January 2026 · 5 min
Give it a quarter, then look
If it cannot get into the daily work in about ninety days, that should be said up front.

A useful shape for this work: one workflow, one owner, one number that matters, about ninety days.
Ninety days is long enough to see how the job really goes, build something people will use, and watch it on a normal week. It is not long enough to stand up a new platform or wait on a data project that has been "almost done" for three years.
Teams sometimes want the five-year map first. The center of excellence. The governance board. Those can come later. A board with nothing running is just a meeting.
A quarter is also fairer to the people doing the work. An open-ended transformation eats their calendar and gives nothing back. A ninety-day project has an end. They can give the time because they know when they get it back, and because the thing being built is supposed to give them hours.
What fits: follow-up after a service visit, the path from first call to getting paid, a handoff between systems already running. What does not: replacing the core system, or "AI across the company."
If a vendor needs nine months before anyone sees it, that is a platform bet. It might be the right bet. It is a different kind of project, and it should be staffed like one. If there is no path into the daily work this quarter, the honest move is to say so now.