A plant manager should start with 3-5 daily KPIs linked to the plant's biggest constraint, then expand to 5-7 for the daily management board [S1][S3].
The operating model matters more than a universal count: every KPI needs a defined formula, a threshold, an accountable owner, and a decision it supports [S1][S2].
Start with 3-5, then expand by operating constraint
Most plants should begin with 3-5 daily KPIs because that is enough to expose the main constraint without turning the morning review into a reporting exercise [S1]. A manufacturing KPI programme should also begin with agreed definitions, clean ERP-MES data, and a decision-oriented dashboard, rather than with a large catalogue of metrics [S2].
The selection rule is simple: choose the few measures that reveal whether the plant can deliver safely, on time, and at the required quality. A plant struggling with equipment losses may start with OEE, unplanned downtime, top downtime reason, FPY, and production attainment. A plant constrained by late orders may instead prioritize schedule attainment, WIP flow, delivery performance, staffing coverage, and safety [S1][S5].
Expansion should follow evidence rather than a fixed dashboard template. If the team can act on five measures, the set is functioning; if it is discussing a sixth or seventh problem that lacks ownership, the dashboard is exposing a control gap rather than creating control [S2].
Use 5-7 KPIs for a daily management board
A daily management board typically supports 5-7 KPIs because the meeting must remain short enough for deviations to trigger action before the next shift [S3]. The same metrics should appear in a shared, consistently updated view, with deviations linked to tasks, owners, deadlines, and visible progress [S3].
The five-category SQCDP structure provides a useful coverage check: safety, quality, cost, delivery, and people. A practical six-KPI board might contain one recordable-incident measure, production attainment, OEE, first pass yield, on-time delivery, and attendance. A seven-KPI board can add the top unplanned downtime reason when maintenance response is a known constraint [S3][S5].
Daily review does not mean every underlying metric must refresh every minute. Shift-level availability, cycle time, and downtime classification can support the daily set, while monthly financial or project measures remain on a slower cadence. One published benchmark treats 15 KPIs as a broader plant framework rather than a daily-review target, with production, quality, maintenance, safety, and delivery-and-workforce coverage [S1][S7].
Compare the main KPI-set approaches
The main options are a 3-5 KPI constraint set, a 5-7 KPI daily board, and a larger 12-15 KPI monitoring portfolio. The correct choice depends on management focus, decision frequency, and the plant's ability to intervene [S1][S3][S5].
The 3-5 KPI approach is strongest for rapid implementation, low cognitive load, and immediate action on one dominant constraint, but it offers less cross-functional visibility. The 5-7 KPI daily board provides broader coverage and supports escalation while preserving a manageable meeting structure. The 12-15 KPI portfolio offers detailed diagnosis across safety, production, quality, cost, people, delivery, and maintenance, but it requires automated consolidation and disciplined prioritization if it is to remain useful [S1][S5][S7].
On management time, the 12-KPI benchmark describes a three-minute scan when all values are assembled in one view and ordered by safety, production, quality, cost, and people. The three-minute result depends on automated data presentation; manually compiling the same figures from emails, spreadsheets, and separate dashboards consumes time without improving the decision [S5].
Choose KPIs that daily action can change
A daily KPI should be specific enough to change a shift decision, measurable from a defined data source, and paired with a threshold that tells the manager when to intervene [S1][S2]. Common production measures include OEE, throughput, and cycle time, with OEE combining availability, performance, and quality into one score [S1].
Quality measures include first pass yield, defect or scrap rate, and the cost of quality. First pass yield is calculated as good units divided by total units started, so a decline can expose process instability before customer complaints appear [S1]. Maintenance measures include mean time between failures, while daily operation benefits from the top unplanned downtime reason and the age of corrective actions [S1][S5].
Delivery and workforce measures should connect to operational exposure, not just administrative completion. Useful examples include production attainment, on-time delivery, attendance, overtime ratio, and overdue action items. The morning benchmark provides concrete warning levels of 100% production attainment, at least 92% FPY, at least 95% attendance, less than 8% overtime, and zero overdue action items; these are presented as one plant dashboard framework rather than universal standards [S5].
Do not confuse a dashboard with a decision system
The primary failure mode is collecting many metrics without assigning decisions, owners, or review cadences, which makes the KPI list ineffective [S1]. A second failure is inconsistent calculation: OEE components and downtime classifications can be interpreted differently by production, quality, and maintenance teams unless definitions are agreed before dashboard construction [S2].
Manual, end-of-shift reporting also loses the control value of daily management. Real-time or automated tracking can expose a slipping metric while intervention is still possible, whereas retrospective reporting often identifies the loss after the operating window has closed [S1].
Different sources set different numerical reference points, so managers should not treat them as interchangeable facts. One manufacturing KPI framework describes many plants scoring around 60% OEE before tracking and treats 85% as the world-class benchmark associated with Seiichi Nakajima [S1]. A separate dashboard framework gives 75% or more as its healthy OEE range and describes 60-75% as typical for most plants [S5]. These references indicate the need to define internal thresholds from the plant's own baseline and improvement plan.
Build a review cadence around exceptions
A daily plant review should scan the agreed KPIs in priority order, stop the routine when a red condition appears, and end with a named action, owner, and due date [S3][S5]. Safety incidents require immediate corrective-action review, and repeated top downtime reasons should trigger escalation when corrective action is not working [S5].
Three cadence levels keep the dashboard manageable: daily for shift-actionable exceptions, weekly for trend and constraint analysis, and monthly for financial or systemic review. This preserves the daily board's purpose without forcing every metric into the morning meeting. The broader 12-KPI morning framework can still function as a reference catalog, with only the relevant measures elevated into the active daily set [S3][S5].
For implementation, the plant should define the 3-5 initial KPIs, confirm calculation rules and data ownership, configure warning and target bands, and run a short trial meeting before expanding to the 5-7 daily board. The guide to tiered daily management with digital boards provides a related framework for separating escalation levels and preventing every exception from reaching senior management.
The trackable signal is action closure: daily deviation count, overdue action count, and mean time from deviation to owner assignment. A daily KPI set is working when the manager can name the plant constraint, identify the responsible action, and verify its effect on the next review, rather than simply adding more metrics [S1][S2][S3].