Building in public: what to share, what to hold, and why the numbers stall

It buys three things at once, and only one of them is marketing.

It buys an audience that exists before you need one. The worst possible time to go looking for attention is launch week, and the people who followed the build are already there on the day there is something to buy.

It buys a forcing function. A public weekly commitment moves work in a way an internal deadline does not, and a surprising number of founders keep the habit purely for this and would tell you so if asked.

It buys a record. Six months of visible reasoning is what a customer, a candidate or an investor reads when deciding whether you are serious. It is slow to accumulate and impossible to fake afterwards, which is exactly what makes it worth something.

What it does not buy is customers on a schedule. That asymmetry is why so many public builders have a large following and a short customer list. If the goal is pipeline this quarter, this is the wrong instrument, and there is nothing wrong with saying that out loud before starting.

Consistency is the entire mechanism, and the enemy of consistency is the improvised post. If you decide what to write on the day you write it, the habit lasts about five weeks.

So fix the slot and fix the shape. Same day, same rough time, same three parts. When the shape is fixed the content largely decides itself, and the work drops from an hour of agonising to twenty minutes of remembering.

A shape that holds up: one decision you made this week and why, one number with the story attached to it, and one thing that did not work. Nothing else. It fits a post, an email, or ninety seconds of talking, which means the same beat can serve every channel you keep.

Write it at the end of the week while the detail is still exact. Specificity is the product here, and specificity has a shelf life of about four days before it decays into a generality anybody could have written.

One post a week held for a year beats daily posting held for three weeks, and it is not close. Most people who quit did not run out of things to say. They ran out of slot, because the slot was never in the calendar in the first place.

The revenue screenshot works once per reader. It runs on novelty rather than information, and novelty is a currency that expires on a fixed schedule.

The reason is structural. A number with no decision attached is not a story. Nobody can learn from it, argue with it, or use it, so the only available response is a reaction. That is why numbers-only accounts collect engagement that never turns into anything.

The fix is a sentence, not a strategy. Post the number, the decision that moved it, and what you would do differently. Three thousand this month, up from twenty two hundred, because we stopped offering the annual discount at checkout and started offering it in the cancellation flow instead. Now somebody can disagree with you, which is the point.

There is a second problem with a numbers feed, and it is the one that quietly ends the habit. It makes the scoreboard the subject, so a flat month becomes unpostable. Flat months are exactly when an audience is most useful and most people go silent, which teaches the readers who stayed to expect nothing from you.

Small numbers with visible reasoning are more interesting than big numbers without it. A founder explaining why four hundred dollars of revenue took nine weeks is teaching something. A screenshot of forty thousand teaches nothing except that forty thousand exists.

Building in public is not a duty to disclose. The working default: share the reasoning, protect the people, and never publish anything you would need a lawyer to help you retract.

Text is the native format for this and it is where to start. It is searchable, quotable, easy to correct, and it takes twenty minutes.

Video adds the one thing text cannot: people hear you think. The hesitation, the emphasis, the specific way somebody sounds describing a problem they genuinely have. Recognition builds faster and the good moments get forwarded, which is most of how a small audience grows at all.

The trade-off runs in both directions. Video is not skimmable, not searchable, and once it is posted a mistake inside it stays a mistake. So use text as the record and video for the relationship, and never let video be the only place your reasoning exists.

The practical failure point is the ratio. Filming ninety seconds takes five minutes. Cutting it, captioning it and finding something to show while you talk takes an hour, and that hour is why week four is silent.

The floor costs nothing: a phone, a window, and the captions built into whichever app you post to. If the beat is text with the occasional video, stay there and spend the hours on the writing.

If video becomes the weekly beat, the hour is the thing to remove. Cutroom takes one spoken take of up to 3 minutes and returns a finished 9:16 MP4: every word transcribed, the hook written, footage placed on the exact words that need showing, captions timed in one of six styles, silence and filler trimmed. A batch is 100 credits and the export is 20 credits per exported minute, so a finished 30 second update is about 110 credits, and Lite is $19.99 a month for 1,250 credits, about 11 of those, cancel in one click. Worth knowing before it becomes the plan: exports are vertical 9:16 only, so a long-form devlog on YouTube is a different job entirely.