These are the actual Google Search Console dashboards for all five sites — 16-month Performance views, straight off my screen.
● GSC screenshot 1 of 5 · aiprofitboardroom.com
This is the main site, aiprofitboardroom.com. 27.5K clicks and 1.08M impressions — over a million times the site showed up in Google. Look at the shape of that blue line: a full year sitting near zero, then from April it just climbs and climbs to 200-plus clicks a day. That tall spike near the end is the best single day — 325 clicks.
● GSC screenshot 2 of 5 · aisuccesslabjuliangoldie.com
This one is the patience story. 6.79K clicks and 592K impressions — but look at the line. From August 2025 all the way to April 2026 it is basically flat on the floor. Eight months of almost nothing. Then the flywheel kicks in and it goes vertical on the right edge. Most people quit somewhere in that flat stretch. The data is why you don't.
● GSC screenshot 3 of 5 · aimoneylabjuliangoldie.com
Site three: 1.8K clicks and 175K impressions. It ticked along at a few clicks a day for months, and you can see the daily bars thickening as the cluster articles land. Not a rocket — a compounder.
● GSC screenshot 4 of 5 · bestaiagentcommunity.com
This site did not exist in this data until late March — a true standing start. 609 clicks and 39.2K impressions from zero. Small numbers, but every one of those is a brand-new site getting found with no ads and no audience.
● GSC screenshot 5 of 5 · juliangoldieaiautomation.com
And the fifth site: 797 clicks and 54.1K impressions — nearly all of it since April, when the flywheel started feeding it. Five dashboards, one system, same shape everywhere: flat, then up.
Now here is the same story as a picture — the main site's climb, animated.
0/wkaisuccesslab · was 30–60
0/wkaiautomation · was 0
0/wkaimoneylab · was ~20
0/wkbestaiagent · 1 click in March
0Kclicks · main site · 16 months
0Kimpressions · 90 days
0posts live
0keywords in striking distance
What is coming
Three things to stick around for.
① The skill file
The whole system runs on one Claude skill file with 13 steps. You will see the actual file.
② The flat months
One site sat flat for four months — the part that fools almost everyone.
③ The failure
16 outreach emails. 9 bounced. Killed in 48 hours. The honest bit is near the end.
The framework
The Five-Site Flywheel.
One keyword and one real case study go in. Five different articles come out, publish across five sites the same day, link to each other, and get indexed the same hour.
Each spin makes the next spin easier — that is why it is a flywheel, not a to-do list.
"One keyword in. Five articles out. Same day, every day."
Old way vs new way
What most people do, and what this does.
Old way1 article a day, if you are fast
One site carrying all the risk
One article a day, written by hand
Guessing keywords from a paid tool
Waiting weeks for Google to index it
No links pointing at the new page
New way5 articles a day, on autopilot
Five sites compounding together
Five unique articles per keyword, same day
Keywords pulled from your own Search Console data
Indexed the same hour, not next month
~25 cross-site links per keyword run
Two Claude accounts run it — one on the laptop, one on the desktop.
Days 1–14 · Setup
Five small sites and one skill file.
Five cheap, fast Eleventy sites on Netlify — no WordPress — each connected to Google Search Console, plus the 13-step blog-post skill that carries all the expertise.
What you're watching: the real 13-step blog-post skill, open in my Agent OS SEO tab — scrolled top to bottom. Transcripts as the source of truth, five unique articles per keyword, CTR titles under 60 characters, schema, embeds, CTAs and the index trigger are all written into the file.
recorded live · my dashboard
THINKING IT?"I would need to be a developer to set this up."
The skill file carries the expertise — you supply a keyword and a couple of paragraphs about a real job you did.
The sites are templates. If you can paste text into a file, you can run this.
Days 15–45 · The cadence
One keyword a day. Five articles a day. Every day.
Real slugs from the daily logs — each one published on all five domains the same day, indexed the same day, with ~25 cross-site links added per keyword run.
What you're watching: the real deploy history in my SEO tab — articles shipped to the five sites, each logged with its live URL and build time.
What you're watching: Agent Kanban in SEO-cluster mode — a planned five-article cluster on a live board, with a Done card previewing the shipped article.
THINKING IT?"Five AI articles a day sounds like spam."
