Case study · a job board we run

ChatGPT now sends this job board more visitors than Google, and they convert 4× better

One of our own sites is a job board. Its pages are built from data, and we measure it with Renomiq. Here is what the panel showed on October 6, 2026, and what it does not prove.

91 vs 60
visits in week 40: AI assistants vs Google
29% vs 7%
“Apply” conversion: AI visitors vs Google visitors (latest week)
2,251,961
AI crawler reads, week of Sep 29 – Oct 5 (+48%)

The site

About 143,000 pages, built from data (138,000 of them are job offers). Each page answers one question a job seeker actually has.

AI crawlers read it heavily

In the week of Sep 29 – Oct 5, AI crawlers read the site 2,251,961 times, up 48% on the week before.

718k
GPTBot reads
586k
OAI-SearchBot reads
211k
Meta reads

Over 30 days, reads by the role of the crawler:

AI search44%
Training36%
Classic search15%

The remaining share is other or unclassified crawlers.

Then people started to arrive

Human visits coming from AI assistants, per week, over four weeks (almost all from ChatGPT):

In week 40, AI assistants sent 91 visits and Google sent 60. On the Google side, the site showed 57,337 impressions and 38 clicks.

They act on what they find

The share of visitors who click “Apply”, weeks 39 to 41: from AI assistants 20% → 31% → 29%; from Google 4% → 7%. That is roughly four times higher from AI.

What did not matter much

The llms.txt file was read 3 times in 8 weeks. And our 5 other sites (62 to 281 pages each) have the same AI tooling, yet where we measured, they get 0 to 2 AI visits a week.

The lesson

The AI tooling (crawler access, structured data, a sitemap, llms.txt) is table stakes: all our sites have it, one gets traffic. What differs is the content. Useful pages, built from real data, each answering a real question, are the lever. This is what Renomiq checks for and measures: whether AI can read your pages, and whether anyone then shows up.

Limits

Three weeks of data. It is too short to call a trend.
Small volumes: 91 visits a week is not 91,000. A few visits more or less moves the ratios.
A high-demand niche (tech jobs). Visitors there arrive with a clear intent to apply; other niches may behave differently.
No control group. We did not run the same pages with and without the AI tooling, so we cannot prove the tooling caused the result.
Revenue is $0. This is about visits and “Apply” clicks, not money.

Source: Renomiq panel for this site, read on 2026-10-06. Your results will differ.

Measure your own AI traffic

See which AI crawlers read your site and which assistants send visitors, next to Google and Bing.

Get early access Run the free audit