Goal outlook

You have a goal with a deadline. The outlook answers one question: at the pace you have actually been going, do you get there?

It answers in three parts — a band, a number of weeks, and the single biggest thing standing in the way — and it declines to answer at all more often than you would expect.

Two gates before it speaks

export const MIN_WEEKS_HISTORY = 2.0;
export const MIN_LINKED_SESSIONS = 3;

Under two weeks of history, or fewer than three deep-work sessions linked to this goal, the engine returns Building and names the next step. Not a cautious estimate, not a wide band — nothing.

That is the difference between a projection and a guess. Two data points draw a line through anything.

The four real bands

The engine computes the progress fraction you need per week from now to hit the deadline, and the fraction you have actually averaged. The ratio between them picks the band.

Needs a change is as strong as the vocabulary gets. It describes the plan, not the person, and the phrase was chosen against a list that forbids the alternatives.

Weeks, not percentages

The second part is weeksDelta — the projected finish against the deadline, in weeks. Positive is late, negative is early.

There is deliberately no confidence percentage. "68% likely to finish on time" implies a distribution nobody computed, from a sample of one person's four weeks. Weeks are a unit you can act on: three weeks late is a number of Saturdays.

The bottleneck

The third part is always set, and it names the biggest real limiting factor with its figure attached — the linked minutes per week, the elapsed weeks, the pace shortfall. One thing, with a number, rather than a list of five suggestions.

Recent linked deep work is carried alongside as corroboration. It is not the driver: the band comes from progress against the deadline, because minutes spent are an input and progress is the outcome, and a tool that grades you on hours logged is a tool you can satisfy without moving.