Learn / Regulated wedge
"SOP not followed" is a design failure, not a discipline failure
It is the finding that shows up more than any other, and the response is almost always the same: retrain the operator, reissue the document, close the CAPA. Six months later the same finding is back, now with "repeat" attached, which an inspector reads far more harshly. The finding is real. The diagnosis is wrong.
Picture the acceptance clerk on the late shift accepting a dangerous goods consignment under IATA DGR rules, with a temperature-controlled pharma pallet in the same build. The SOP for standard acceptance is fourteen pages in a shared drive. The DGR exception is in a different document. The cold chain handling note is a third. Under time pressure, with the aircraft's cut-off approaching, the clerk does what works, from memory, the way they were shown. The auditor later finds a step out of sequence and writes "SOP not followed."
Was that a discipline problem? The clerk did the job correctly and safely. What they did not do was reconcile three disconnected documents in real time, at night, against a scenario none of them fully covered. That is not a person failing a process. That is a process failing a person, and calling it a discipline failure guarantees it repeats.
Why non-adherence is designed in
Most SOPs share a set of properties that make them hard to follow, and every one of them is a design choice, not a worker attribute:
- They describe the happy path only. Drawing every exception into a document is miserable, so the exceptions live in people's heads. But the exceptions are exactly where adherence breaks, because that is where the document goes silent and the operator has to improvise.
- They are not usable at the point of work. A fourteen-page PDF is not a thing you consult while building a pallet. So people learn the job once, from a person, and the document becomes reference material nobody references.
- The current version is ambiguous. If the operator cannot instantly tell which version is live, they will use the one nearest to hand, and sometimes it is the wrong one. Now "SOP not followed" is technically true even though they followed an SOP.
- They do not route. A single flat document cannot resolve to "your steps, for this shipment, in this situation." It shows everything to everyone, which means each person mentally filters it, and mental filtering is where steps get dropped.
Put those together and non-adherence is not an accident. It is the predictable output of the design. When following the SOP is harder than doing the job right, capable people do the job right and skip the SOP. The paper loses to reality, every shift.
The CAPA that actually closes
Corrective and preventive action is where this either gets fixed or gets deferred to the next audit. The difference is entirely in where you point the root cause analysis.
The weak CAPA reads: root cause, operator error; corrective action, retrain the operator; preventive action, remind the team. It closes fast and reopens faster, because nothing about the process changed. The next clerk, on the next late shift, hits the same three-document scramble and produces the same finding. Inspectors have seen this pattern a thousand times, and a repeat finding with a retraining-only history reads as "root cause never addressed."
The strong CAPA starts from the process. Root cause: the acceptance procedure did not resolve to the dangerous goods and cold chain exceptions at the point of work, and the current version was not unambiguous. Corrective action: redesign the procedure so it routes to the exact steps for a DGR-plus-cold-chain shipment, make the live version the only one reachable, and capture the step completion as evidence. Preventive action: apply the same routing to the other acceptance scenarios and put an owner on keeping them true. That CAPA changes the terrain, so the behaviour changes without anyone being asked to try harder.
This is the same logic that runs underneath a good temperature excursion investigation: the goal is that the correct action is the one the process makes easiest, under pressure, every time. And it is why GDP inspection readiness cannot be a binder: an inspector who traces a real shipment is testing whether the followed process and the written process are the same thing.
Make adherence the default, and prove it
The durable fix has two halves. First, design the process so the right route is the easy route: one master process that resolves to the exact steps for this shipment and this scenario, exceptions included, so nobody has to reconcile three documents from memory. Second, make adherence provable: capture who completed which step, against which version, so "SOP followed" is a record you can show, not a claim you defend. When both are true, the finding does not just close. It stops recurring, because the condition that produced it is gone.
Where FLOW fits
FLOW holds the SOP as one living master process that resolves to the exact route for the situation in front of the operator, exceptions and all, with owners, versions, sign-off, and an audit trail. The current version is the only one anyone can reach, and step completion is captured as evidence, so non-adherence stops being designed in. It keeps your QMS as the controlled document store and feeds it a process that is provably alive and followed. That is the objection-handle for the QA leader who says "we already have SOPs": you keep them, and FLOW is the living process that proves they are current and adhered to. The category argument is in the process system of record, and the process graveyard audit will tell you in ten minutes how many of your SOPs are quietly being worked around right now.
Common questions
What does an "SOP not followed" audit finding mean?
It means the auditor observed a gap between the written procedure and the actual practice: a step skipped, a different sequence used, an exception handled off-script, or a version in use that was not the current one. It is among the most frequently cited observations in GxP and GDP inspections. Importantly, it usually does not mean the operator was careless. More often the written SOP did not match the real work, was not findable at the moment of use, or did not cover the situation in front of them.
How do you write a CAPA for SOP non-adherence?
Start the root cause analysis from the process, not the person. Ask why the SOP was not followed: was it unclear, out of date, not accessible at the point of work, or silent on the scenario that occurred? A strong corrective action changes the process so the correct path is the easy path: fix the step, route the exception, make the current version the only one people can reach, and record adherence. A weak CAPA that only retrains staff and reissues the same document tends to reopen, because it treats a design problem as a behaviour problem.
Why do staff not follow SOPs?
Rarely defiance. Usually the SOP describes an idealised happy path that does not survive contact with a real shipment, or it is a long document nobody can navigate under time pressure, or multiple versions exist and the wrong one is nearest to hand, or the exact scenario (a dangerous goods exception, a temperature excursion) is not in the document at all so the operator improvises. When following the SOP is harder than doing the job correctly, people do the job and the SOP loses. That is a design outcome.
Can retraining close an SOP non-adherence finding?
Retraining can close the immediate finding on paper, but if the underlying process is still hard to follow, the same gap tends to recur and repeat findings are viewed far more seriously than first ones. Retraining is defensible as one action inside a CAPA, but as the whole CAPA it signals to an inspector that root cause was not really addressed. Durable closure comes from making the correct route the path of least resistance, then evidencing that it is followed.
Stop retraining people to follow an unusable process.
Bring one SOP that keeps generating findings to a 30-minute pilot session. Leave with it living in FLOW: routed, current, and provably followed.
Book a pilot →