Run-to-run reproducibility also varies. Perplexity has the highest citation stability (mean Jaccard overlap of 0.67, with 51% of citations appearing in every run), while Claude has the lowest (0.49 overlap, 33% stable) ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)). This means a brand's citation presence in Claude is more volatile — appearing in one answer and disappearing in the next — while Perplexity citation positions are stickier once earned.

## Being Cited Does Not Mean Being Recommended

A critical distinction emerges from industry-level citation data: citation and recommendation are separate signals. Research by SolCrys across 13,739 citations and 1,534 domains found that the most-cited sources (Reddit, Wikipedia, TechRadar) do not correspond to the most-recommended brands ([SolCrys, 2026](https://solcrys.com/ai-citation-vs-recommendation)). Lily Ray's study of 100 B2B "best software" queries confirmed the gap: when AI Overviews cited a brand's self-promotional listicle, that brand was excluded from the actual recommendation 69% of the time ([SolCrys, 2026](https://solcrys.com/ai-citation-vs-recommendation)).

Semrush's ghost citation study quantified a related pattern across 115 prompts and 14 countries: 61.7% of AI citations are "ghost citations" — the source is linked but the brand name never appears in the answer text ([Semrush, 2026](https://semrush.com/blog/the-ghost-citations-study)). ChatGPT shows this most starkly, with an 87% citation rate but only a 20.7% mention rate. Gemini inverts the pattern: 21.4% citation rate but 83.7% mention rate ([Semrush, 2026](https://semrush.com/blog/the-ghost-citations-study)).

For Machine Relations measurement, this means citation count alone underestimates visibility in some engines and overestimates it in others. A brand with high citation rates in ChatGPT may be invisible to users who never click the source links. A brand frequently mentioned by Gemini without formal citations has high visibility but no measurable link equity.

## 46% of AI Citations Point Back to Recommended Vendors

Across 143,424 AI citations tracked by Orbator over 28 days, 46.2% resolve to vendor-controlled domains rather than third-party reviews, Reddit threads, or analyst reports ([Orbator, 2026](https://dev.to/henrik_telle/we-tracked-143424-ai-citations-nearly-half-point-back-to-the-vendors-being-recommended-3g6f)). The tracked engines — ChatGPT, Claude, Gemini, Perplexity, and Grok — showed that vendor content dominates citations because it tends to be the most detailed, frequently updated, and search-optimized.

The study measured 271 software categories weekly and found that only 16% of the time do all five engines agree on the same recommendation, creating significant inconsistency for buyers consulting different AI assistants ([Orbator, 2026](https://dev.to/henrik_telle/we-tracked-143424-ai-citations-nearly-half-point-back-to-the-vendors-being-recommended-3g6f)).

Attrifast's data confirms the pattern: on branded queries, vendor sites hold 27.4% of citation slots and earn the top citation position in 41% of answers, compared to 11.7% vendor share and 8% top-position rate on non-branded queries ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

This is not manipulation — it is structural. AI engines cite what they can verify, and vendor sites often contain the most complete, current product information. But the effect is that brands with strong owned content have a measurable citation advantage, especially in SaaS and B2B verticals where vendor share already reaches 21.3%.

## Citation Source Shifts Are Reshaping the Competitive Landscape

The source types AI engines rely on are changing. From May 2025 to May 2026, Reddit's share of all AI citations grew by 3.6 percentage points (7.8% to 11.4%), while editorial reviews fell by 2.4 points (24.7% to 22.3%) and long-tail blogs declined by 2.8 points (22.1% to 19.3%) ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

| Source Type | May 2025 | May 2026 | Change |
|---|---|---|---|
| Editorial Reviews | 24.7% | 22.3% | -2.4 pp |
| Long-tail Blogs | 22.1% | 19.3% | -2.8 pp |
| Vendor Sites | 13.4% | 14.7% | +1.3 pp |
| Reddit | 7.8% | 11.4% | +3.6 pp |
| News/Press | 8.9% | 9.1% | +0.2 pp |
| Wikipedia | 9.1% | 8.9% | -0.2 pp |
| Forums/Q&A | 7.1% | 7.8% | +0.7 pp |
| Academic/Government | 6.9% | 6.5% | -0.4 pp |

Reddit's rise is not uniform across verticals. In consumer electronics, Reddit already commands 23.1% of citations. In legal, it holds just 1.4%. The practical consequence: a brand investing in Reddit community presence will see disproportionate citation returns in consumer verticals and near-zero returns in legal or insurance.

