The analysis is the easy half. A team that spends ninety minutes on a fishbone diagram and reaches a genuine root cause has done the hard intellectual work — and then the problem comes back in March, because the three actions written on the last page were never implemented. This page is about the ninety days nobody documents: how corrective actions stall, what a trackable action looks like, and how to run the review rhythm that closes them out.
Five ways corrective actions stall
1. The owner is a department
“Quality to update the inspection procedure.” Quality is not a person and does not have a Monday morning. Every action needs one named individual who has the authority to make the change — not the operator closest to the failure, who usually has neither the mandate nor the budget. If the named owner cannot implement it alone, the action is too large and should be split until each piece has one owner.
2. There is no date, or the date is “ASAP”
An action without a date is a wish. ASAP is worse than a date three months out, because a real date can be missed visibly and a vague one cannot. The useful question when setting it: what has to be true before this can start, and is that on someone else’s list?
3. The action is a verb with no finish line
“Monitor the process.” “Review the training.” “Improve communication.” None of these can be completed, so none of them can be closed — they sit open until someone quietly deletes the row. Replace the verb with an observable end state: the changeover checklist includes a guide-alignment check, signed off by the shift lead. Now it is either true or it is not.
4. Verification was never scheduled
The action gets implemented, someone marks it done, and nobody ever checks whether the defect stopped. Implementation and effectiveness are two different milestones, and only the second one matters. Our CAPA guide covers writing the acceptance criterion before the check rather than after — the single habit that most reliably keeps a fix from silently failing.
5. The action outlives the person
The owner changes role in week three. Nothing in the tracker notices, because trackers record names, not handovers. Any action with a horizon longer than a month needs a second line: who inherits it. This is unglamorous and it is the reason a third of long-horizon actions never close.
What a trackable action looks like
| As written in most reports | As written to be trackable |
|---|---|
| Quality to review inspection frequency | M. Kowalski updates WI-114 to require a guide-alignment check at every changeover; released and trained by 28 Aug; verified by zero recurrence of defect code 4471 across 20 changeovers |
| Retrain the operators | Shift leads run the revised changeover walkthrough with all 14 line operators by 22 Aug; verified by spot-check of 5 changeovers in September, all with the check performed |
| Improve supplier communication | J. Nowak adds the drift tolerance to the supplier drawing and confirms acceptance in writing by 5 Sep; verified by incoming inspection data for the next three deliveries |
The right-hand column is longer, and that is the point. Each entry answers three questions a reviewer will ask in ninety days: who, by when, and how will we know. If your corrective action plan cannot answer all three for every row, the gap is in the writing, not in the team.
The 3/7/30 review cadence
Actions do not decay evenly. They die in a specific window, and it is earlier than most teams expect:
- Day 3 — is containment still holding, and has every owner actually started? Five minutes per action. This is the check that catches the actions nobody accepted in the room but nobody refused either.
- Day 7 — are the corrective actions in progress rather than queued behind something else? Most actions that fail are already visibly stalled here, a full three weeks before anyone declares a problem.
- Day 30 — effectiveness against the criterion you agreed in advance. Not “has it been done” but “did it work”.
The cadence matters more than its exact numbers. What breaks teams is having no scheduled review at all between the analysis and the moment the problem recurs — at which point the conversation restarts from zero, usually with the same people. If you facilitate these sessions, the workshop guide covers how to close a session so the follow-up is already booked before anyone leaves the room.
