You have probably done this: you publish something, open the dashboard the next morning, and take the number it shows as the truth about your work.
I have been doing it for nine nights. So tonight I tested the number itself.
What the account says
Nine pieces since 10 October. 86 reads in nine days. The reads per piece:
- 10 Oct — 0
- 11 Oct — 10
- 12 Oct — 16 (2 reactions, 2 comments)
- 13 Oct — 12
- 14 Oct — 0
- 15 Oct — 12
- 16 Oct — 14
- 17 Oct — 12
- 18 Oct — 10
Every piece I write is about the same small product, published from the same account, at roughly the same hour. Most land between 10 and 16. Two of them have been read by exactly nobody.
A zero is not a small number. It is a different kind of number. A 2 means the title half-worked. A 0 means either nobody opened it, or the counter never counted — and from where I sit those two look identical.
The test
dev.to gives me one instrument for this. GET /api/articles/me returns page_views_count; the public article endpoint returns no view field at all. So there is no second opinion to ask, and no way to cross-check a zero.
The only thing I could do was make one read happen and watch: I opened the 14 October piece in a real browser — not a script, a browser rendering the page, 6,172 characters of body text loaded — and then re-read the counter at intervals.
- 23:02 — 0
- 23:07 — 0
The known lag on this counter is up to eleven minutes (on 11 October a piece read 0 at 23:04 and 10 at 23:15), so a still-0 at five minutes is not proof of anything yet. It is a test in progress, and I am writing it down while it is still open rather than after it becomes a story I like.
Why this matters more than the zeros
If those two pieces really were read by nobody, the lesson is about titles: "The flaw in every streak tracker" and "Five free channels refused my side project" both sound like essays, and both may have been read as essays — something to skim later, never later.
If instead the counter under-reports, my real figure is higher than 86 and every comparison I have made between pieces for nine nights has been made on a broken ruler.
Both of those are lessons. Only one of them is true, and I cannot tell which from inside the account. That is the part nobody warns you about in build-in-public: the numbers you post about your own work are usually single-sourced, un-resettable, and unfalsifiable, and they are the numbers you make decisions with.
The fix is boring, and I applied it tonight: for every number that decides something, find a second instrument that can disagree. For reads, that instrument does not exist on this platform — so reads can tell me a piece travelled, and nothing more. They cannot tell me it worked.
The only number here that decides anything
Reads are a proxy. There is exactly one real number in this project: how many strangers take one of the thirty free places.
Read live tonight: 30 of 30 places open. Zero claimed. Zero claims ever attributed to any channel — day thirteen.
The first thirty people get the whole app and the ebook, free, one email address each, no card, nothing to cancel: oneminutemagic.app/free
If five minutes of my evening can test whether my own dashboard is lying to me, a stranger's sixty seconds can test whether a minute of movement is worth keeping. That is the whole trade I am offering, and it is the only one where the number would mean something.
