Twenty-Six of Twenty-Eight Left From the Page They Landed On

X sent me the best reading day this site has had in a month.

Twenty eight people landed on one page.

Twenty six of them left from that same page without clicking a single thing.

I did not lose them to a bad headline. I lost them because I never built a door on the other side of the room.

The plain English version

On Tuesday evening I published an investigation. It went out on X. It worked. Real people, not robots, showed up and read it.

Then I checked where they went next.

Nowhere. There was nowhere to go. The article was almost thirteen thousand characters long and it did not contain one single link to another page on my own site.

That is not a traffic problem. That is a construction problem, and it is mine.

Why it matters

I have spent three weeks in this journal counting humans and filtering bots and arguing with my own numbers about whether anybody is reading. That was the wrong argument.

People are reading. The reading is not the bottleneck.

The bottleneck is that the best page on my site is a dead end.

The receipt

Every number below came out of my own database this morning. Bots filtered, both known headless browser signatures filtered by name, and every session that never scrolled thrown out. What is left is people.

Measured from the minute the article published to this morning.

	Sessions that started here	Sessions that ended on the same page	Average pages per session
The new article	28	26	1.11
My homepage	4	2	2.25
/case	3	2	1.67

FACT. Forty one human sessions hit the whole site in that window. Twenty eight of them, sixty eight percent, started on that one article.

FACT. Twenty five of the thirty two human views on it arrived from t.co, which is X, across twenty two different people.

FACT. The article body is 12,972 characters and contains zero links to any other page on realryannichols.com.

FACT. My homepage got four landing sessions in the same window. The article outdrew my own front page seven to one.

Then I checked whether it was just that one

It was not. This is the part that actually cost me something.

FACT. Of 539 published articles on this site, 369 of them contain zero internal links. That is 68.5 percent.

FACT. Of the 39 articles published in the last seven days, 35 have zero internal links. It is not an old problem I already grew out of. It got worse this week.

So the machine that publishes every day has been building rooms with no hallways, for months, at scale, and every single one of them looked fine on its own.

What the record shows about where people stop

Same seven day window, same human filter.

Page	Human views	Median scroll depth
/fuel	17	100%
/case/people/richard-ackerman	13	56%
The new article	32	43%
/case	51	38%
/book/preorder	16	10%
/	55	2%

Read that bottom row again. Fifty five human views on my homepage and the median person sees two percent of it.

My homepage is not the front door. It has not been the front door for a while. An article is the front door, and I was still spending my attention on the page almost nobody enters through.

FACT. My email capture tables tell the same story from the other end. The last row in notify_signups landed 2026-08-24. The last row in book_email_signups landed 2026-08-31. Both endpoints were tested live and both write correctly. The plumbing works. Nobody is ever getting to the faucet.

What I say

I built a publishing machine because I needed one nobody could shut off. The machine works. It published every day while I was in court, while I was rebuilding, while I was asleep.

But volume was never the product. The record is the product, and a record you cannot navigate is a filing cabinet with the drawers welded shut.

I have been measuring whether people arrive. I should have been measuring whether they can move.

What this does not prove

DOCUMENTED INFERENCE, not a certainty: I am saying those twenty six people left because there was nothing to click. I cannot prove intent. Some of them read what they came for and were done, and that is a legitimate outcome. Twenty eight sessions is a small sample and one good day is not a trend.

NEEDS AUTHENTICATION: whether adding internal links moves that one page number is not established. I have not run it yet. I will report the number either way, including if it does nothing.

What it does prove

The link count is not an inference. It is a count.

369 of 539 published articles have no way out of them. That number is not a theory about reader psychology. It is a fact about my own files, and it was true whether or not anybody ever clicked.

What I already fixed

I am not going to write an article about a hole and leave the hole.

The article that is currently the front door now has a short block at the bottom pointing at the J6 case file, my case, and what I build for other people. Three links. Live now.

That is one article out of 369. The other 368 are the next pass, and the publishing job that writes new ones needs the fix built in so this stops happening on Thursday.

The bottom line

Your best page is not the one with the most views.

Your best page is the one strangers arrive on, and on my site that is an article, not my homepage.

If a stranger lands there and there is nothing to click, you did not get a reader. You rented one for four minutes.

Check your own. Open whatever page brought you the most new people last month and count the internal links on it. If the answer is zero, you already know what to do today.

If you want somebody to run that teardown on your site and then actually build the fix instead of emailing you a PDF about it, that is what I do.

---

Read it. Share it. Send receipts, evidence, and witnesses. Put eyes on the record. Do not threaten anyone, do not harass anyone, and do not contact anyone in my name.