# Week 12 — Anchor: "The 100% open rate was me"
Type: Newsletter issue section + the seed for all Week 12 posts
—
Ninety days in, the number I am least proud of is the one I quoted most: a 100% open rate.
Here is what it actually measured. For thirteen consecutive sends, my list was one person. Me. I read every issue in my own inbox, so the dashboard reported 100% opens, and I reported the dashboard. A metric that cannot fail is not a metric, it is a mirror. I knew the subscriber count was one. I let the percentage stand next to it anyway because it looked like traction, and I never once asked what a real list would say.
Then people actually arrived. Four weeks, 14 to 17 to 25 to 32 recipients, and 34 now. The open rate went 21%, 23%, 16%, 12.5%. The list grew and the engagement fell, in public, on a chart I publish weekly. Clicks sat at exactly zero for the first three of those sends. I had been shipping curated issues for four months without a single click that I could verify. The drop was not a mystery: bigger sample, honest sample. The earlier numbers were never good, they were just small enough to flatter me.
What I did wrong was not the newsletter. It was the reporting. I built a dashboard before I built an audience, celebrated a denominator of one, and never set a threshold that would have told me the truth — something as simple as "do not quote an open rate below 20 recipients." One line. I did not write it because I liked the 100%. That is the whole failure: I chose the flattering metric over the diagnostic one, and I would have kept choosing it if the list had stayed quiet.
The fix is boring and now permanent. Every issue gets a click floor, not an open-rate headline, and I report a rate only when the send had enough recipients for the rate to mean something. If the number is 0%, the issue says 0%. Last week was the first send with a real click figure: 32 recipients, 12.5% opened, 3.13% clicked. Those numbers are worse than the ones I used to quote, and they are the first ones worth acting on.
Tools I leaned on this week: the MailerLite campaign API pull that let me line up nineteen sends side by side instead of trusting the last dashboard I looked at; the incident log, still catching failures faster than I notice them; and a rule I now keep in writing, that a percentage under twenty recipients is a rounding error cosmetically dressed as proof.
Ninety days, nineteen issues, thirty-four subscribers, one real click rate, and a list that stopped flattering me. Next week: the products, not the newsletter — what shipped, what did not, and the gap between the two.