What a mechanism actually is
A mechanism is just something that makes a commitment hard to forget. It does not need to be software, a dashboard, or a company process. In my case the mechanism was a sentence I said out loud to my manager in our weekly one-on-one: 'I will do X by Friday.' That was enough. We both remembered it.
The parts that matter are simple: a clear promise, a specific date, and a person who hears it. Drop the person and the date tends to slide. Drop the date and the promise stays fuzzy. Keep those two and almost any format works — an email, a Slack message, or a note in your weekly goals.
- A clear promise: name the result, not just the activity.
- A specific date: not 'soon' or 'this month'.
- A person who hears it: someone you would be embarrassed to disappoint.
The promise mechanism
The simplest mechanism I ever used required no software. Before starting something that mattered, I told my manager: here is the task, here is the result I will deliver, here is the date. Unprompted. He had not asked for it and often did not need it.
That was the point. The value was not in his oversight — it was in the fact that the commitment had left my head and entered someone else's. A promise you make to yourself can be renegotiated silently. A promise someone else is holding cannot.
Behavioural research keeps landing on the same finding: commitments that are written down and reported to another person get completed at meaningfully higher rates than private ones. Gail Matthews' study at Dominican University found written goals plus weekly progress reports to a friend outperformed goals merely thought about. Public-commitment studies show the same effect, and show it is strongest when the witness is someone you actually care about — a close relationship beats a broadcast to strangers.
Founders have no manager — pick one anyway
The whole apparatus that used to keep you moving — a boss, a review cycle, a performance doc — is gone the day you start a company. Most founders replace it with willpower, which works for about six weeks.
Replace it instead with a person. The term for this is an accountability partner: someone who has agreed to hear your commitment and ask you about it later. They do not need authority over you. They need to be someone you would be embarrassed to disappoint.
- A co-founder — convenient, but you're already inside each other's excuses.
- An investor or advisor — a monthly update is already a mechanism; use it deliberately.
- A prospective customer — "I'll show you the import flow on the 14th." The strongest of the lot, because the cost of slipping is external and real.
- A former manager or colleague — knows your work, has no stake, will not soften the question.
- A peer founder — trade commitments both ways; a standing 20-minute weekly call is enough.
- A partner or family member — best for the personal goals you will otherwise sacrifice first.
How to write a promise that can't be fudged
A promise fails when it is vague enough that both parties can walk away with different memories of it. Write it so a stranger could referee the outcome.
- State a result, not activity. Not "work on pricing" — "pricing page live with three tiers".
- One date. Not a range, not "end of the month-ish".
- One named witness. Written down next to the promise.
- A stated consequence when it slips — even a small one. "If it misses, I tell you why before the date, not after."
Inspect on a cadence, not on a feeling
The inspection is the part founders skip, and it is the part that makes it a mechanism rather than a wish. Attach every promise to a recurring review that happens whether or not you feel like it.
Weekly is the right frequency for most commitments — long enough to make progress, short enough that a slip is still recoverable. Use the weekly review here: each promise is a weekly goal, the review is the inspection, and the status is either done, moved with a reason, or killed.
Say the reason out loud when a date moves. A promise that slips silently three times was never a mechanism; it was a mood.
When a mechanism decays
Mechanisms rot, and they rot quietly. The inspection gets rescheduled once, then dropped. The owner changes roles and nobody inherits it. The metric on the dashboard drifts until it measures something adjacent to the outcome rather than the outcome itself.
When you notice decay, do not patch it with more discipline. Either rebuild the loop — restore the inspection, reassign the owner, fix the metric — or kill it and admit the thing was not important. A half-alive mechanism is worse than none, because it produces the feeling of coverage without the coverage.
Using this in FounderVision
Turn each promise into a weekly goal, and put the witness's name in the notes: who you told, what you said you'd deliver, and by when. That is the tool.
If you work with a manager or a team here, share the item so it appears in their review — that is the inspection, and it happens without you having to volunteer for it. If your witness is outside the app, the weekly review is still the moment you send them the one-line status.
The milestones on your timeline are the long game. Mechanisms are how any given week actually moves them.