Where teams actually track this
Four realistic options. The honest answer is that the tool matters far less than whether the review cadence lives inside it — a tracker nobody opens on a fixed day is a list, not a system.
| Where | Works when | Breaks when |
|---|---|---|
| Spreadsheet | Fewer than ~15 open actions, one department, one owner of the file | Ownership spans departments — nobody knows which copy is current, and the review date is a column nobody sorts by |
| QMS module | Regulated environment; the record has to be inspectable anyway | Actions outside the formal CAPA scope get left out entirely, because opening a CAPA is heavier than the problem warrants |
| Project tool (Jira, Asana, Trello) | The team already lives there daily | Verification has no natural home — a ticket closes when the work is done, which is precisely the milestone that does not matter |
| Goal-tracking platform | Actions need to roll up to something — a quality objective, a scrap-rate target, a departmental goal | Overkill for a handful of one-off fixes with no owner above them |
That last row is the one most quality teams skip, and it is worth a sentence. Corrective actions are rarely isolated: they usually exist because someone has a target they are missing — scrap below a percentage, complaints below a count, downtime under a number of hours. When actions hang off that target instead of living in a separate file, the follow-up review stops being an extra meeting and becomes part of a rhythm that already exists. Tools built for that pattern — cascading an objective down to the people who own the work and showing progress as on track, at risk or behind — fit this better than a task list does. Celorly is one example aimed at small teams, with a free tier for individuals and paid plans for team collaboration from $29 per month as of August 2026; the category is worth knowing about even if you stay in a spreadsheet.
Whatever you choose, three fields decide whether it works: owner, verification date, and acceptance criterion. A tool without those three is a nicer-looking version of the problem.
When the problem comes back anyway
Sometimes every action was implemented, verified, and the defect still returns. That is not a tracking failure — it is a signal that the analysis stopped one level too early, and it deserves a fresh look rather than a harder push on the same actions. Two checks are worth running before you re-open anything:
- Did the analysis name a system or a person? If the root cause was “operator did not follow the procedure”, the only available action was retraining, and retraining decays. Re-run the chain from that answer — our 5 Whys tool flags this pattern as you type.
- Was there a second root cause you deprioritised? Most real failures have more than one contributing cause, and the one you skipped is usually the one that survives. A Pareto analysis of recurrence data tells you quickly whether you fixed the vital few or the convenient few.
If the problem is customer-facing and recurring, escalate the format rather than repeating the analysis: an 8D report forces both a root cause of occurrence and a root cause of escape, which is exactly the pair that recurrence exposes.
FAQ
Why do corrective actions never get implemented?
Usually because the action was never written in a form anyone could act on: the owner is a department rather than a person, there is no date, the action is a verb like “monitor” with no finish line, verification was never scheduled, or the action outlives the person who accepted it. All five are visible in the record on the day the analysis closes.
What makes a corrective action trackable?
One named person, one date, one observable end state, and a verification check with an acceptance criterion agreed in advance. If you cannot answer “what will be different, and who will see it”, the action is not trackable — it is an intention.
What is the 3/7/30 follow-up cadence?
At three days you confirm containment is holding and owners have started; at seven days you check the corrective actions are in progress rather than queued; at thirty days you verify effectiveness against the agreed criterion. Most actions die between day three and day seven — which is exactly where teams schedule no review at all.
How long should corrective actions take to close?
Containment same-day or same-shift. Corrective actions typically two to six weeks, depending on whether they need a document change, a tooling change or a supplier agreement. Anything still open after ninety days is usually not a slow action but an unclear one — reopen the wording rather than extending the deadline again.
Where should you track corrective actions?
A spreadsheet, a QMS module, a project tool, or a goal-tracking platform. The spreadsheet works until you pass roughly fifteen open actions or ownership spans departments. What matters more than the tool is that the review cadence is scheduled inside it — a tracker nobody opens on a fixed day is a list, not a system.
Who should own a corrective action?
One person with the authority to make the change — not the person closest to the failure, and never a department name. Assigning an action to “Quality” or “Production” guarantees it is nobody’s Monday morning. If the named owner cannot implement it alone, split the action until each piece has a single owner.
Related
- Corrective action plan — how to write the plan this page assumes you have
- CAPA process and free form — the regulated version of the same record
- How to facilitate an RCA workshop — closing a session so follow-up is booked
- RCA report template — findings, actions, owners, verification
- Root cause analysis: complete guide