Prediction grounded in the task
SHERPA uses a task representation, usually hierarchical task analysis, to examine plausible errors. Activities are classified and linked to relevant error modes, avoiding brainstorming that becomes detached from the work people actually perform.
The approach belongs to human and organisational factors. Its purpose is to anticipate system vulnerabilities, rather than label employees as careless or claim that an error will certainly occur simply because it appears in an analysis table.
Distinguishing activity types
The technique distinguishes actions, information retrieval, checking, communication and selection. Classification directs the question: omitting a check differs from communicating incorrect information or choosing the wrong option within an interface presented to the user.
The team should justify why a proposed error is credible in the scenario. Applying every category to every step would create a lengthy, unhelpful document, particularly if it ignored information, design, time and the conditions under which the activity occurs.
Consequences and recovery opportunities
After identifying an error, analysis considers its consequences, detection opportunities and measures that could reduce it. A later review counts as recovery only if it can reveal the problem in time and support an effective response.
For example, shift handover analysis might consider omission of a restriction still in force. Control would not stop at reminding people to communicate properly: it could improve the information record and confirmation of what the incoming team has understood.
Applications and limitations
SHERPA can support procedure design, interfaces and task-based training. It requires methodological competence and collaboration with people familiar with the activity because the credibility of predicted errors depends on understanding the actual operating context.
Its output is neither a universal error inventory nor an automatic probability measurement. Keep the reasoning behind each scenario and review it when work changes. The method demonstrates value by turning specific findings into verifiable improvements, rather than counting hypothetical errors in a worksheet. Proposed controls should subsequently be checked with the people expected to use them in practice.
