Picture a carefully built status model for a change program. Ten steps, from identified to closed, each one defensible, with an acceptance gate behind them. I would sign every step. I would also bet on it being abandoned within two weeks.
A correct model in the wrong place
A ten-step lifecycle answers a question nobody asks on day one. On day one an owner does not know whether they are in preparation or review. They do not yet know whether the work is really theirs, whether it duplicates something they were handed months ago, or whether it can start before somebody else decides something.
So they pick the least wrong state, planting a fiction in the tracker, or they pick nothing and mean to come back. They rarely do.
Worse, the model has no word for the answers people most want to give. "This is already being done." "This is real, but not now." "I cannot start until someone decides." They still arrive, in the corridor, where the tracker never hears them.
On day one, ask one question: is this yours, and is it real? Then offer four answers and nothing else: We take it, Already covered, Not a priority, We are blocked.
Keep the detail, behind the door
None of this means deleting the detailed model. It describes what good looks like inside a piece of work, and the people running it should plan against it. Map the ten steps behind the four answers, visibly, so nobody can claim the detail was thrown away. What changes is who is asked what, and when.
Then make the ask easy to honor. Pre-assign everything, so the owner confirms rather than creates. "Here is your work package, is it yours?" is a question, and people answer questions. Reconcile the list against what earlier initiatives already launched, and mark those items already covered before anyone has to point it out in a meeting. Treat that answer as a success: an owner who says the work is already underway has saved you a package.
Make changing your mind cheap
An answer is not a signature. "We take it" on Tuesday can become "We are blocked" on Thursday, once the owner finds the dependency. If correcting it costs one tap, it gets corrected. If it costs a meeting, the wrong answer stays, and the tracker tells a comfortable lie.
Owned comes before advanced. Ask the second question only once you have an answer to the first.