ChatGPT Sites SEO Test After 3 Months: What 285 Impressions and 4 Clicks Show

In August, I started testing ChatGPT Sites on three real websites. Two were built entirely with ChatGPT Sites, while the third combines existing WordPress content with a ChatGPT-built page. I documented the starting point in ChatGPT Sites SEO Test: Two Websites and One Hybrid and said I would return with real results instead of assumptions.

This is the first results update. I used the same built-in three-month date range in Google Search Console for all three properties: August 9 through November 8, 2026.

Here is the important distinction: the two websites built entirely with ChatGPT Sites produced 178 impressions, all 4 clicks and 16 indexed pages. The mixed WordPress and ChatGPT property contributed the remaining 107 impressions and 17 indexed pages.

Google found all three properties, indexed pages on each one and tested those pages in real search results. However, four clicks from websites that changed during the test cannot prove that ChatGPT Sites helps or hurts SEO.

Why Small Numbers Are Still Useful

Many published SEO case studies begin after a website is already receiving thousands of impressions or clicks. That is not how most new small-business websites begin. They start quietly. Google discovers a few pages, experiments with where to show them and sometimes sends only a handful of visitors.

I am a wireless consultant learning SEO by working on my own websites. I am not presenting these numbers as a controlled scientific study. I am showing exactly what appeared in Search Console, including the limitations and mistakes, because that is more useful than pretending the test was cleaner than it was.

Two Important Changes During the Test

SingleMale.net began changing direction

SingleMale.net originally contained lifestyle content for single men, including minimalist living and apartment fitness articles. I later began repositioning the site toward dating and matchmaking. Most of the impressions in this report came from the older lifestyle pages, so these results do not measure the newer direction in a meaningful way.

GangsterMobile.com is a mixed property

GangsterMobile.com is tracked as a domain property in Search Console. Its report includes existing WordPress pages and the ChatGPT-built content at audit.gangstermobile.com. Most of the impressions came from the established WordPress pages. The ChatGPT-built audit page received only 10 impressions.

That means GangsterMobile.com is not a clean WordPress-versus-ChatGPT comparison. I will not use it as proof that either platform performed better.

Three Months of Search Console Results

Source: Google Search Console Performance and Page indexing reports, August 9–November 8, 2026.

I intentionally left the combined average position blank. Averaging the positions of three unrelated websites would create a number that sounds precise but does not provide a useful comparison.

Results by Property

SingleMale.net: the most impressions came from the old subject

SingleMale.net received 3 clicks from 128 impressions. Google reported 13 indexed pages and 6 pages not indexed. The homepage produced 2 clicks from 40 impressions, while the minimalist living guide produced 1 click from 26 impressions.

The compact apartment gym page received 37 impressions but no clicks. Its overall average position was 22.1. That suggests Google was giving the content some visibility, but most appearances were not high enough to attract meaningful traffic.

The important limitation is the subject change. These numbers primarily belong to the older lifestyle content. I need to measure the matchmaking direction separately rather than mix the two periods together.

TheCellularConsultant.com: fewer impressions but a relevant query

TheCellularConsultant.com received 1 click from 50 impressions. Search Console reported 3 indexed pages and 4 pages not indexed. The homepage accounted for 43 impressions and the single click.

The leading visible query was “AT&T business wireless consultant,” with 4 impressions and no clicks. Four impressions is a very small sample, but the query closely matches the type of business visitor I want to reach.

The property’s average position was 9.3. I am not treating that as the ranking position for this specific query or as proof of a stable first-page ranking. With only 50 total impressions, a few appearances can move the property-wide average substantially.

GangsterMobile.com: many unindexed pages and no clicks

GangsterMobile.com generated 107 impressions and no clicks, with an average position of 39.8. Search Console showed 17 indexed pages and 49 pages not indexed.

Most impressions came from existing WordPress pages. The ChatGPT-built audit page received only 10 impressions, so the property does not provide enough evidence for a fair platform comparison.

The 49 not-indexed pages are the most important issue to investigate next. I have not yet reviewed every exclusion reason, so I will not guess whether those URLs are duplicates, redirects, low-value archives or pages Google simply has not chosen to index.

What the Data Confirms

• Google discovered and indexed pages on both websites built entirely with ChatGPT Sites.
• Those two properties generated 178 impressions and all 4 clicks recorded in this report.
• Search Console recorded 4 impressions for a business-intent query relevant to my wireless consulting work.
• Publishing and submitting a website does not guarantee that every known page will be indexed.

What These Results Do Not Prove

• They do not prove ChatGPT Sites is better or worse for SEO than WordPress.
• They do not explain why individual pages were indexed or excluded.
• They do not prove the website platform caused the low traffic.
• They do not measure leads, sales or revenue.
• They do not measure the newer matchmaking direction on SingleMale.net.
• They do not evaluate page speed because I have not completed those tests yet.

New domains, limited content, competitive searches and changes in subject can all influence performance. This report cannot separate those factors.

What I Will Test Next

1. Review the indexing exclusions. I will examine the “Why pages aren’t indexed” reports, beginning with GangsterMobile.com.
2. Inspect selected URLs. I will compare an indexed and unindexed page from each property using Google’s URL Inspection report.
3. Run PageSpeed Insights. This was part of my original test and still needs to be completed.
4. Separate the changed content. I will measure SingleMale.net’s matchmaking content from its own starting point.
5. Isolate the ChatGPT-built audit content. I will track the audit subdomain separately instead of allowing the larger WordPress property to hide its results.
6. Track business outcomes. Future updates should include legitimate contact-form submissions and sales conversations, not just search clicks.
7. Repeat the comparison at six months. Using the same reports will show whether visibility is growing or remaining flat.

What This Means for a Small-Business Website

My first three months show that Google can find, index and display a website built with ChatGPT Sites. That is the starting line, not the finish line.

The platform can help build the website, but it does not remove the slower work: choosing subjects real customers search for, creating useful pages, monitoring indexing and improving the site based on evidence.

If you are starting a new website, connect Google Search Console immediately. Record the date and purpose of every major change. Pay attention to the searches that match your business instead of judging success only by total impressions.

For longer-term context from an established WordPress website, read my 16-month ChatGPT-assisted SEO case study for WirelessConsultant.net. That is not a controlled comparison with these newer sites, but it shows how search visibility can develop over a much longer period.

Most importantly, give the test enough time. Four clicks are not a victory, but they are also not proof of failure. They are the first measurable signal in a longer experiment.