About 98 Out of Every 100 Visits to My Websites Weren’t People
By Scott Steinmetz ·
My print shop's website showed more than 66,000 page views last month. That sounds like a busy little store. It wasn't. When I counted only real people sitting at a real browser, the number was closer to 1,300. The rest were robots: search engines, price scrapers, security scanners, and programs poking at the site looking for a way in.
It was the same story everywhere I looked. My portfolio site showed almost 33,000 page views, and about 180 came from people. A local shop whose site I manage looked like it had 33,000, and about 600 were human. Even allowing for people whose ad blockers hide them from the count, roughly 98 out of every 100 "visits" weren't visits at all.
That matters because every decision about a website starts with "how many people saw it?" The free counters most hosting plans include count every robot. The popular analytics tools count people more accurately, but they usually come with tracking cookies, a consent banner, and a dashboard the business owner never opens. I wanted honest numbers, delivered to the people who need them, without any of that.
How the System Works
- Counts Only Real Browsers: The count comes from a tiny, privacy-friendly measurement the browser itself sends, so robots that never load the page in a browser don't show up. No cookies, no consent banner.
- One Email a Month: On the first of every month, each business gets a one-page email: real visitors, how often it showed up in Google and how many people clicked, what they searched for, which pages they read, how fast the site loads on a phone, and whether it was ever down.
- A Short Note in Plain English: At the top of each email, a short paragraph explains what changed and what's worth doing next, so nobody has to read a chart.
- A Check Every Five Minutes: A small program checks each site around the clock, so "was the site up?" has a real answer instead of a guess.
Overcoming the Friction
1. The First Draft Was Too Sure of Itself
My first test report proudly announced the site had been up 100% of the month. It had only been checking for less than an hour. It also restated every number in the table right below it, which is exactly the kind of report people stop reading. Now the note only mentions uptime once there's a full month behind it, skips comparisons when there's no earlier data to compare against, and explains what the numbers mean instead of repeating them.
2. Behind-the-Scenes Pages Were Showing Up as "Popular"
One of the "most-visited pages" in the first test was an address no person ever reads. It's a page the website uses to talk to itself. Same with a line saying the site had sent visitors to itself. Small things, but a business owner seeing those would rightly wonder what else was off. Those are filtered out now, so the list shows only pages a customer would recognize.
The Business Win
Both client businesses whose websites I look after now get an honest monthly picture without logging into anything, and I get one for all of my own sites. It costs nothing to run and needs no upkeep from them. The biggest change is simply knowing the real audience: a shop with 1,300 real page views a month makes very different decisions than one that thinks it has 66,000.
If you've ever looked at your website's visitor count and felt good about it, ask where that number comes from. Next, I'm running a free security checkup on the two servers that run all of this, to see what an attacker would notice first.