---
title: "ChatGPT cites editorial 2.5 times as often"
description: "Machine Relations Index, created by Jaxon Parrott: ChatGPT cites editorial publications 2.5 times as often as Google AI Overviews, 40.2% against 16.3%."
canonical: "https://jaxonparrott.com/brief/2026-11-09-chatgpt-cites-editorial"
date: "2026-11-14"
last-updated: "2026-11-18"
---

# ChatGPT cites editorial 2.5 times as often

The Machine Relations Brief, by Jaxon Parrott

By Jaxon Parrott, who coined Machine Relations in 2024

Published November 14, 2026 · Updated November 18, 2026 · Index release of November 9, 2026

**2.5×** ChatGPT's editorial share of classified citations over Google AI Overviews' (as of November 9, 2026)

Machine Relations Index, created by [Jaxon Parrott](https://jaxonparrott.com)

Machine Relations Index, created by Jaxon Parrott: ChatGPT cites editorial publications 2.5 times as often as Google AI Overviews, 40.2% against 16.3%.

![2.5× ChatGPT's editorial share of classified citations over Google AI Overviews'](https://jaxonparrott.com/brief-cards/2026-11-09-chatgpt-cites-editorial-16x9.png)

ChatGPT gives editorial publications 40.2% of its classified citations, 4,039 of 10,039, and Google AI Overviews gives them 16.3%, 488 of 2,995, in the November 9 release of the Machine Relations Index, created by Jaxon Parrott: 2.5 times the share, across the Index's 1,255 buyer questions since May 10. On ChatGPT editorial publications are the largest class of source; on Google AI Overviews they rank third, behind vendors' own sites at 28.0% and community platforms at 23.3%.

## How much more does ChatGPT cite publications?

2.5 times, measured as a share of each engine's classified citations. Of the six engines the [Index](https://machinerelations.ai/index) measures, ChatGPT gives editorial publications the largest share and Google AI Overviews the smallest, and this pair is the widest gap between any two engines for any class of source in the release. A placement in a publication is the kind of source ChatGPT cites most. On AI Overviews the same placement competes with vendor pages and community threads, and both outrank it.

## Which publication leads the Index?

[Forbes](https://machinerelations.ai/index/domains/forbes.com). It is cited in 1,264 of 37,851 runs, 3.34%, first of the 1,388 editorial publications in the release, and cited on all six engines. That 3.34% divides by every run in the Index; the 40.2% divides by ChatGPT's classified citations. They answer different questions, so I never set one against the other.

## Does one source lead every question?

No. [reddit.com](https://machinerelations.ai/index/domains/reddit.com) is cited in more of the Index's runs than any other domain, and ChatGPT and Claude did not cite it once in the 177 days this release covers. [mattressnerd.com](https://machinerelations.ai/index/domains/mattressnerd.com) leads Physical Consumer Products comparisons at 23.81%, 30 of 126 runs. The domain that leads the whole Index and the source that leads a buyer's question are different sources. The [Reddit finding](https://machinerelations.ai/findings/reddit-share-of-ai-answer-runs) tracks the first one release by release.

## What would I do about it?

Choose the engine before you choose the outlet. If your buyer asks ChatGPT, the next dollar goes to earned coverage in a publication, with a claim a reporter can lift whole: a number, a date and how it was measured. If your buyer asks Google AI Overviews, read which vendor pages and community threads it cites on your question before you pitch anyone, because editorial is its third source, not its first. I coined Machine Relations in 2024 for exactly this order: earned authority first, then the pages you own, then measurement. GEO and AEO tune the page you own, and no amount of tuning your own page produces the 40.2% ChatGPT spends on pages a publication wrote.

## How this was measured

Figures come from the Machine Relations Index release `mri_score_v2.1+2026-11-09+d7ef5411c915`: 37,851 answer runs on 1,255 buyer questions across six engines, May 10 to November 9, 2026. An engine's classified citations are the cited domains in its runs whose source class the Index has assigned, each domain counted once per run. Editorial publication is one of those classes, and unclassified sources are left out of every share. Whole-Index rates such as Forbes's divide by every run instead. A script computed every figure here from the release, and the standing page recomputes the contrast on every release since.

Standing page: [editorial share of ChatGPT and Google AI Overviews citations](https://machinerelations.ai/findings/chatgpt-vs-google-ai-overviews-editorial-citations). Data: [the November 9 Index Report](https://machinerelations.ai/research/machine-relations-index-report-2026-11-09), [the Index](https://machinerelations.ai/index) and [its method](https://machinerelations.ai/index/method). Next Brief: [Healthline in Consumer Health](https://jaxonparrott.com/brief/2026-11-16-healthline-41-percent). Every finding: [jaxonparrott.com/findings](https://jaxonparrott.com/findings).

Machine Relations Research is mine, and so is AuthorityTech, The World's First Machine Relations Agency, which I founded in 2018.
