---
id: "lib-2026-10-23-publish-stats-get-cited-the-tactic-that-put-my"
type: "library"
title: "Publish stats, get cited: the tactic that put my data all over the web"
lang: "EN"
org: "Peec AI"
source_type: "Practitioner post"
date: "2026-10-23"
week: "2026-W43"
url: "https://nosov.ai/library/lib-2026-10-23-publish-stats-get-cited-the-tactic-that-put-my/"
original_url: "https://peec.ai/blog/publish-stats-get-cited-the-tactic-that-put-my-data-all-over-the-web"
tier: "B"
tier_name: "Useful, limited"
tier_reason: "A GEO researcher's live experiment shows invented statistics being retrieved by ChatGPT and cited by real institutions, but it is a single uncontrolled site with self-reported analytics from a vendor that sells AI-search monitoring."
summary_basis: "full-text"
original_minutes: "10"
tags: ["ai search", "geo", "data fabrication", "retrieval"]
---

# Publish stats, get cited: the tactic that put my data all over the web

Tier B (Useful, limited): A GEO researcher's live experiment shows invented statistics being retrieved by ChatGPT and cited by real institutions, but it is a single uncontrolled site with self-reported analytics from a vendor that sells AI-search monitoring.

Summary based on the full text.

## In short

A post by Metehan Yesilyurt, GEO researcher at Peec AI (bylined 7 October 2026). In March 2026 he launched stateglobe.com with 50,000 statistics pages generated by GPT-4.1 in a single day; every number was invented and none was checked. Google deindexed the site within months, but ChatGPT kept fetching its pages to answer live questions, and the fake numbers were cited in an academic paper, a bank's investment research and other publications. He calls the effect "citation laundering".

## Key numbers

- 50,000: invented statistics pages generated in one day with GPT-4.1 on a new domain
- about $50: total cost, including 15,000 added listicles
- around 72,000: ChatGPT-User requests (live retrieval during answers) in the last 30 days, of over 600,000 bot requests
- 330+: referring domains linking to the fake-statistics site after 7 months
- 10,000+: sessions referred from ChatGPT since March, per GA4

## What it reports

- Setup cost about $30 for the statistics pages and $20 for 15,000 listicles; a basic AI design template; a prompt listing countries and topics plus two sentences.
- Google: almost fully indexed in a week, then visibility collapsed after two updates; indexed pages now get 5 to 20 impressions.
- In the last 30 days: over 600,000 bot requests, around 72,000 from ChatGPT-User, the agent ChatGPT uses to open pages in real time while answering.
- GA4 shows 10,000+ ChatGPT-referred sessions since March; Ahrefs and Semrush show links from 330+ sites, including Shopify, DHL and Bluehost.
- Citations that link back: an academic paper, investment research from Qatar's largest financial group, an Indian economics publication, a Canadian influencer community report.
- On 7 October the site ranked second for visibility on tracked prompts in Peec's tool, behind Statista.
- A second page with verified, sourced statistics also drew quick links from sites that appear to use AI-written content.
- Explanation offered: retrieval rewrites prompts into "statistics" sub-queries, rerankers score relevance not truth, and nothing in the pipeline checks the number.

## How they got it

One live site, observed through Cloudflare bot logs, GA4, Ahrefs, Semrush and Peec's own visibility tracker. The site now carries a disclaimer.

## Limits

- One site, no control group; the 10,000+ sessions figure versus the earlier "more than 10,000 AI sessions" is the author's analytics, not audited.
- Downstream citations counted only when they link back; scale of unlinked reuse is unknown.
- The author works for a vendor selling AI-search monitoring and optimisation, and the post ends with tactics for getting cited.
- The tactic may stop working once published.

## Why it matters for companies

Anyone using AI-assisted research for reports, investor material or policy papers can import fabricated numbers with a plausible source attached. Teams should require a primary source for every statistic in AI-drafted documents and check that the source itself has a method. For communications and marketing teams, the same mechanism means accurate, well-sourced statistics pages are now a channel to be cited by AI answers; the ethical line is the data, not the tactic.

Original: https://peec.ai/blog/publish-stats-get-cited-the-tactic-that-put-my-data-all-over-the-web