What I publish and what I do not

I publish mistakes and the mechanics of how something got built. I do not publish revenue in any form, and no client name goes out until that client has said yes to it in writing. The rule underneath both is one sentence: share what you learned, withhold what someone could use against you. The rest of this is why those categories in particular, and where the line moves when the call is harder.

The incentives push building in public somewhere specific. Publish whatever performs, stop thinking about it, and within a few months you have a highlight reel with the word transparency written across it: numbers going up, and every setback arriving already resolved with a lesson attached. Nobody sets out to write that.

What I publish and what I do not

The left column has a reader in it. The right column has a counterparty in it. Most of the hard calls are about which column something belongs to, and the test is the same every time: who is the worst reader this could reach.

Why mistakes travel furthest and cost the least

A mistake has an unusual property. It is close to the most useful thing I can hand a reader, and close to the cheapest thing I can give away.

Cheap, because the cost has already been paid. The money is spent, and the file is already in the bin. Publishing it creates no new exposure. It converts a loss that already happened into something another person can use, and that is the only free transaction available here.

Useful, because advice with no mistake inside it cannot be checked. Price properly is unfalsifiable, and every reader nods at it. A usable version names the decision and the moment it was still open, so they can hold it against the decision sitting in front of them this week. A sentence about how you felt afterwards gives them nothing to check.

That is the part most people skip. Regret is the easiest thing to write and it does a reader no good. The decision is harder to put down, because writing it down means admitting the alternative was visible at the time.

Mistake posts are also self-limiting, which protects you. You can only write one honestly once you are past it, so the writing lands after the danger rather than in the middle of it. A problem that is still live belongs in a conversation with the person involved. Writing it up while it is unresolved makes it harder to solve, and it turns the other party into a character in your content.

Revenue is the highest-engagement and highest-risk thing I could post

Revenue posts perform, which is the difficulty. You collect the reward the day you press post. The bill arrives later, from somebody you were not thinking about when you wrote it.

Four things happen when a figure goes out. None of them are visible on the day.

Competitors learn the size of the prize and roughly how hard the door is. A number tells a stranger what the category pays and where you sit inside it. That is a favour I would rather not do for free.

Clients read it differently from how you meant it. A client who knows the monthly figure knows what share of it they are, so they also know what losing them would do to you. That is a negotiating position, handed over voluntarily, to the person sitting opposite you at renewal.

The people you work with read it too. A public number sets an expectation about what a fair share looks like, stripped of every cost that sits between the top line and what is left at the end.

And it cannot be unpublished. Screenshots outlive edits. A figure posted once gets read as current for years, including by people deciding whether to work with you at a point when it is no longer true in either direction.

Most of the revenue screenshots I see are advertising something else, usually a course about how to get the screenshot. That is what the genre is for.

There is a tax wrinkle on top of all this. Figures get read against registration thresholds, and anybody can look those up. Post a number anywhere near one and the next question you get is a compliance question, from strangers, in public. You can invoice as a sole proprietor and register when you cross the line. That schedule is private and it stays private.

Why I ask for client permission every time

Permission for one post is not permission for the category. A client who agreed to a written case study did not agree to a screenshot of the same work in a thread months later, and the second use is the one that produces the phone call.

So I ask by message instead of on a call. A message leaves a dated yes I can find again. I ask again when the format changes, and if the answer is slow, the answer is no.

VALORAE runs in my name, and what counts as private is not the same across it. Arc clips founders whose faces are already the product, so the raw material was public before we touched it. Media manages social accounts where the account carries the client's own name, and whether the work behind it gets mentioned is the client's call to make rather than mine. Cast is a community for editors and clippers, which is not a client relationship at all.

I hold that line harder than the contracts do, and there is a reason for it. I started building at fourteen and I am sixteen. A contract signed by someone my age is not something either side can lean on. Depending on where you are it may bind neither of us, or only the adult across the table.

The consequence is behavioural. When the paperwork underneath you is thinner than it looks, the discipline has to live in habits, and publishing habits are the cheapest place to put it. What keeps a relationship intact is that I do not say the thing in the first place. Everything you post also gets read through the age lens, which is a tax of its own, and I have written about it separately.

What about numbers that are not mine?

Those I publish, with the source attached, and I publish them often.

Analyst coverage of the clipping economy in early 2026 put creator distribution at roughly one dollar per thousand views against roughly twenty-five for traditional paid social, with more than 780 active campaigns running on a single platform. I used both figures in writing about whether the agency model is finished, because they carry the weight of that argument.

A cited number can be checked and argued with. It belongs to the reader as much as to me. My own numbers cannot be checked by anybody, so they do not work as evidence at all. They work as an assertion about my competence, which is the weakest kind of claim to put in public and the easiest one to be wrong about later.

What building in public is for

The ethos on the rest of this site is output over potential, and builds over decks. Writing counts as a build when it is made of specifics. It counts as a deck when it is made of adjectives.

The audience here is narrow. People building something themselves, and people deciding whether to work with us. Both can recognise the difference between a post that cost something to write and a post assembled to perform. The assembled kind gets more engagement. It buys less, and the trade looks fine for about a year.

The other thing worth keeping straight is that publishing a plan is a promise made to people who never asked for one. Announce a build and you have attached an audience to a decision that should have stayed reversible. I write about things after they exist, or after they failed, and almost never about what is coming.

What I would do if I were starting this week

1. Keep one private file for every number you want to post, and post none of them.
2. Write the mistake up once it has stopped being painful, and name the decision in it rather than the feeling.
3. Ask for client permission by message so the yes is dated, and ask again when the format changes.
4. Read the draft once as the worst plausible reader, which is usually a competitor, or a client two weeks from a renewal conversation.

None of that makes the writing less honest. Every line on that list is a question about who pays for the sentence.

The private file is the most useful document I keep, and it has an audience of one.