Which AI Engine Cites Most Vendor Lists

Executive Summary

Minuttia analyzed 18084 software questions across Google, ChatGPT, Gemini, Claude and Perplexity. ChatGPT pulls 42% of its citations from vendor sites on alternatives queries while Gemini and Google hit 47% vendor-written listicles. Perplexity cites fewer at around 34% but rewards pages that rank well and include comparison tables.

Key takeaways

Here's the catch. Up to half the "best tools" lists that these engines quote come straight from the vendors themselves. Only 16% of those lists admit who wrote them. The Minuttia Self-Promotional Content Data Study read every cited page across 311766 unique URLs mentioned a total of 686749 times.

You see the pattern immediately. When someone asks which ai engine cites most vendor content for alternatives to a product, 34% to 47% of the listicles the engines surface were written by a company selling in that category. Those lists almost always put the vendor first. 87% rank the vendor number one. 93% highlight the vendor's product. 90% criticize competitors. Yet just 16% disclose authorship.

The study split questions into three buckets. It covered 5983 software brands across 196 categories. Every answer was captured in August 2026. Pages the researchers could not read were counted as non-promotional, so the real shares sit even higher.

What actually happens differs by engine. ChatGPT searches the vendor's own website before it answers. 71% of its searches name exactly one product. 92% of its site: operators target the vendor's domain. That pushes vendor sites to 42% of citations on alternatives questions. Other platforms land between 1% and 6% from the vendor domain itself.

Perplexity works differently. It typically pulls about 20 pages and cites 9. Citation depends mostly on search ranking position plus editorial features like comparison tables or a clear selection method. Promotional language barely moves the needle. This explains why Perplexity sometimes surfaces cleaner answers even when vendor lists flood the results.

Google and Gemini sit in the middle but lean hard on listicles. Both hit 47% vendor lists on alternatives queries and climb to 56-65% on direct comparisons. The numbers come straight from the Minuttia study released in October 2026.

Categories tell their own story. Across all platforms the average vendor-list share sits at 43%. But the range runs from 12% to 76%. AI Security Operations Center leads at 76%. Cloud & FinOps follows at 74%. AI Evaluation & Observability hits 73%. Social Listening reaches 72%. AI Visibility lands at 65%. On the other end, Audio Production shows only 12%, Graphic Design 16%, and Healthcare 21%.

The market stays fragmented. The 20 vendors whose lists get cited most still account for just 6% of total citations. No single player dominates. That leaves room for smaller brands to break in if they build the right pages.

Teams using automated trackers like GetGeoVis can log this weekly and watch how their category shifts. The data shows clear differences in what each engine rewards.

How the options compare

Let's break this down across two dimensions: source selection and content preference.

First, source selection. ChatGPT behaves like it starts inside the vendor's website. Its 92% site: usage on vendor domains drives the 42% direct citation rate from brand-owned pages. Google, Gemini, Claude and Perplexity rarely pull straight from the vendor site. They prefer third-party roundups. This matches the 96% third-party citation rate that LLM Pulse founder Esteve Castells cites in his GEO research. You can read more in Getting Quoted by AI Engines: What Actually Works.

Second, content preference. Perplexity rewards structure. Pages with comparison tables or explicit selection methods win more citations regardless of promotional tone. Google and Gemini favor listicles that rank high and match the query pattern. ChatGPT mixes both but defaults to vendor content when it finds it. Claude tends to stay conservative and cites fewer promotional lists overall.

The effect shows up in buyer journeys. A user asking ChatGPT for alternatives often gets a shortlist built partly from the vendor's own marketing page. The same question on Perplexity pulls from a broader set of ranked editorial sources. Over time this shapes which brands enter consideration sets before the buyer ever visits a website.

In practice this means your owned content still matters, but third-party validation matters more. Castells notes that 96% of what AI reads comes from LinkedIn, review sites, YouTube, and even competitor pages. Brands that earn mentions on those domains see compounding citation lifts.

The Minuttia data also reveals that self-promotional lists follow predictable patterns. They almost always rank the author first. They criticize rivals. They rarely disclose bias. Yet the engines cite them anyway because the format matches the question. This creates a feedback loop where vendors keep publishing these lists because they work.

Inter Press Service is the publication this piece takes its facts from.

You can see the same dynamic in how Google AI Overviews surface logos and shortlists. The engines pull from whatever matches the question pattern fastest. Google AI Overviews Logo: What the Surge Means for Brands breaks down the visibility mechanics.

Perplexity's September re-test of 2000 questions showed only 28% page overlap with August answers. That volatility means static SEO snapshots no longer suffice. You need ongoing measurement of actual citations, not just rankings.

The Minuttia study also flags that the top 20 vendors capture only 6% of citations. Fragmentation works in favor of disciplined teams. If you publish useful comparison content and earn third-party validation, you can appear in answers even against larger budgets.