Every article is built from a real case study — a true job, with real numbers — not thin filler.
1,496 posts are live and impressions tripled in 90 days. Thin content dies; grounded content compounds.
The unfair advantage
Your own Search Console beats every keyword tool.
After a few weeks, GSC fills with gifts: gaps where Google shows you but you have no page, and leaks where you rank but the title loses the click.
What you're watching: a live Search Console pull inside my Agent OS Research tab — 5,000 real queries analysed, then striking-distance and low-CTR cards painted from the data, each with a one-click "Use topic" button.
recorded live · real GSC data
"Stop guessing keywords. Google is already telling you what to write."
THINKING IT?"Doesn't running the Agent OS burn a fortune in tokens?"
No — that is the biggest myth about it. The everyday 90% runs on free local models and free APIs, and for the frontier work it drives the Claude CLI you already pay for — your Claude subscription includes it, so you are not paying twice.
Inside the Boardroom there are full token-optimisation tutorials, so you cut usage to the bone and never think about it again.
The patience beat
The flat months almost fool everyone.
aisuccesslabjuliangoldie.com sat flat at 30–60 clicks a week from January to April — publishing the whole time. Then three weeks: 81 → 295 → 405.
THINKING IT?"What if my site just stays flat forever?"
New sites sit flat for months — that is normal, and it is why most people quit at ten posts.
The difference here: Claude keeps publishing daily through the flat months, so the waiting costs you no working time.
What failed · No.1
The outreach campaign died in 48 hours.
June 7: an automated backlink outreach campaign with guessed targets — big sites, invented editorial addresses. Here is the honest anatomy.
THINKING IT?"Doesn't a failure like that wreck the whole system?"
No — it was fenced off and killed in 48 hours, and the publishing flywheel never touched it.
That is the point of a circuit breaker: one bad experiment dies alone.
What failed · No.2
Ranking for keywords nobody clicks.
A batch of local SEO pages — "essex seo", "el cajon seo" — ranked and got zero clicks. Ranking without intent is a vanity win. Cut.
essex seo · ranked · 0 clicks ✕el cajon seo · ranked · 0 clicks ✕
What failed · No.3
Four sites fighting over one keyword.
The "deepseek harness" topic ranked across four of the five sites — the variants split each other. The fix: one site owns a topic deeply, the others support with links, not clones.
What failed · No.4
Cheap models wrote cheap articles.
Cheaper model routes produced worse articles — free output got flagged, long-generation models hit rate limits. The fix that stuck: the Claude CLI you already pay for, with a reliable writer profile.
THINKING IT?"Can't I just run this on the cheapest model?"
I tried. The articles were worse, and worse articles do not rank — quality of the writer matters more than saving pennies per article.
You already pay for Claude; the CLI comes with it. Use the good writer.
Days 46–75 · Optimisation
Stop only publishing. Start fixing.
CTR titles rewritten on click-leak pages, one clear CTA on every article, schema and internal links tightened, testimonial videos rotated with no duplicates.
Same rankings — four times the clicks. That is what fixing the leaks buys.
Days 76–90 · The new surfaces
AI search is already paying out.
The site ranked No.1 for "best ai skool community" with an AI Overview placement — and one page got recommended No.1 in Google's AI Overview less than 24 hours after publish.
When other sites mention your brand next to your key terms, AI engines learn to connect you — that is where the compounding goes next.
Three beliefs to drop
What is really holding you back.
Wrong: "AI content gets punished by Google."
Right: 1,496 posts are live and impressions tripled in 90 days — Google runs AI answers at the top of its own results. What dies is thin content, and every article here is built from a real case study.
Wrong: "I would need to be technical to run this."
Right: The skill file carries the expertise — you supply the keyword and a couple of paragraphs about a real job you did. Over 3,000 business owners are inside AIPB, and some started with zero AI experience.
Wrong: "90 days is too long to wait."
Right: The flat months run on autopilot. The old way costs 90 days of your hands — this way costs 90 days of the calendar.
Don't take my word for it
158 pages of members who already broke through these exact beliefs — real businesses, real wins, in their own words.
The AI Profit Boardroom is where the actual Five-Site Flywheel lives — the 13-step skill file, the Agent OS SEO tab, the roadmap and the weekly calls. Same system you just read about, taught hands-on inside.
