What readers keep, after seven years of writing
I pulled 30 days of traffic from my old site, the month ending July 23, 2026. Most of what sits on that site was written in 2019 and 2020. Back then the work was web projects and local print campaigns. I had not opened most of those articles in years.
The list came back in an order I would not have predicted.
Four pages took almost all of it: the homepage, an article on writing with ChatGPT, a walkthrough of manual mode photography, and a guide to door hanger campaigns. Three of the four have nothing to do with the business I run now. Read together, they say something consistent about what a reader carries away from a piece of writing.
My most-read article is the one I am retiring
The ChatGPT guide draws more readers than any other page on the site. It hands you the exact commands to turn a video transcript into a finished draft in five minutes. It tells you how many words the paid version could hold, and where the free version gave out.
Every one of those sentences was correct the week I wrote it. Each one describes a version of a tool that has been replaced several times since.
Readers still arrive. That page drew 56 of them last month, more than any other. They stay about 18 seconds.
About 18 seconds is how long it takes to confirm you are in the wrong place.
The task itself has moved. The work that decides an AI draft's quality now happens before the prompt. You supply your own material, and you answer the questions the model asks you. A page of commands cannot carry that.
A piece pinned to a tool's behavior lasts about as long as the behavior does.
The piece strangers cite is the one that did the arithmetic
In March 2021 I published a guide to door hanger campaigns. It carries a full budget: $1,415 for 4,000 hangers, split across demographic research, design, printing, and distribution. It works the labor out in the open. A canvasser places 75 hangers an hour and is paid $15 an hour. Covering 4,000 doors takes 53 hours, which comes to $800.
It also names the two conditions where the math holds. And it names where it does not: a business that cannot carry an $18 acquisition cost, and homes that have asked not to be solicited.
Other marketing sites now cite that article for the cost of a door and the cost of a customer.
The starting figures were a printer's published estimate. The arithmetic on top of them was mine. One of those sites reports a response range as something I have watched in the field. My article reports it as the range campaigns fall into. Numbers travel further than the sentences around them, and they do not always travel with their conditions attached.
Readers rarely keep an article for its view on print. They keep the one that saved them an hour with a calculator.
The piece that holds attention longest has nothing to do with marketing
A walkthrough of shooting in manual mode holds readers longer than anything else on the site. ISO, aperture, shutter speed, and how the three trade against each other. It drew 53 readers last month and held each of them about 52 seconds. That is close to triple what the ChatGPT guide holds.
Those are small numbers over one month, so I read them as a direction rather than a finding.
Here is the direction. That piece was written by someone holding a camera, for someone about to hold one. Each section does something for the reader in the next ten minutes. Nothing in it depended on being current.
Readers keep what they can use
The two survivors share three qualities.
They are specific. Both carry numbers in the title and in the body, and a reader can check every one of them. They are honest about where the advice runs out. The door hanger guide names the businesses it does not suit. And they are usable inside an hour, with no tool version to match and no involvement from me.
The ChatGPT guide was specific and usable on the day it published. Its specifics belonged to somebody else's product.
What the traffic report does not measure
None of those pages produced a consultation. Not one, in seven years. A traffic report measures attention and stops there.
The subjects are also wrong for the work I do now. Most of that archive is being refreshed, redirected, or taken down. A marketing practice for expert firms has no business ranking for door hangers.
The three qualities travel to any subject, including the subjects that serve the work I do now.
Seven years of writing, and two pieces still working. Neither survived on style. Each one carries a number I had to sit down and work out. The number sits in plain sight, where a stranger can check it. That is the question I put to my own drafts now, before anything goes up. Is there something in here a reader could take away and use tonight?