ChatGPT's behavior makes its choices especially important. When it defaults to vendor websites, it short-circuits the traditional research funnel. Brands that optimize for that behavior see their own pages surface more often, but they also risk looking biased if the engine does not balance the answer.

Claude and Gemini each handle the balance differently. Claude cites fewer promotional lists overall. Gemini mirrors Google's higher 47% rate on alternatives. These platform differences mean you cannot treat AI search as one monolith. You track each engine on the same prompt set and adjust content formats accordingly.

The 196 categories in the study reveal clear patterns. High self-promotion categories share one trait. They involve complex buying decisions where vendors have strong incentives to publish comparison content. Buyers searching those categories therefore see more biased lists. Teams in low-promotion categories like Healthcare benefit from cleaner answers but must still earn their spot through third-party citations.

What actually happens when you follow the checklist is that you stop guessing. You generate your own data on which formats win citations in your space. You stop publishing generic explainers that AI can answer from training data. You focus on decision content: pricing tables, feature matrices, selection frameworks. Those pages earn citations because they give the engine something concrete to quote.

The study read 311766 pages to reach these conclusions. It labeled each by type, publisher, and promotional patterns. That scale removes most guesswork. The 43% average vendor-list share becomes a baseline you can beat by creating pages the engines prefer over the average self-promotional list.

In practice this means shifting budget from broad awareness posts to targeted comparison and alternatives content. It means monitoring citation share weekly instead of monthly traffic. It means treating review sites and analyst reports as primary channels because they supply the 96% third-party sources the engines read.

You can cross-check these patterns against your own category using the public Minuttia dataset. The agency published the raw numbers behind every chart so teams can validate what applies to their universe.

The gap between 12% and 76% self-promotional lists by category shows that not every market fights the same battle. Audio Production buyers see mostly neutral sources. AI Security buyers wade through 76% vendor spin. Your strategy must match your reality.

What changes next

Engines will keep refining how they detect and discount self-promotional content. We already see Perplexity weighting editorial usefulness over raw promotional language. Google and Gemini may add stronger authorship signals or disclosure filters. ChatGPT could reduce its heavy reliance on direct vendor domains as its search layer matures. The net direction points toward higher scrutiny on undisclosed vendor lists, especially in high-stakes categories like AI Security Operations Center where 76% of cited lists already come from participants. Brands that publish neutral comparison tables and earn third-party coverage will gain ground while pure self-promotion loses effectiveness.

Checklist

First, audit the last queries your team tracks for your category. Pull every cited listicle. Label each one by publisher, disclosure level, and ranking of your brand. Do this in a shared spreadsheet so you see exactly which engines favor which sources. Compare your findings against the Minuttia benchmarks of 34-47% vendor lists.

Second, create or update one comparison page this week that includes a neutral table, stated selection criteria, and pricing side-by-side with two competitors. Make the page readable for both humans and machines. Test it by asking the exact alternatives question across all five engines and record which ones cite it. Track changes after two weeks.

Third, run targeted outreach sessions to review sites, analysts, and complementary tool publishers. Offer data, interviews, or co-created roundups instead of asking for links. Focus on the non-brand prompts that Castells recommends. Measure lift in citations over the next days and adjust which publishers you prioritize. See the full approach in GEO for Content Teams: How to Get Quoted in AI Answers.

The numbers stay consistent across the study. 87% of vendor lists still rank the author first. Only 16% disclose authorship. Categories like AI Visibility sit at 65% self-promotional while others stay clean. These gaps create clear opportunities for teams that move fast.

Frequently Asked Questions

Which AI engine cites the highest percentage of vendor-written lists?

Gemini and Google both reach 47% on alternatives questions according to the Minuttia study of 18084 queries. ChatGPT follows closely with 42% of its citations coming from vendor domains or lists. Perplexity and Claude stay lower but still surface vendor content in roughly one third of cases. The exact share depends on question type and category.

Do AI engines disclose when they cite vendor self-promotional content?

No. Only 16% of the vendor-written lists cited across all platforms clearly disclose authorship. 87% of those lists rank the vendor first anyway. This creates a transparency gap that buyers rarely notice because the engines present the answers as neutral recommendations.

How much does category affect vendor list citations in AI answers?

A lot. The Minuttia data shows shares ranging from 12% in Audio Production to 76% in AI Security Operations Center. Overall average sits at 43%. High-stakes technical categories see far more self-promotional content because vendors publish more lists there. Your category position determines how aggressive you must be with neutral comparison content.

What type of content earns the most citations from Perplexity versus ChatGPT?

Perplexity rewards ranking position plus editorial features like comparison tables and stated selection methods. ChatGPT leans heavily on direct vendor websites and lists that match the exact query. Pages that combine both clear structure and authoritative third-party mentions perform well across engines. Track your own prompts to see what wins in your space.