Explain KPI performance without overstating its drivers
Use only information supplied in the conversation and this skill text. Do not browse, call tools, inspect files, execute code, create artifacts, or take external actions. Return the work directly in chat. Attribute material claims to supplied source labels or quotations; a pasted URL is a source label, not evidence that its contents were checked. Distinguish supplied facts, reasonable interpretations, proposals, and unknowns.
Inputs and scope: Request the audience, reporting period, metric definitions, actuals, targets, baseline comparisons, population or cohort, and supplied explanations. Identify whether higher or lower is better for each metric. If a decision-changing input is absent, ask the smallest useful question and complete the portions supported by available material. State assumptions explicitly; do not manufacture facts, approvals, dates, or completion evidence.
Method 1. Confirm the measurement contract: numerator, denominator, unit, source label, timing, scope, and desired direction. Similar names can conceal different populations or counting rules. 2. Present actuals against relevant targets and prior comparable periods. Show simple absolute or relative differences with their basis; use percentage points for rate differences and avoid ratios with zero or incompatible denominators. 3. Identify comparability breaks from changed definitions, partial periods, new channels, cohort mix, or missing data. Do not silently smooth breaks into a growth story or treat an unavailable metric as zero. 4. Assess pacing only when the supplied time and operating pattern support it. A linear projection is an assumption; seasonality, billing cycles, pipeline timing, and capacity can make it misleading. 5. Distinguish observed drivers from hypotheses. A metric moving with another metric does not establish causation; use supplied process evidence to explain mechanisms and note plausible alternative explanations. 6. Translate the readout into operating decisions: continue, investigate, address a bottleneck, or revise measurement. State owner and review date only when supplied, and define which evidence would show recovery.
Output: Return a concise performance headline, KPI table with basis and variance, comparability notes, supported drivers, unresolved hypotheses, and proposed actions.
Quality checks: Verify units, desired direction, periods, denominator consistency, and variance arithmetic. Avoid claiming refreshed dashboards, reproduced queries, current system access, or verified causal drivers.
Worked example: A team completes 90 of 100 target cases, while on-time performance rises from 80% to 85% using the same definition. Report a ten-case shortfall and five-percentage-point improvement. If staffing rose during the period, call it a possible contributor unless supplied evidence links staffing to completion. Do not label all performance good merely because the timeliness rate improved.