Four stalls that better tracking cannot fix.
Ownership, sequencing and honest task states close a specific gap. There are plans they cannot move at all, and it is worth recognising one early rather than spending three months maintaining a list against a wall.
One. The person at the centre has not agreed
This is the commonest one and a tidy task list hides it best. Adult children build a plan. Every step is owned, dated and correctly ordered. Every step fails at the same point: he does not want an aide in the house, or she will not stop driving, or the assessment is cancelled for the fourth time by the only person entitled to cancel it.
Follow-through watches whether a step moved. A refusal registers as a step that did not move, so the system does exactly what it was built to do and raises it again. And again. Now software is reminding a family weekly, in writing, about something their father has already declined. The tracking is not neutral here: it turns a decision he is allowed to make into an item that reads as a failure.
The honest handling is a closed state meaning declined by him, with the reason recorded. Not stalled. Not pending. This is the state families are most reluctant to use, because entering it feels like giving up, and it is usually the opposite: it is what makes the rest of the plan accurate again.
Two. The stall is a disagreement in coordination clothes
Naming an owner works because work nobody has taken on does not progress. It stops working the moment the family does not agree the work should happen at all.
Then an empty owner field is not a gap, it is a position. One sibling thinks the house should be sold and one does not, so the item called arrange a valuation sits unassigned for six weeks while everyone can see whose name is not on it. A tool built to make ownership visible has become a scoreboard in an argument it has no standing to settle.
The tell: an item discussed at every check-in and assigned at none of them. That is not a coordination failure and no amount of follow-through repairs it. It needs a decision somebody has to make, a conversation somebody has to have, and sometimes a mediator or a professional care manager who is not related to anyone in the room.
Three. The constraint is money, or somebody else's queue
Ordering by consequence helps when order is the problem. It does nothing when the answer is that the thing costs money the household does not have, or that the agency processing the determination is slow for everybody in the queue, not just for this family.
Follow-through can establish that a wait has run past the point where a second call is reasonable, and it can make sure that call is made by a named person on a set date. It cannot produce a determination and it cannot produce funds. Carrying a funding gap as an open task with a date beside it is worse than leaving it off, because it makes the item look tractable and postpones the harder conversation about what the plan looks like without it.
The item nobody wrote down
The three above are visible once you know to look. This one is not, and it is the real ceiling on the approach.
A follow-through system watches the list. It has no way of knowing what is absent from the list. A plan can be current, fully owned, every state accurate, and completely silent about an appeal window nobody knew existed or a program nobody thought to ask about. The tidiness is what makes it dangerous: a well maintained plan feels like a complete one, and the two are unrelated.
Which is why a list needs checking at least once against something outside the family's own knowledge, rather than only against what they already knew to ask for.
Four. Nobody is willing to hold it
The whole approach assumes one person will carry the plan. Most families have somebody who will, and it is usually the person already doing everything else. Some do not: distance, estrangement, shift work, or four siblings who each sincerely believe they are the one contributing.
No system creates a coordinator. Handing an unowned plan to software produces an unowned plan with better formatting, and the weekly summary then reads as an accusation aimed at nobody in particular. If genuinely nobody will hold it, the real options are paid help, a smaller plan that one person can actually carry, or an explicit decision that some of it will not get done. All three are legitimate. Expecting the tool to cover the gap is not.
Why publishing this is the point
A method that claims to work everywhere is not a method, it is a sales pitch. Follow-through closes one gap well: the distance between a plan being right and anything happening, when the family agrees on the plan and somebody will hold it. Outside those conditions it produces a very well organised record of a plan going nowhere, which costs a family the one thing they have least of.
If a plan matches any of the four above, stop maintaining it and address the actual blocker first. The list will still be there afterwards, and it will start working again the moment the thing underneath it moves.
Related: how to tell a plan has stalled, the six task states and why ownership decides it.