Gist Research
Explore how AI recommends brands, finds creators and cites published work.
New from Gist Research · CRM
A CRM study on citations, recommendations and the questions that change the answer.
Perplexity · B2B sales question
28 of 28 saved answers showed this separation. Selected question; study limits apply.
Featured study · October 6, 2026
For a question about cleansers, ChatGPT and Perplexity both recommended Vanicream first. They linked to entirely different websites.
Read the studyResearch and evidence
Full category reports and focused research features, with their question panels, collection dates and evidence limits.
A linked source and a fully supported answer are different outcomes. Two research papers show how to measure the gap without treating a citation count as an accuracy score.
Read the articleRESEARCH · 2024 STUDYThe study behind GEO's “up to 40%” visibility result tested how supplied sources appeared in an AI answer. Its design helps explain what that finding can support when you plan work on a website.
Read the articleRESEARCH · OCTOBER 14 COLLECTIONEight saved ChatGPT API answers show why the question, recorded searches, returned sources and final citations need separate reading.
Read the articleRESEARCH · OCTOBER 14 AND 19 COLLECTIONSSaved answers discussed both approving website changes and controlling internal AI agents. Counting every named tool as a marketing competitor would miss that distinction.
Read the articleRESEARCH · OCTOBER 29 DIAGNOSTICIn one Gist diagnostic, 540 AI Overview collection attempts produced 364 usable saved answer records. Those records contained 205 distinct extracted text strings within question groups. Each count answers a different question.
Read the articleRESEARCH · OCTOBER 29 DIAGNOSTICIn Gist's October 29 self-audit, explicit lookup questions often produced an owned link. Only two saved answers to the 15 unbranded questions included an owned link. Both answered the same question, and both added claims the cited research does not establish.
Read the articleESPRESSO RESEARCH · OCTOBER 2026Breville leads the recommendation review. Nespresso’s presence differs across broad, pod and budget questions. Explore findings from 3,933 saved AI answers.
Read the highlights and get the full reportFULL CATEGORY REPORTHow AI answers compare skincare brands, products and routines. Read the full category study, with repeated-answer comparisons, Google AI Overview findings and the evidence behind them.
Read the report and download the PDFFULL CATEGORY REPORTHow AI answers handle costume costs, retailer choices, decoration and reuse. Read the full study of 1,168 ChatGPT and Perplexity API answers to 30 questions, with Google coverage assessed separately.
Read the report and download the PDFRESEARCH FEATUREA costume shopper may need help deciding what to rent, what to assemble and what to do with it afterward. Those questions bring different businesses—and sometimes no store at all—into AI advice.
Read the articleRESEARCH FEATUREA retailer can be offered as a place to shop without receiving a source link. Another can be cited while the answer never suggests buying from it.
Read the articleRESEARCH FEATUREAveeno, Thayers, Innisfree and other names outside Gist’s tracked set still entered the saved product comparisons.
Read the articleRESEARCH FEATUREMissing product names, repeated sunscreen lists and a persistent platform-source pattern tell three different stories.
Read the articleCHATGPT RESEARCH REPORTA ChatGPT-only study of 344 saved answers to 30 authored questions, separating car-care providers and resources from used-car buying.
Read the reportCHATGPT RESEARCH REPORTA ChatGPT-only study of 280 saved answers to 30 authored questions about search, agents, representation, selling and research.
Read the reportMore research
Research · October 6, 2026
Perplexity usually named more product families for one cleanser question. The first brand stayed the same.
Read the studyMethods · October pilot
See the questions we asked and how we compared the answers.
Read the pilot methodsOur methodology
A mention, a citation and a visit describe different things. Our research keeps the question set, counting rules and comparison limits beside the findings.
Read our methodologyMore from Gist Research
What AI says about a sponsor matters before and after a creator partnership. Explore the questions to ask.
Read the studyChoose which optional analytics and advertising Gist can use. You can change your choices in Privacy settings. Privacy policy