Google's own AI surfaces show similar vertical-specific patterns. Google AI Mode self-cites its own properties in approximately 17% of e-commerce answers, forcing brands to compete against Google's own product surfaces for citation visibility ([SEONIB, 2026](https://seonib.com/blogs/google-ai-mode-self-citation-17-percent-ecommerce-seo-7-strategies)). Cicero Studio's analysis confirmed that Google AI Mode references its own content in roughly one in six answers across commercial queries ([Cicero Studio, 2026](https://cicero.studio/en/blog/google-ai-mode-self-referencing-seo)). This self-citation pattern does not affect all verticals equally — commercial and e-commerce queries trigger it far more than informational or educational queries.

## What Vertical Citation Patterns Mean for Machine Relations

These findings confirm that Machine Relations is not a single optimization problem but a vertical-specific discipline. The data supports five structural conclusions:

**1. Citation slot count determines competitive intensity.** SaaS brands compete for 5.1 citation slots per answer; healthcare brands compete for 2.4. More slots means more opportunity for challenger brands to earn citations alongside incumbents.

**2. Source concentration determines strategy type.** In healthcare (71.2% top-ten concentration), the strategy is displacement — earning authority in the same institutional surfaces AI engines already trust. In SaaS (28.4%), the strategy is distribution — maintaining presence across dozens of surfaces because no single domain dominates.

**3. Source type determines channel priority.** Healthcare brands need academic and government citations. SaaS brands need editorial reviews. Travel brands need Reddit and UGC. Fintech brands need financial publisher placements. The channel allocation should follow the vertical's source-type distribution, not a generic best-practice list.

**4. Engine selection changes the math.** Perplexity gives brands 2.7x more citation opportunities than Gemini. Claude's citation slate is 36% volatile, meaning citation positions turn over faster. Brands serving audiences that prefer specific engines need to weight their strategy toward those engines' citation patterns.

**5. Vendor content is a legitimate citation surface — in the right verticals.** In SaaS, vendor sites earn 21.3% of citations naturally. In healthcare, they earn 5.2%. The Machine Relations response should match: invest in owned comparison content for SaaS, invest in earned institutional placements for healthcare.

## Methodology

This analysis synthesizes three independent studies:

**Attrifast (2026):** 1,200 buyer-intent prompts across 12 verticals, run 3 times each across ChatGPT, Claude, Gemini, and Perplexity (14,400 observations total), generating approximately 51,723 citation events across 8,917 unique domains. Measurement window: April 12 to May 14, 2026.

**Orbator (2026):** 143,424 citations tracked across ChatGPT, Claude, Gemini, Perplexity, and Grok over a trailing 28-day window. 271 software categories measured weekly with per-category citation shares and 95% Wilson confidence intervals. Published under CC BY 4.0.

**Presenc AI (2026):** 38,000 sector-tagged prompts across 14 industries, measured monthly via ChatGPT, Perplexity, Gemini AI Overviews, and Claude. Hybrid model and human review. Published quarterly.

**Similarweb (2026):** ChatGPT citation rates by category from the 2026 Generative AI Landscape Report, covering U.S. desktop usage. Reported via [Search Engine Journal](https://searchenginejournal.com/chatgpt-links-out-most-on-travel-questions/583824).

**Semrush and Kevin Indig (2026):** 3,981 domain appearances across 115 prompts in 14 countries, measuring citation and mention behavior separately across ChatGPT, Gemini, and Google AI Mode ([Semrush, 2026](https://semrush.com/blog/the-ghost-citations-study)).

**SolCrys and Lily Ray (2026):** 13,739 citations across 1,534 domains over a 7-day window, plus Lily Ray's study of 100 B2B "best software" queries across three measurement dates ([SolCrys, 2026](https://solcrys.com/ai-citation-vs-recommendation)).

**Kakimov et al. (2026):** Observational framework for auditing citation behavior in AI-generated search summaries, applied to Google AI Overviews using the MS MARCO Web Search dataset. Presented at ICML ([Kakimov et al., 2026](https://proceedings.mlr.press/v318/kakimov26a.html)).

