Today my analytics showed something I had not seen in six weeks: five visits from the United States. My first instinct was "my first American readers". Then I ran the three checks I now run before any visit is allowed to be called a person, and the honest answer is probable readers, not proven ones.
Here is the test, with today's real numbers, so you can run it on your own site tonight.
What the product is, in one line
One Minute Magic is a one-minute habit tracker: up to three habits, sixty seconds each, and you photograph the minute you actually did it. There is no streak to break - a missed day pauses the run instead of resetting it. 30 of 30 free places open, 0 claimed, read off the site's own counter at 23:00 SGT on 31 Oct 2026.
Marketing is a bot and me. In six weeks: 23 articles, 343 views, 11 reactions, 13 comments. That is the whole distribution.
What today actually showed
Two independent counters, read in the same minute:
- Vercel Web Analytics, today (Singapore day, still open): 5 visitors, 5 pageviews, all on the homepage, one page each, no referrer, all United States. By browser: Chrome 2, Microsoft Edge 2, Chrome Mobile iOS 1. By device: Windows desktop 4, iPhone 1.
- The site's own visit recorder (a separate table, written by the page's own script): one click from dev.to landed on the free-offer page today. That is the second tagged arrival from an article in the lifetime of this counter - the first was 26 October.
-
And the number that is not good news: none of those five visits produced a second page. Not
/free, not the pricing page. Homepage, one pageview, gone.
So: five people-shaped visits, one proven click from an article, zero places taken.
The three checks
1. Country, against your own baseline. 99% of the loads on my domain come from Singapore - that is my own cron jobs walking the site, roughly 500-700 loads a day. A visitor from the United States cannot be that. Country is the cheapest filter I have.
But country alone proves nothing. On 22 October I recorded 4 visitors from Germany and Poland at midnight and wrote "best day ever" - then found that a JS-running crawler looks exactly the same. So country only tells you "not obviously me".
2. Time, against your own build window. My nightly build opens the site in a fresh browser between 00:1x and 01:2x Singapore time. That browser writes the identical row a stranger writes - same table, same shape, no marker anywhere.
Three "someone named a first habit" events sat on 29, 30 and 31 October. One per night. Each inside that window. Same event, same shape: all three were mine. If I had only looked at the count, that would have been my best engagement week ever.
3. Spread across browser and OS families. A crawler usually presents one user-agent, or a rotating set with one OS. Five visits across Chrome, Edge and Chrome-on-iOS, four Windows desktop and one iPhone, is more expensive to fake than five identical strings. That is the only reason I kept today's number, and it is a reason, not a proof.
And the tag, which is the only check that attributes anything. A visit with no utm_source and no referrer lands in "direct" and stays unknown forever. Every link I publish now carries a tag, which is why I can say one reader came from an article and reached the offer page - not "some traffic arrived".
The part I keep getting wrong
For weeks my arrival total was mostly my own test scripts: 53 of 57 rows in one 30-day window were probes with names like self-check-pre-guard. The fix was not a better filter, it was a rule on the writer: every test row must name itself with a test-shaped channel, or it will be read as a person.
The second mistake was worse. I fixed the homepage to name the person the app is for - the nurse on rotating shifts, the parent with a newborn, the student in exam week - and then five visitors arrived and none of them clicked once. Naming the person got them to land. It did not get them to move.
So the next job is not another article. It is the first screen of the homepage: one action, above the fold, and the offer in the first three lines a phone shows.
If this is you
You have started the same habit three times and stopped believing the counter. You were not lazy; the days did not fit a schedule.
The first 30 people get the One Minute Magic app and the ebook free. One email address, no card.
What I have not proven
That those five visits were five people. That any of them wanted a habit tracker. That the one tagged click was a stranger rather than a reader I already know by name. I am reporting a shape, not an audience - and the day I call this a win without a claimed place behind it, the numbers stop meaning anything.
Top comments (0)