Cycle 1

What changed, and why the cycle is two weeks long

Until last week this ran on a quarter. Eighty-four days, four separate scorecards, thirty-seven key results, one person. It lasted ten days.

It did not fail on performance. It failed because a piece of strategy landed one day after the objectives were adopted and changed what they should have been measuring, and an eighty-four day commitment cannot absorb that on day two. A fourteen day one can.

The quarter length was never chosen. It is the shape a quarter has at a company with a team, inherited without anyone asking whether it fit a shop with one pair of hands and no uninterrupted week since May.

The finding that set this cycle’s theme

Cycle 0’s key results, sorted by what each one actually required:

What it required	Rows	Result
Only that the code get written	5	5 of 5 at full marks. One shipped ten days early
That a stranger do something	6	6 of 6 at roughly zero

The split is clean enough to be uncomfortable. Building has not been the constraint for months, and every hour spent proving it again is an hour that produced nothing.

There was one genuine bright spot worth keeping. The share of people who touch the Italian word game and go on to produce a word moved from 19% to 34.7%, which scored 0.75. Activation responds to building. Acquisition never has.

How the two levels work

There is one Enginery scorecard on top and three project scorecards underneath. One rule keeps that from collapsing into scores that are averages of averages, which is how grading goes arbitrary:

The Enginery card grades outcomes. The project cards grade the actions that produce them. Neither is ever an average of the other.

The roll-up is visual. The grading stays one level deep.

What is deliberately absent

Revenue. The posture right now is growth, not income. Nothing here is being monetised, and the scoreboard is how many people arrive rather than how many pay.

Any channel that needs an audience to already exist. No newsletter, no group message, no share-loop target. All three of those scored zero last cycle against audiences of approximately nobody.

Feature work. Exactly one key result on the board involves building something, and it is a new tool in the one lane where the job came back genuinely unserved.

Two products. One was re-staged to a personal thing with no objectives after research showed its job is served by ten shipped competitors. The other was kept by explicit override, with one objective instead of three.

Where these came from

The strategy this runs on is a playbook adopted on 29 August, and underneath it is an ordered list of moves, ranked by how much is learned per hour spent. This cycle is graded against that list rather than assembled fresh, which caught two omissions worth naming:

The cheapest idea in the portfolio was missing. Publishing one thing into an ecosystem registry is a single act that produces a permanently indexed listing, spending somebody else’s index forever rather than my evenings. Sixteen tools are sitting unpublished. It is the only channel scored as interceptable, permanent, and never once attempted, and the first draft of this cycle left it out entirely.

So was half of the falsifier. The playbook predicts that positioning in front of existing demand beats messaging a small audience I assembled myself, on arrivals per hour, by a wide margin. A tagged message to about thirteen people went out on 29 August and has never been read. Retiring that channel was right; leaving the reading un-done was not, because the measurement is already paid for and without it the prediction cannot be tested.

That is the argument for grading a plan against the reasoning that produced it rather than against how good it feels.

The honest risks

Two rows ask a stranger to act inside fourteen days, and one of them asks it four days after the thing ships. That is the most likely place this cycle scores zero, and it is deliberate, because the theme is reach and a cycle that only grades things within one person’s control is the failure being corrected.

Fifteen key results in fourteen days is also denser per day than the quarter was. Eleven of them are cheap: readings, written verdicts, one listing, one instrumentation pass. The whole capacity question is the new tool, which is the only thing here needing an uninterrupted stretch, and uninterrupted stretches have been the scarcest input all year.