The Workaround Is the Real Workflow

There is a laminated card taped to a workstation somewhere right now, curling at the corners, listing the seven steps required to order a scan that the system was sold as making effortless. Somebody printed it. Somebody laminated it — which means it survived long enough to be worth protecting. It is, quietly, the most honest piece of documentation in the building: a usability report nobody commissioned, written by the people who use the thing, taped to the thing, ignored by everyone with the power to change the thing.

Hospitals run on these. The paper list beside the electronic one. The shared login that keeps the desk moving when individual access lags three weeks behind the rota. The copy-paste template that turns forty minutes of documentation into four. The instant-messaging group that carries the question the official channel would have answered tomorrow. Safety teams tend to see this catalogue as a problem to be solved — a list of violations, deviations, non-compliances to be stamped out. That reading is almost exactly backwards. Each workaround marks, with forensic precision, the place where the designed system and the real work pulled apart. They are not the disease. They are the symptom doing its job: pointing at where it hurts.

A field guide to the workaround

You can find the same specimens in almost any clinical environment, because they grow from the same conditions.

Paper that refuses to die. The electronic system was meant to retire the handwritten list, and instead the handwritten list sits beside the screen, updated by hand, carried in a pocket. Not from nostalgia. The paper is faster to scan, survives a frozen terminal, and fits the way a team actually holds a shift in its head. The official record lives in the system; the working record lives on the paper. Everyone knows this. Nobody writes it down.

Shared credentials. Access control assumes one human, one login, neatly provisioned the day they start. Reality delivers a locum who arrived this morning, a student who needs to see the screen, a password reset that will clear by Thursday. So a credential gets shared, because the alternative is a patient waiting while the system catches up with the staffing it was never told about. The control was designed for an org chart. The shift was staffed by whoever turned up.

Documentation after the fact. The system wants each entry made at the point of care, in the moment, one field at a time. The work happens faster than that and in a worse order, so the notes get batched — reconstructed at the end, from memory and scribble, when there is finally a chair and a free minute. The timestamps say the record was made contemporaneously. The record was made at the end of a long run, which is a different thing, and occasionally a consequential one.

Unofficial channels. The sanctioned route for a quick clinical question is slow, formal, or buried three menus deep. So the question goes to an instant-messaging group instead, where it is answered in ninety seconds by someone who happens to be looking at their phone. The service moves because the unofficial channel moved it. The governance policy describes a communication system that, on the ground, half the conversations route around.

None of these is exotic. That is the point. They are so ordinary they have stopped registering as workarounds at all — they have become, simply, how the work is done.

Why the workaround exists

There is a distinction from the study of human factors that explains nearly all of this, and it is worth stating plainly. There is the work as imagined — the clean, sequential process that lives in the design document, the policy, the training slide, the procurement demo. And there is the work as actually done — interrupted, parallel, improvised, performed under time and noise and fatigue by people carrying six things at once. The two are never identical. The gap between them is not a sign of bad staff or a bad system. It is a structural fact about complex work: the designer cannot foresee every condition, so the people at the sharp end close the difference in real time. The workaround is what closing the difference looks like.

Which means the workaround is locally rational. Stand where the person stands — this team, this patient, this Tuesday — and the shared login is not a breach but a solution; the paper list is not defiance but competence; the batched note is not laziness but triage of an impossible minute. They are solving the problem in front of them with the tools they have, and solving it well enough that the service keeps running. The behaviour looks irrational only from a distance, from the clean room where the policy was written, by someone who has never had to make the system work at four in the morning.

And much of it traces to a single design failure: the system was built for the audit, not for the shift. It optimises for the things that get inspected — the complete field, the individual login, the contemporaneous timestamp — rather than for the thing that has to happen, which is the work. Build for the inspection and the humans will quietly build, alongside it, the thing that actually does the job. They always do.

The double nature of the workaround

Here is where the comfortable reading breaks down, because the workaround is genuinely two things at once, and any honest account has to hold both.

It keeps the service running. That is real, and it is not a small thing — strip out every workaround tomorrow and the place would seize within the hour. But the same workaround that keeps the service running also carries latent risk, and the risk is of a particular, dangerous kind: it is invisible to the system that is supposed to be managing risk. The shared login means the record no longer reliably says who did what. The paper list holds information the electronic record doesn't, which means a patient's true picture is split across two systems, one of which governance cannot see. The batched note compresses a shift's worth of judgement into a tidy retrospective that reads more certain than the moment actually was.

The workaround is undocumented, so it cannot be examined. It is untrained, so each person reinvents it slightly differently, and the safe version and the unsafe version look identical from outside. It is invisible to governance, so it accrues risk in a blind spot — quietly, reliably, until the day it doesn't. And on that day, when harm finally arrives, the investigation discovers the gap between policy and practice and names that as the finding. The shared login becomes the cause. The paper list becomes the cause. The instant-messaging group becomes the cause. The workaround that held the service together every other day becomes, on the worst day, the thing that gets blamed for the service falling apart — and the deeper truth, that the workaround existed because the official system never fit the work, goes unwritten, because it isn't on the form.

Reading the workaround as data

So treat the workaround as what it actually is: data. Free, unsolicited, brutally honest field research that staff conduct continuously and for nothing, marking on the real map every point where the designed route was unusable. A team that wanted to know exactly where its system fails the work could hardly commission better intelligence, and it is already lying around, taped to the workstations.

This reframes what to do about them, and the first move is counterintuitive: stop hunting them punitively. The instinct — find the workarounds, name the deviations, enforce the policy — feels like safety and is its opposite. Punish the workaround and you do not remove it, because the underlying problem that produced it is still there, still has to be solved, still gets solved the same way. All you remove is your own visibility of it. The behaviour goes underground. It carries on, now hidden, performed more carefully out of sight, no longer available as data — and you have converted an open signal into a covert one, which is the single worst trade in safety. You wanted fewer workarounds. You got the same number, now invisible.

For anyone building clinical software, the lesson lands harder still, and it is the same lesson healthcare startups so often miss about the doctors they are building for: the workflow you designed is a hypothesis, and the workarounds are the result of the experiment. Before you design the workflow, shadow the work. Watch what people actually do — not what the policy says, not what the stakeholder in the meeting described, not the demo path, but the real, interrupted, four-in-the-morning version, with all its paper and shortcuts and side-channels. Every workaround you find is a feature the real world is begging for, or a constraint the design imposed that the work cannot bear. Designed well, the software absorbs the workaround into the supported path, and the laminated card comes down because nobody needs it any more. Designed in a meeting, the software ships, and within a fortnight a new card goes up.

What this means

The reflex to read workarounds as indiscipline is the most expensive mistake in clinical safety culture, because it throws away the best signal the system produces and trains the people producing it to stop. The workaround is not the failure. It is the visible, generous, locally rational trace of a failure that already happened upstream, in the gap between the work somebody imagined and the work somebody actually has to do.

Which leads to the uncomfortable inversion worth ending on. A system with no visible workarounds is not, as it appears on the dashboard, a safe and compliant system. It is far more likely a system where the workarounds have simply gone underground — out of the daylight, out of the data, still load-bearing, still risky, now invisible to everyone whose job is to see them. The laminated card taped to the workstation is not the thing to be ashamed of. The thing to be ashamed of is never having read it.