APS Scheduling KPIs: Benchmarks for Adherence, Stability, Utilization and Changeover

Use these benchmarks as reference points for spotting a gap worth investigating, not as targets. Your own trend, counted the same way every day, says more about your schedule than any single published figure.

Every figure shows its basis. A superscript number links to a public source in the Sources list, and the sentence gives that source's context. A figure marked no cited source is kept as originally published because no public source stating it has been verified, so treat it as illustrative.

At a glance

Schedule adherence

Schedule adherence is the headline KPI: the percentage of jobs completed in the planned sequence and time window. Typical plants land at 70 to 85 percent no cited source.

Competent operations hit 88 to 92 percent no cited source, and world-class sits above 95 percent no cited source. The most common lever is widening the frozen fence to 3 to 5 days so the shop stops chasing intraday changes no cited source.

How to measure

• Divide jobs completed on plan by jobs scheduled inside the frozen window.
• Count it daily.
• Avoid monthly averages, where good and bad days cancel.

What moves it

• The length of the frozen fence.
• Discipline about intraday changes inside the fence.
• Stabilizing the product families where misses cost the most.

Put a cost on missed adherence, and pick which product families to stabilize first, with the Schedule Adherence Cost Impact calculator.

Plan stability

Plan stability, sometimes called schedule nervousness, measures how many jobs move between planning runs. World-class nervousness is under 5 percent of jobs shifting per regen no cited source, and typical is 15 to 25 percent no cited source.

Anything above 30 percent means the plan is noise no cited source. High nervousness usually traces to demand signals or safety stock parameters changing faster than the fence protects.

How to measure

• Count jobs whose date or sequence changed between two planning runs.
• Divide that count by the jobs in the plan.
• Score every regen, so you can prove a parameter change reduced churn.

What moves it

• A two-zone fence: firm the near term completely and allow flex only beyond lead time.
• How often demand signals change inside the fence.
• How often safety stock parameters change.

Score each regen with the Production Schedule Stability calculator rather than relying on shop-floor sentiment.

Capacity utilization at the constraint

Capacity utilization at the constraint is where output lives, but the target is not full load. World-class constraint utilization runs 85 to 95 percent, with deliberate protective capacity held to absorb variation no cited source.

Non-constraint resources should run lower, often 60 to 75 percent, because loading them to the constraint's level just builds queue no cited source. Overall equipment utilization above 95 percent at every station is a red flag for a plant that will thrash the moment demand shifts no cited source.

For context, the Federal Reserve measured U.S. manufacturing capacity utilization at 75.7 percent in August 2026, 2.5 percentage points below its 1972 to 2025 average¹. That is an industry-wide measure of plant capacity, not a per-resource target.

How to measure

• Divide loaded hours by net available hours for each resource.
• Report the constraint separately from non-constraint resources.
• Keep the coefficient of variation of load across parallel machines under 15 percent no cited source.

What moves it

• Protective capacity held deliberately at the constraint.
• Non-constraint loading kept below the constraint's level.
• Load balanced across parallel machines.

Compare load with finite capacity for each resource with the Finite Capacity Load calculator.

Changeover ratio

Changeover ratio, the share of available time consumed by setup, separates flexible plants from rigid ones. Typical operations lose 15 to 25 percent of capacity to changeover no cited source.

Strong single-minute exchange of die (SMED) programs cut it to 8 to 12 percent no cited source, and world-class high-mix shops hold under 6 percent no cited source. The Lean Enterprise Institute defines the SMED target as reducing changeover times to a single digit, or less than 10 minutes².

For example, a plant averaging 45 minute changeovers that reaches a 20 minute target recovers (45 - 20) / 45 = 56 percent of its changeover time at the same number of changeovers, or over half its lost capacity.

How to measure

• Divide setup hours by available hours for each resource.
• Log setup minutes and setup count separately.
• Size the sequencing opportunity apart from the SMED opportunity.

What moves it

• SMED work that reduces setup minutes.
• Smarter sequencing of like families, which reduces setup frequency.

Rank the sequencing opportunity separately from the SMED opportunity, so you invest where payback is faster, with the Changeover Sequence Savings calculator.

On-time delivery to commit

On-time delivery (OTD) to commit is the customer-facing benchmark, and it should exceed internal adherence by a small margin if your promises carry buffer. Typical OTD sits at 85 to 92 percent no cited source, and world-class holds 98 percent or better no cited source.

The gap between adherence and OTD reveals whether you are missing plan or over-promising. Measure OTD against the original commit date, not the revised date, or you flatter the number.

Improving it rarely means running faster. It means quoting against real finite capacity, so you stop promising dates the schedule cannot support.

How to measure

• Divide orders delivered by the original commit date by orders delivered.
• Never reset the commit date when an order slips.
• Compare OTD with schedule adherence every month.

What moves it

• Promise dates quoted against finite capacity.
• Buffer built into each promise.
• Schedule adherence, which sets the floor for OTD.

See whether the schedule can support a date before you promise it with the Finite Capacity Load calculator.

Bottleneck output attainment

Bottleneck output attainment tracks actual constraint throughput against its rated rate, and it caps every other number. World-class attainment is 90 to 95 percent of the constraint's demonstrated capacity no cited source.

Typical plants leave 15 to 25 percent on the table through starving, blocking, and minor stops no cited source. A 3 point gain in constraint attainment usually beats a 10 point gain anywhere else no cited source.

How to measure

• Divide actual constraint output by its demonstrated capacity.
• Log starving, blocking and minor stops as separate losses.
• Measure it at the resource that actually limits the plant.

What moves it

• Staged material at the constraint.
• A decoupling buffer in front of it.
• No scheduled changeovers during peak demand windows.

Translate a rate improvement at the constraint into system output with the Bottleneck Schedule Impact calculator.

Labor and material coverage

Resource coverage KPIs matter when labor or material, not machines, bind the schedule. Labor coverage, crewed hours divided by required direct hours, should sit at 100 to 110 percent including planned absenteeism of 3 to 8 percent no cited source.

For comparison, BLS data for 2025 put the U.S. manufacturing absence rate at 2.9 percent of full-time wage and salary workers, with 1.6 percent of usual hours lost³. Those measures exclude vacation, personal days and holidays, which a planning allowance may include.

Material coverage inside lead time should hit 98 percent or better for A-class items no cited source. Plants that hit machine utilization targets but ignore these two routinely publish schedules that collapse at execution.

How to measure

• Divide crewed hours by required direct hours for each shift.
• Divide A-class items covered inside lead time by A-class items required.
• Show the binding resource on the dashboard, not only what is easiest to measure.

What moves it

• Crew plans that allow for planned absence.
• Material availability checked before a job is released.

Sequence the levers by payback

Turn the KPIs into an improvement roadmap by sequencing the levers by payback.

• First, stabilize the plan, targeting under 15 percent nervousness, because a nervous schedule corrupts every downstream metric no cited source.
• Second, protect and elevate the constraint toward 90 percent attainment no cited source.
• Third, attack changeover toward under 12 percent of capacity no cited source.
• Fourth, align labor and material coverage above 100 percent no cited source.

A plant that moves adherence from 82 to 93 percent and constraint attainment from 78 to 90 percent typically frees 10 to 15 percent more sellable output from the same assets no cited source.

Quantify the investment side so KPI gains become a funded project rather than a wish list. APS ROI Payback

How to build your own baseline

• Pick one KPI, write its definition down, and keep it fixed.
• Count daily and per resource, not as monthly plant averages.
• Collect several months of data before you set a target.
• Compare against your own trend first and published ranges second.
• Record what changed in the plan whenever the KPI moves.