Snapshot · proof window · Jul 5, 2026

Google has served this site 3,299,205 times.

One operator. One site I own. I published AI-written pages and watched what Google did with them for eleven months. They ranked. Now I’m testing how far that traffic travels.

Follow along Read the numbers

Impressions / mo

699,504

↑ 16% vs. May ’26

Avg. position

Lower is better. Down from 46.

Keywords ranking

9,513

↑ 17% month over month

Top-10 keywords

718

Jul ’26

Ranked in the top ten results.

Live proof

The numbers, as Google reports them.

These are the site’s own Search Console and analytics numbers, refreshed on a real cadence, good month or bad. As of Jul 5, 2026.

Interactive proof

Follow the system month by month.

Not every page wins. The system works at the portfolio level.

All-time ledgerSince Jun ’25

Impressions: 3,299,205
Clicks: 36,894
Keywords ranking now: 9,513
Top-10 keywords (Jul ’26): 718
SERP features (Jul ’26): 1,621
Avg. position now: 8.1
All-time CTR: 1.1%
Sessions: 36,621

Most of those impressions were earned while pages still sat on page two and three, so the all-time CTR reads low at 1.1%. Now that they rank higher, the next lift is clicks, then getting the best pages linked and cited elsewhere.

Growth readThe curve

64×

10,943 impressions in Jun ’25 to 699,504 in Jun ’26. The dips stay in because they happened.

Traffic readJun ’26

9,003

↑ 32% vs. May ’26

Clicks, all-time: 36,894
Pageviews: 45,104

How visitors behaveGA4 · Jun 2025 – Jun 2026

44.9%

Organic bounce rate

2m 21s

Avg. time on page

1.2

Pages / session

Organic search44.9% bounce

Referral43.9% bounce

Direct57.7% bounce

AI referrals49.3% bounce

Under half of organic visitors bounce, and they stay more than two minutes a session. Direct traffic bounces more, and AI assistants now send their own engaged traffic, up from almost nothing a year ago.

Showing the latest verified snapshot; the live feed reconnects automatically.

Looking for the site?

The property stays anonymous on purpose.

You won’t find the site or its articles linked here, and that is deliberate. Every number Google and GA report is public. Which site earned them stays private, so a year of work can’t be reverse engineered into a copycat. If we work together, the case study is yours.

The story

It barely existed eleven months ago.

In Jun ’25 this site earned 10,943 impressions, rounding-error traffic, average position around 27, pages deep where no one ever clicks. Eleven months later, Google showed it 699,504 times in a single month.

Not every page won. A few never climbed out of the long tail. The real earners were a smaller set, and they carried everything else; the rest mostly fed internal links to the pages that could. Over time it was obvious which topics Google actually rewarded, so I leaned into those.

The April ’26 core update knocked the site back. Impressions flattened and average position slid off its March best. May took back the impression and click highs, even as average position eased to the low nines: more queries ranking pulls more mid-table spots into the average, not fewer wins. That dip stays on the page because a case study you can’t see losing in is just an ad.

Why no domain name? The niche pays my bills, and naming it just invites copies. So the proof stays aggregate: totals and full-month trends, each dated when it was pulled.

Phase 2

The pages rank. Now I’m testing whether the traffic travels.

Phase 1 was the hard part: could AI-written pages earn impressions and clicks from Google on their own, no ads, no shortcuts? They did. That’s what you’re looking at.

Phase 2 is narrower. Take the pages that already win and put them where they earn links and repeat readers, without turning the site into a content farm.

Rank first.Compound next.

Who it’s for

The kind of operator I’d build this with.

A good fit

This is for you if…

You own a content site, or are about to, and want organic traffic that compounds.
You know your niche well enough to judge whether a draft is actually good.
You think in months and quarters, the way the chart above was earned.
You care how it actually works, more than a course or a quick-win promise.

Not a fit

Probably not, if…

You need results this quarter. Month one of this case study was 10,943 impressions.
You want to flood a domain with a thousand pages a week and see what sticks.
You just want a list of prompts or a “secret method.”
You’d be publishing into a niche you couldn’t fact-check yourself.

Follow the story

Follow along while it’s still small.

I’m documenting this in the open as it plays out. Leave your email for the occasional note, good month or bad. And if you’d ever want a system like this for your own site or company, say so and I’ll reach out. No pitch, no sequence.

Through it: page one.