How follow-through works

1. One plan, and only one

   A household assessment produces exactly one plan. Running the check again returns the plan that already exists rather than creating a second one.

   This sounds like a technical detail and is not. Two plans for one household means two lists, two sets of owners, and a sister working from the version that does not have the transport call on it. The constraint lives in the database, so it holds even when two people submit at the same moment.

2. Findings become tasks

   Each item on the plan is converted into something specific. A task carries the organisation to contact, a phone number where one exists, roughly how long the call takes, and what to have ready before making it.

   Three rules govern this conversion, and all three cost us output:

   • An item the household is clearly ruled out of never becomes a task. The purpose of ruling something out is to hand back their time. Putting it on a list takes it away again.
   • An uncertain item does become a task. Worth a call to find out is a real action. Treating uncertainty as nothing is how households lose money they were owed.
   • Nothing is invented. If the source does not say what documents to bring, the task says nothing rather than sending somebody to hunt for paperwork they may not need. An unknown effort stays unknown instead of becoming a guess that reads like a fact.

   The same program surfaced twice is one phone call, not two.

3. Order by what unblocks what

   Tasks are sequenced by consequence, never by how valuable an item looks. Anything with a real deadline goes first. Then anything another task is waiting on. Then the steps most likely to succeed. Then the shortest, because a small finished thing creates the momentum that gets the large one done.

   The record request that takes ten minutes to make and six weeks to arrive belongs near the top, even though it is the least impressive item on the page.

   Why only two or three are visible at once.

4. A name against each one

   An owner is a plain label that somebody types. It is not an account, an invitation or a permission, and there is deliberately nothing for it to grow into.

   A task with nobody's name on it is the strongest single predictor that the step will still be open when the plan closes, which is the finding the whole design is built around. Why ownership decides it.

5. It ends on a fixed date

   Every plan is created with an end date already set. When that date arrives the plan closes, and anything still open is marked as needing a fresh look rather than being quietly carried forward forever.

   A plan without an end is ongoing coordination wearing a disguise. What happens at the end.

What is deliberately absent

The clearest way to describe a boundary is to say what was left out on purpose. None of the following exists anywhere in the follow-through mechanism:

• No way to add a task by hand. Every task derives from an item on the plan. The moment a person can type in an arbitrary task, this is a to-do list, and a to-do list has no end.
• No second plan. One per assessment, structurally.
• No calendar, no availability, no scheduling. Not a missing feature. Their absence is what keeps this finite.
• No accounts and no invitations. Owners are text, so there is no membership to accumulate.

Where the record is kept

Every change appends a new entry: a task moved, an owner set, a plan closed. Nothing is edited in place and nothing is deleted to tidy up, because the history of what a household was told and when is not something anyone should be able to quietly revise afterwards.

Corrections are recorded as corrections. A re-check is a new reading, sitting alongside the old one rather than replacing it.

Last reviewed 29 July 2026