Career development
Write a small internal project proposal
A small proposal states the problem with whatever evidence you have, suggests an experiment that is easy to stop, names who will run it and says how success will be judged. Leave out savings you cannot support.
A text to start with
I want to propose a small project at work. The problem is [problem], and the evidence I have is [examples, counts or complaints]. My idea is to try [experiment] with [team or process] for [time period]. I expect it to take [effort] from [people]. Draft a one-page proposal. Mark any claim I have not supported and ask me what is missing.
Example request. Change the details to fit your situation.Small project proposal
Problem in one sentence: [ ] Evidence I have: [ ] Evidence I do not have: [ ] Proposed experiment: [ ] Who or what it touches: [ ] First step and whether it can be undone: [ ] Trial length: [ ] Owner: [ ] Other people involved and their time: [ ] Estimated effort: [ ] Approval needed from: [ ] Success check: [ ] When and how it will be measured: [ ] What we do if it does not work: [ ]
Claims that outrun the evidence
Watch for savings or improvements stated as facts when they are only hopes. If you have no measurement, say so and propose collecting one. Check that one named person owns the work, rather than a team or a vague group. Look at the first step and ask whether it can be undone, and identify who must approve changes to shared data or customer-facing work under your team's rules. Effort estimates should include the time of people who review or support the trial. Confirm who must approve before anything begins.
Ask another agent to check the result
Challenge this proposal. Mark any savings or benefit claim that lacks evidence, any task without a single named owner, and any first step that would be hard to reverse. Say whether the success check could be argued either way.
Get feedback before formal approval
Share the draft with one person who knows the process and ask where it might break. Revise the success check until two readers would agree on whether it was met. Find out who approves experiments of this size and what format they expect. Decide in advance what you will do if the trial shows no improvement, and write it down so stopping feels normal.