HasmordResearchOctober 2026 · AI assistants

Who ChatGPT and Perplexity cite for shopping questions, and how it differs from Google

We asked ChatGPT and Perplexity the same 60 shopping questions from our Google AI Overviews study and recorded every source they cited. All three cited brand websites in about half of their answers or more; beyond that they quote different parts of the web.

Published October 17, 2026 · Corrected October 18, 2026 · Data collected October 17, 2026

Correction, October 18, 2026. The Google AI Overviews numbers on this page come from our Google study, which first counted only the short source list of each overview and missed the sites cited next to its paragraphs. After the recount, Google cites brand websites in 49% of overviews, not 22%, so the gap with ChatGPT (57%) and Perplexity (53%) is much smaller than we first reported, and we removed that claim. The ChatGPT and Perplexity numbers are unchanged.

57% / 53%of ChatGPT / Perplexity answers cited a brand’s own site (Google AI Overviews: 49%)
53%of ChatGPT answers cited a retailer page (Google: 27%)
0YouTube pages cited by ChatGPT or Perplexity (Google: 85% of overviews)
8%of the sites ChatGPT cited were also cited by Google for the same question

Key findings

How often each engine cited each type of site

Share of answers that cited at least one site of each type: Google AI Overviews (55 overviews), ChatGPT (49 answers with sources) and Perplexity (60). An answer can cite several types.

Google AI OverviewsChatGPTPerplexity

Review site or media
89%78%98%
Brand site
49%57%53%
Retailer
27%53%32%
Video (YouTube)
85%0%0%
Forum or social
67%0%8%
Professional or institution
27%20%23%
Independent blog
35%14%3%
Source typeGoogle AI OverviewsChatGPTPerplexity
Review site or media89%78%98%
Brand site49%57%53%
Retailer27%53%32%
Video (YouTube)85%0%0%
Forum or social67%0%8%
Professional or institution27%20%23%
Independent blog35%14%3%
Median sources per answer759

Most cited sites: ChatGPT — walmart.com, rtings.com, tomsguide.com, target.com, nordstrom.com. Perplexity — nytimes.com, cnn.com, goodhousekeeping.com, forbes.com, rtings.com. Google AI Overviews — youtube.com, reddit.com, facebook.com.

What this means for brands

The three engines share two kinds of source, brand sites and review sites, and differ on the rest: video and forums in Google, retailers in ChatGPT, review sites and media in Perplexity.

Method and limitations

Questions: the same 60 shopping questions as the Google study: 5 in each of 12 categories, in four shapes (“best X for Y”, “X vs Y”, “is X worth it”, “how to choose”). We wrote them to sound like real buyers; they are not a random sample.

Collection: one run per question on October 17, 2026. ChatGPT: chatgpt.com search results for a United States location, collected through DataForSEO’s LLM Scraper; the model reported was gpt-6. Perplexity: the Sonar model through its API with United States web search, collected through DataForSEO. Google AI Overviews: the Google study, same day.

Counting: each cited site counted once per answer. For ChatGPT, every source it listed; search results it read but did not cite, images and product cards are not counted. For Perplexity, every cited link. For Google, every site in the overview’s source list and next to its paragraphs, with Google Shopping product links counted separately. Site types were assigned by one person at Hasmord with the same scheme as the Google study, without a second coder. Percentages for ChatGPT use the 49 answers that cited at least one source.

Limits: the 60 questions are written research questions, not a sample of real searches, so the percentages describe these questions only. Answers change between runs, between logged-in and logged-out users, and between an assistant’s app and its API; this is one run on one day, and we are re-running the same questions to measure the change. The raw rows are in the CSV.

Where Hasmord fits

Hasmord is the AEO-native sales platform. Buyers ask questions across AI assistants, search, and sales conversations. Hasmord uncovers where your answers about your products and services fall short, turns evidence into better answers, finds potential customers for you, and connects every lead to your sales process.

It asks your buyers’ questions in ChatGPT, Claude, Gemini and Perplexity, reads Google AI Overviews for the questions you track, and shows which sources each answer relied on, so you can see where your facts need to be right. See the AEO/GEO platform.

Frequently asked questions

Do ChatGPT and Perplexity cite brand websites for shopping questions?

Yes. In our 60-question study, a brand’s own site was cited in 57% of ChatGPT answers that had sources and 53% of Perplexity answers. Google AI Overviews cited one in 49%.

Do ChatGPT, Perplexity and Google cite the same sources?

Rarely. Of the sites ChatGPT cited, 8% were also cited by Google’s AI Overview for the same question; for Perplexity it was 13%. Each engine needs to be checked on its own.

Does ChatGPT cite YouTube or Reddit for shopping questions?

Not in our study. ChatGPT cited no YouTube or Reddit pages in 60 shopping questions. Perplexity cited no YouTube pages and cited forums in 8% of answers. Google AI Overviews cited video in 85% and forums in 67%.

Is a brand website worth optimizing for AI answers?

In our study all three engines cited brand websites: in 57% of ChatGPT answers with sources, 53% of Perplexity answers and 49% of Google AI Overviews. Keep the facts on your own pages complete and the same as on your retail listings.

Related: who Google’s AI Overviews cite · how to get cited by ChatGPT · Perplexity shopping · appear in Google AI Overviews.