What a Working Arrangement Looks Like

Rather than a summary, a description of what the arrangement these notes point toward actually looks like. Organisations asking the same question at team level can compare their process with the data-flow guide: define the purpose, make the record visible to the people in it, and interpret time data alongside finished work rather than on its own.

The shape of it

No continuous system. Short bounded investigations, run when there is a question, with a stated end date.

A question written down before the method is chosen.

A method matched to the question and to the accuracy it needs, not to what is available.

And a decision attached: what you would do differently depending on the answer.

The default investigation

Two weeks of sampling: six to ten prompts a day, five seconds each.

Estimate first, from memory, before looking at the result.

Categorise afterwards, from the raw entries, never in advance.

Read the record as a narrative before summing anything.

And stop at the end of the fortnight whether or not it produced something.

What is kept permanently

A reference class: what each piece of work was and how long it actually took. Three lines per item.

Sleep: bedtime and wake time.

And a monthly line for any long project: done, next, stuck.

Nothing else runs continuously, and none of the three costs more than a minute a week.

The numbers worth having

Your overrun factor, from twenty estimate-versus-actual pairs.

Your billable proportion, if you invoice for time.

Your longest uninterrupted stretch, per day, during any investigation.

And the share of the day that was unplanned, which is the input to every realistic commitment.

Changes

One at a time.

Sized for an ordinary day, not a good one.

With a rule for the day you miss it.

Measured against one figure, four weeks either side, same method.

And dropped without ceremony if it does nothing.

What is deliberately absent

Continuous activity capture running for its own sake.

Categories chosen in advance.

Streaks.

Comparison against anybody else's figures.

Records of other people's behaviour.

And any measurement of things done purely because you enjoy them.

The test

Could you say, right now, what question your current tracking answers?

When did it last change a decision?

Could you stop for a month and name what you would lose?

And is anything running that you have not read in four weeks?

An arrangement that fails these is a habit rather than a tool, and the note on tracking becoming the problem covers what to do about it.

What runs permanently

Three things, none costing more than a minute a week: the reference class of past work, sleep times, and a monthly line per long project.

Everything else is a bounded investigation with an end date.

Worth keeping in mind

An arrangement that cannot answer what question it serves, when it last changed a decision, or what stopping would cost, is a habit rather than a tool.

The short version

Nothing here runs continuously except three things that cost under a minute a week between them. Everything else is a bounded investigation with a question and an end date.

In short

Short bounded investigations with a question and an end date, a method matched to the accuracy needed, one change at a time measured against one figure.

That is the whole arrangement, and it costs very little.

The point

Could you say what question your tracking answers, when it last changed a decision, and what stopping would cost? An arrangement failing those is a habit.. For a complementary perspective on work, measurement or planning, consult Atlassian.