Deciding what to promote through which channel, in what order, and how to tell afterwards which channel actually deserves next quarter's budget. How we test →
Written, reviewed and free to take. No run behind them, so no claim about what they do to an output. What that means.
A launch plan usually starts as a list of channel names in a document, with no order, no owner and no score attached to any of them. Work then begins on whichever channel the person writing the list is most comfortable with, which is a preference dressed as a priority. The skills in this category exist to interrupt that at the two points where it costs the most.
The first point is before anything gets built. A channel deserves effort based on how many qualified people it can realistically reach inside the promotion window, how closely its audience already wants this category of thing, how much prior evidence exists that it converts for offers like yours, and how much time and cash it needs before the first result arrives. Those four inputs can be scored honestly in about twenty minutes, and scoring them changes what gets built. Without that step, the widest-reaching channel wins by default, and reach is the axis most likely to be quoted at its theoretical maximum rather than its realistic figure. A list of forty thousand with a three per cent open rate reaches roughly twelve hundred people.
The second point is after the first results land, when someone decides by feel whether to scale a channel or drop it. Two channels are only comparable if they are measured on the same number, and for anything sold by subscription that number cannot be collected at the moment the campaign ends. The margin on a subscription arrives in instalments, so a reading taken when the attribution window closes is a projection, and the figure that should drive a budget decision is the one taken later, once billing data has shown what churn did to that cohort.
There is a third thing these files share, which is a bias against vanity counting. A giveaway that produces twenty thousand entries and four hundred buyers has not grown a list, it has rented an audience of prize hunters and will show it in the open rates two months later. A launch that hits its traffic target because one partner sent a large email has not proven the plan works, it has proven that partner works, and repeating it without that partner will fail. Each skill here is built to separate the number that flatters from the number that predicts.
Nothing in this category can see your analytics, your billing data or your email platform. Every figure these skills work with is one you supply, and a plan built on an optimistic input produces a confident plan that is wrong.
Measured means the skill was given a realistic task on real material, then the identical task was run again with the skill removed, five runs each way. Each output was graded alone, against a rubric written by someone who had never seen the skill, by a session that was not told the other arm existed. Whether it passed was decided by a rule written down before any run executed. Those pages carry the worst case, the median, the p value, and what the skill costs you as well as what it buys.
Not yet measured means exactly that. It is written, it has been read, it is free to take, and we have run no experiment on it, so we make no claim about what it does to an output. It is not a skill that failed. Skills that failed are not published at all, in either state, and their numbers are in the results table.
Measuring one skill properly costs roughly twenty model sessions. We are working down the queue and moving skills from the second group into the first. Read the full method, or go back to all skills.