Threat and error management outside the flight deck
How an observation method from airline safety can be used on customer calls and housing inspections, and the limits to bear in mind when doing so.
Threat and error management (TEM) began in aviation as a way of describing what happens on ordinary flights. Accident investigation works back from something that went wrong, whereas TEM looks at the routine flights where a crew meets a late runway change or a misheard instruction, deals with it and lands safely, and asks which of those small events tend to lead somewhere worse.
In a line observation audit, a trained observer sits on the flight deck and records each of these as a sequence. Something outside the crew's control adds complexity, which is a threat. The crew does something, or misses something, that takes the flight away from what was intended, which is an error. If neither is caught, the aircraft can end up in an undesired state, such as being too high or too fast on approach. Over a few hundred flights, the sequences start to show where a crew's usual checks are catching problems and where they're being missed.
Other settings
Not much of that is specific to aircraft. A customer on a call mentions a bereavement, which is a cue the agent didn't create and has to respond to. The agent moves on to the balance owed, which is an error. The customer agrees to a repayment plan they can't afford, which is the undesired state, and a complaint arrives a few weeks later. A housing inspection can be described in the same way, with a hazard found on site, a step missed in recording its likely cause, and a repeat report six months on.
The words change from one setting to the next, and SOMAT lets each organisation use its own, but the order of events stays the same. That's what makes the records comparable, because when every observer records events in the same order and against the same categories, you can count them and compare one team with another fairly.
| Step | Flight deck | Contact centre | Housing inspection |
|---|---|---|---|
| Opens | Threat | Customer cue | Hazard found |
| Slip | Crew error | Agent error | Inspection error |
| Ends | Undesired aircraft state | Unsuitable outcome | Hazard unresolved |
Limits
Two things aviation has learned about running these audits apply just as much elsewhere. The first is that people behave differently when they think they're being judged, so observation only gives a fair picture if those being observed trust that it won't be used against them. Line audits in aviation are run on that basis, with records kept anonymous and separate from any disciplinary process, and I'd want any pilot outside aviation to agree the same terms before the first session.
The second is that observers need training and a shared understanding of the categories, otherwise two observers watching the same call will record it differently and the comparisons stop meaning much. That's why every pilot includes observer training and an independent evaluator who reviews sessions as they close, and why the sequence rules in SOMAT limit which events can follow which.
It's also worth saying that TEM records what happened, and the reasons for it still have to be worked out by people who understand the job. A pattern of missed cues might come from gaps in training or from call targets that leave no time to listen, and the data on its own won't say which. We cover how we report findings, and the limits we put on them, in a separate article.
A first scheme
For a first pilot, I'd suggest one scheme covering one kind of work, such as calls about arrears or inspections for damp and mould, with a short list of cue and error categories agreed in the first few weeks. That should be enough to be useful without asking observers to learn a long list, and the categories can be refined once the first few hundred sessions show which ones are pulling their weight (and which nobody ever uses!).