Machine Relations Index data referenced for G2 citation authority metrics uses the MRI v2.0 methodology, which reports citation rates per source-segment pair with confidence grades (A, B, C, or collecting) based on evidence accumulation across six engines.

*Last updated: July 28, 2026*

## FAQ

### Which industry has the highest AI citation rate?

SaaS has the highest blended citation density at 5.1 citations per answer across four engines, according to Attrifast's 14,400-observation study. Healthcare has the fewest at 2.4 citations per answer, but those citations concentrate heavily in institutional sources like NIH and Mayo Clinic ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

### Why does ChatGPT cite travel content more than other verticals?

ChatGPT includes source citations in 22.6% of travel answers compared to a 6.8% overall average, according to Similarweb's 2026 Generative AI Landscape Report. Travel queries tend to be decision-heavy and location-specific, which triggers more citation behavior. Reviews and UGC account for 54.1% of travel citations ([Search Engine Journal, 2026](https://searchenginejournal.com/chatgpt-links-out-most-on-travel-questions/583824)).

### Do AI engines agree on which sources to cite?

Rarely. Across 271 software categories, all five engines (ChatGPT, Claude, Gemini, Perplexity, Grok) agree on the same recommendation only 16% of the time ([Orbator, 2026](https://dev.to/henrik_telle/we-tracked-143424-ai-citations-nearly-half-point-back-to-the-vendors-being-recommended-3g6f)). Citation reproducibility also varies: Perplexity shows the highest run-to-run stability (0.67 Jaccard overlap), while Claude shows the lowest (0.49) ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

### How much of AI citation share goes to vendor-owned content?

46.2% of 143,424 tracked AI citations resolved to vendor-controlled domains, according to Orbator's 28-day study across five engines. The share is higher for SaaS (21.3%) than healthcare (5.2%) at the vertical level. On branded queries specifically, vendor sites earn the top citation position in 41% of answers ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026); [Orbator, 2026](https://dev.to/henrik_telle/we-tracked-143424-ai-citations-nearly-half-point-back-to-the-vendors-being-recommended-3g6f)).

### Is Reddit becoming more important for AI citations?

Reddit's share of all AI citations grew from 7.8% to 11.4% between May 2025 and May 2026, the largest increase of any source type. In consumer electronics, Reddit already holds 23.1% of citation slots. The effect is vertical-specific: Reddit holds 14.6% in travel but only 1.4% in legal ([Attrifast, 2026](https://attrifast.com/blog/ai-search-citations-by-vertical-2026)).

## Attribution

This research is published by Machine Relations Research, the research program of machinerelations.ai — the public research and standards initiative that publishes the glossary, research, evidence, and measurements for the Machine Relations discipline. Provenance and editorial standards: https://machinerelations.ai/about

## Machine-readable related links

### Related concepts

- [Machine Relations Index (MRI)](https://machinerelations.ai/glossary/machine-relations-index)
- [Machine Relations (MR)](https://machinerelations.ai/glossary/machine-relations)
- [AI Visibility](https://machinerelations.ai/glossary/ai-visibility)
- [AI Citations](https://machinerelations.ai/glossary/ai-citations)

### Supporting research

- [Earned Media vs. Owned Content: AI Citation Rates and Top Sources Ranked (2026)](https://machinerelations.ai/research/earned-vs-owned-ai-citation-rates-2026)
- [AI Search Citation Factors: What Determines Which Brands AI Engines Cite (2026 Data)](https://machinerelations.ai/research/ai-search-citation-factors-2026)
- [AI Citation Behavior Across Models: Why One AI Visibility Strategy Fails Across Gemini, Claude, Perplexity, and SearchGPT](https://machinerelations.ai/research/ai-citation-behavior-across-models-2026)
- [Citation Absorption vs Citation Selection: Why Getting Cited Is Not the Same as Getting Used](https://machinerelations.ai/research/citation-absorption-vs-selection-ai-search-2026)

### Framework context

- [Machine Relations Index](https://machinerelations.ai/index)
- [Machine Relations Stack](https://machinerelations.ai/stack)
- [Evidence Base](https://machinerelations.ai/evidence)