A project is the only honest way to make something new. Everything else is repetition.
For a century we managed the repetition of making. Now repetition is nearly free, and what is left is the new, which was always our job. This is what I think the discipline becomes, and how to start living it on Monday.
The model
Seven declarations
Intent is the plan.
A plan lists tasks, and tasks change every week. Intent says what must be true when we are done, and it should survive the tasks. Anyone can generate tasks now. Deciding what is worth the outcome is the rare act.
In practice: one page before any work. The outcome, who it is for, what it must never break, and the date of the first go or stop decision. If it does not fit on one page, the project is not understood yet.
Done is a sentence, not a feeling.
What cannot be checked cannot be finished. Write the acceptance checks first, so the sponsor and the owner agree on done before anyone is invested in a different answer.
In practice: three to five pass or fail checks that someone other than the builder can run. "A customer completes the task alone in under five minutes" is a check. "Stakeholders are satisfied" is not.
Cut work by what one mind can hold.
Every handover loses meaning, and every extra person adds lines of communication. Slice the work so that one named owner can explain their slice from motive to detail, and give that owner as many people and machines as they need.
In practice: one name per outcome. Interfaces between slices are written down as contracts. If a slice needs a meeting just to be understood, it is too big.
Attention is the budget.
Hours are abundant. What is scarce is judgment good enough to sign for something. Plan it like money, because it is the one resource that cannot be bought overnight.
In practice: count review hours per owner each week. Put critical reviews where people are fresh. Nobody signs what they cannot explain. When owners are overloaded, cut scope before adding capacity.
Fluent is not true.
Work will look finished long before it is. The new enemy is not slow work but confident, unchecked work, so authors never mark their own work done.
In practice: an independent check on everything that matters, by a second person or a second system whose job is to break it. Sample the rest. Review the plan itself on Friday as well as the product.
Give odds, not dates.
A plan is a forecast, not a vow. Compare it with how similar projects actually ended, and update it weekly. Nobody was ever hurt by an honest probability, only by a comfortable lie.
In practice: report "even odds by June, nine in ten by September", and let the sponsor commit to the second number. Track forecast error over time and show it.
Kill early, kill proudly.
The first stage of a project should buy information as cheaply as possible. A project stopped in week three worked: it cost little and taught everything. A portfolio that never stops anything is not managing risk.
In practice: write the stop conditions at the start, release the people and money the day a project stops, and say thank you in public.
Where it fits in a company
It needs no reorganization. It is a layer on top of what exists. The monthly status deck becomes one screen: the intent, the checks passed, the current odds and the review load. The PMO stops collecting status and starts auditing the three things that predict failure: a vague intent, unchecked work and a forecast that never moves.
It adds to regulated and safety-critical work, it does not replace its documentation. And it will hurt in one place: projects that run on comfortable dates. Sponsors who want a date will resist an honest range, so do not begin with the project everyone is afraid of.
How to start on Monday
- Pick one project and write its intent on one page.
- Write three to five acceptance checks, owned by someone who will not build it.
- Replace the next status report with a range and what would change it.
- Name the stop conditions and the date of the next go or stop decision.
- Count the review hours. If someone is over capacity, cut scope.
Do it for one project for one quarter. If it is not better, drop it.
We do not manage projects. We bring things into the world: on purpose, on time, and true.