Efficiency's denominator is a resource—time, steps, or cognitive effort—and the report must say which

What it is

Efficiency's normative form is "resources consumed per unit of output," and the resource must be named: time, operation steps, and cognitive effort are three different denominators that do not convert. "This flow is faster" is an incomplete statement; the complete one is "time per output dropped 20%, step count flat, cognitive load unmeasured."

Why it happens

Different denominators point at different cost structures and can move in opposite directions: fewer steps usually raise per-step choice and memory demands (steps down, effort up); shorter time can come from cutting confirmation steps (time down, error risk up). Unnamed-denominator claims proliferate because every denominator flatters someone—steps favor automation, time favors practiced users, cognitive load favors complex flows. The normative "must say which" exists so the caliber is checkable and auditable.

Studying it

Fix three items in the protocol: the denominator (which resource), the numerator (what counts as one unit of output), and the measurement conditions (proficiency and context). Time uses timestamps from task start to goal state; steps count input events with a stated policy on background events; cognitive effort uses subjective scales (NASA-TLX) or dual-task estimates. Comparisons must share the denominator; cross-denominator results can only be stated side by side, never converted.

Where it stops holding

Every denominator has blind spots: time is contaminated by environmental interruptions; steps miss hidden background automation; cognitive-effort scales carry subjective calibration bias. The output-unit definition also biases—"per record" and "per hour" reorder the ranking. No universal conversion rate exists between resources; any conversion embeds value weights, which is a decision, not a measurement.

Applying it

• Keep an efficiency-metric register: every metric states its denominator, output unit, and measurement conditions, declared wherever the number appears.
• Ban cross-denominator conversions in conclusions; when multiple resources matter, report the numbers in parallel.
• Re-check optimization claims across all three calibers (time, steps, effort) so a single caliber cannot decorate the whole.

Related

• Same group: B5.09.1 Effectiveness is measured by accuracy and completeness of task outcomes, not by completion rate alone · B5.09.3 Satisfaction is a subjective judgment, correlated with performance but not inferable from it · B5.09.4 Gains on the three dimensions can offset each other; report them separately instead of merging into one score
• Nearby: B4.12 Keystroke-Level Model and operator constants · Q4 Measurement and Reliability
• Search terms: efficiency measures · nasa-tlx · operations per task