Loading
An hour belongs to a phase somebody recognises — "data migration", not "PRJ-4471". People can place their own day against work they remember doing; they cannot place it against an accounting reference, so they approximate.
That approximation is what margin is later calculated from, which makes the naming of phases a finance decision rather than an administrative one.
A week logged on Friday is written from memory, and memory keeps the long meeting and drops the four interruptions. Long tasks inflate, short ones disappear, and the shape of the week is wrong in a way no one can see.
Loop keeps a timer and a weekly grid for this reason. The grid is the fallback for people who move between things, not the plan.
Whichever they will actually do. Loop offers both: a timer for people who work in long uninterrupted blocks, a weekly grid for people who move between things all day. A grid filled in daily beats a timer nobody remembers to start.
To the phase, not the task. Phase level tells you what is overrunning; task level costs more to maintain than it returns, and it is the level people quietly abandon within a month.
They fill it in. The number that matters is the phase total across a project, not whether a particular Tuesday was logged on Tuesday — chasing perfect daily compliance is the usual way time tracking gets abandoned altogether.
Half an hour with your own data usually saves reading three of these. The guides will still be here afterwards.