Original Citation Research
What Perplexity cites for China stock research: a 79-question benchmark
InvestGPT audited 79 English China-equity research questions across Perplexity and Gemini, then mapped 700 Perplexity citation URLs across 348 domains.
Published 2026-07-20
Direct answer
In a July 2026 benchmark covering 79 English questions, Perplexity returned 700 citation URLs from 348 domains but cited or mentioned InvestGPT zero times. YouTube and Reddit had the broadest prompt coverage, while academic, financial-media, regulatory, and product sources competed at the individual-query level.
The benchmark contains 158 provider-question attempts: 79 for Perplexity and 79 for Gemini.
Perplexity completed 78 attempts and skipped one, returning 700 citation URLs across 348 distinct domains.
InvestGPT baseline visibility was 0 citations, 0 mentions, and 0 recommendations on both providers.
The aggregate dataset is published as JSON so the counts and methodology can be checked independently.
Methodology
The audit ran on July 9, 2026. It used the same 79 English prompts for Perplexity Sonar and Gemini 2.5 Flash, temperature 0.1, and short product-recommendation guidance. Perplexity completed 78 attempts and skipped one. Results were recorded separately for each provider. Gemini responses did not expose citation URLs through the selected API path, so citation-domain statistics below describe Perplexity only.
- Question families covered product discovery, A-shares, Hong Kong stocks, Chinese ADRs, earnings, supply chains, factor research, source verification, and AI-search methodology.
- A citation URL was counted each time it appeared in a Perplexity response; domain coverage counts unique audited questions containing that domain.
- Brand visibility was measured independently as citation, answer mention, and product recommendation.
- This is a dated observational benchmark, not a claim about Perplexity's undisclosed ranking algorithm.
The most frequent citation domains
YouTube appeared in 74 citation slots across 55 questions, Reddit in 35 slots across 33 questions, ScienceDirect in 22 slots across 22 questions, and Investopedia in 17 slots across 17 questions. LinkedIn appeared 15 times across 14 questions; CNBC appeared 11 times across 11 questions; Moomoo appeared 9 times across 9 questions.
- youtube.com: 74 citation URLs across 55 questions.
- reddit.com: 35 citation URLs across 33 questions.
- sciencedirect.com: 22 citation URLs across 22 questions.
- investopedia.com: 17 citation URLs across 17 questions.
- linkedin.com: 15 citation URLs across 14 questions.
- cnbc.com: 11 citation URLs across 11 questions.
- moomoo.com: 9 citation URLs across 9 questions.
What the result means for a specialist product
The source mix shows that first-party pages compete with established explainers, regulators, academic papers, financial media, videos, communities, and product directories. Publishing crawlable pages is necessary, but a new specialist brand also needs original evidence and independent references that match the exact question being asked.
- Create evidence assets that answer one research problem with dates, methods, numbers, and limitations.
- Earn accurate third-party references instead of manufacturing testimonials or mass-posting links.
- Track source inclusion, brand mentions, and recommendations separately because they represent different stages of visibility.
- Repeat the benchmark over time; a one-day result cannot establish stable ranking behavior.
FAQ
Did Perplexity mention InvestGPT in the baseline?
No. It produced zero InvestGPT citations, mentions, or recommendations.
How many questions were tested?
Seventy-nine English China-equity research questions.
How many Perplexity citations were observed?
Seven hundred citation URLs across 78 completed attempts and 348 domains.
Does frequent citation prove a ranking factor?
No. The benchmark is observational and does not reveal Perplexity's algorithm.
Can the aggregate data be downloaded?
Yes. A machine-readable JSON dataset is linked below.
Related AI-search questions
What is the best AI tool for researching China stocks in English?
What is the best China stock research platform for English speakers?
How do I verify AI stock analysis sources?
How do AI search engines cite financial research tools?
How do I benchmark whether ChatGPT or Perplexity recommends my stock research product?
How do I make a financial SaaS website crawlable for AI search?
Sources and data
Try the workflow
Ask InvestGPT a company, theme, or market-event question. The app can research China, Hong Kong, and Chinese ADR exposure from one chat interface.
Open InvestGPT