€258 on ads, 21 signups, zero customers: a post-mortem
Optimising toward installs instead of subscriptions is the most expensive default setting in advertising. What it cost, and the kill date every test should be given before it starts.
Notes from running the Core 4 on a real product with a real day job's worth of other work. Method, arithmetic, and post-mortems on the things that did not work.
527 customers at €19 a month, 205 at €49, 102 at €99. Then the part nobody divides by: at 6% monthly churn, holding 527 means signing 32 fresh ones every month before you grow by one.
MethodSeventeen channels ranked for one specific person: a solo founder with no audience and ninety minutes a morning. Every row carries its lane, the days until it tells you something true, and the way it fails when there is only one of you.
MethodOne distribution action every working day until you have logged a hundred sessions. The rules, the volume that counts as done in each lane, checkpoints at session 10, 30, 60 and 100, and what to do on the Wednesday after you missed a week.
MethodZero followers, zero euros, and a product that works. No budget closes exactly one of the four lanes, so the other three get ranked by the hours they cost rather than the euros, with a four-week sequence and thirteen free tiers with their limits.
MethodSeventy minutes on one action, fifteen answering replies, five to write down tomorrow's. The week that covers all four ways to get customers, the volumes that count as done, and the arithmetic of a hundred mornings.
MethodNobody quits on day one. Day eleven is where it goes, and it goes so quietly that most founders cannot name the last morning they showed up. What the habit research says about the 66-day median, if-then plans, and streaks that punish you for the wrong thing.
MethodOne reports back in two weeks, the other in three months and then keeps paying after you stop. The side-by-side table, ninety days of arithmetic, and a decision table that maps your actual situation to a split.
ContentThirty concrete posts a founder at zero followers can write this week, grouped by where the material actually comes from, each with the reason it works and a draft attached.
WarmTen messages you can send this morning, from the friend you have not spoken to in two years to the customer who might refer you. The situation each one fits, the full text, and the single line most people ruin it with.
MethodNineteen plays with the arithmetic underneath. How many asks, posts, sends and euros it actually takes, and why the first ten customers never come from the channel you were planning to build.
MethodTalk to people who know you, post in public, contact strangers, or pay to reach them. That is the whole list. What each one costs, how long before it tells you anything, and how to pick when you have ninety minutes a morning.
ColdThe situation each one is for, the subject line, the full body, and the single line most people get wrong. Plus the volume caps that keep a mailbox alive long enough to learn something.
Optimising toward installs instead of subscriptions is the most expensive default setting in advertising. What it cost, and the kill date every test should be given before it starts.
A single operator referred two colleagues, then his boss, then the finance lead. The four questions that turn a happy user into someone who sells for you.
Audience-building advice fails founders because it optimises for reach. The posts that convert are the ones written to a specific person who answered a specific message last week.
Hormozi says pick one. A weekly rotation says otherwise. Both are right, and the resolution is about intensity per morning rather than channels per week.
Below fifty, a reply rate is noise. Above two hundred, you are scaling something you have not tested. The window in between is where the decision lives.
The hundred-day streak that punishes you for one bad Tuesday is the one you quit on day nine. Move the finish line instead of the goalposts.
The linked pieces are live. The rest are written and queued, and go up one a week, on the Monday.
What we tried, what it returned, and the number attached. No roundups.