If you’re scaling from pilot runs to tens of thousands of units per month, your biggest risk isn’t price—it’s believing brochure capacity. This guide gives you a field-proven playbook to verify real, repeatable capacity using simple math (takt, cycle time, OEE), hard evidence (the last 8–12 weeks of records), on-site timing, a short pilot run, and contract levers to lock in what you’ve validated.
What you’ll achieve
A confident go/conditional/decline decision on a factory’s ability to produce your SKUs at rate and quality
A two-week validation plan (pilot run) with pass/fail gates
A capacity math worksheet you can run in minutes
A document request list and an on-site walk path that surface hidden bottlenecks fast
Contract language pointers to secure reserved and surge capacity
Who this is for
Operations/Procurement Managers and Founders at growth-stage brands ramping from ≤1,000 to 50,000+ units/month
Difficulty and time
Difficulty: Moderate (you’ll be timing lines and reconciling data)
Time: 1–2 weeks for pre-screen + on-site audit; 2 weeks for pilot validation; 1 week to finalize decision/contract
Why this works
You’ll triangulate claims against historical production records, standard capacity math, and a live run. The OEE formula you’ll use (Availability × Performance × Quality) is widely accepted in manufacturing KPI practice and aligned to common definitions used across industry; U.S. manufacturing extension guidance also emphasizes standard KPI capture for reliability as a best practice, as summarized by the NIST MEP Lean & Process Improvement program (2024–2025).
Step 1: Learn the quick capacity math you’ll use
You only need a few numbers to sanity-check capacity.
Key terms
Takt time: available production time per period divided by required output. It’s the “beat” of customer demand.
Cycle time (CT): actual time to complete one unit at a station or line.
OEE: Overall Equipment Effectiveness = Availability × Performance × Quality.
UPH/UPM: units per hour/minute.
Mini-worksheet (fill these in)
Demand and time
Weekly demand (units): ____
Available production time per week (seconds) = working days × shifts/day × hours/shift × 3,600 = ____
Takt (seconds/unit) = available time / weekly demand = ____
Bottleneck timing (on site)
Observed cycle time at bottleneck (sec/unit) = average of multiple timed samples = ____
Raw UPH = 3,600 / cycle time = ____
OEE adjustment
Availability (fraction) = 1 – unplanned downtime / planned production time = ____
Performance (fraction) = (ideal cycle time × total count) / operating time = ____
Quality (fraction) = good units / total units = ____
OEE = Availability × Performance × Quality = ____
Adjusted UPH = Raw UPH × OEE = ____
Weekly output potential
Hours actually run per week on your product (or equivalent) = ____
Estimated weekly output = Adjusted UPH × hours run = ____
Quick pass/fail screen
If observed cycle time > takt, the line cannot meet demand without adding hours/shift(s), improving CT, or increasing parallel capacity.
If OEE at the bottleneck is <60–75% (context dependent), expect chronic shortfalls unless top losses are addressed.
Why trust this math
These definitions and components of OEE and its subfactors (Availability, Performance, Quality) are standard in manufacturing KPI practice; they’re consistent with common interpretations used by industry bodies and training aligned to KPI frameworks and MOM standards, and are reinforced by continuous improvement resources like the NIST MEP Lean & Process Improvement guidance (2024–2025).
Step 2: Pre-screen with a specific document request (last 8–12 weeks)
Ask for the exact records that prove sustained output. Don’t accept summaries—ask for native exports/screens.
Request these by line/workcenter
Line schedules and production reports (planned vs. actual, weekly)
Shift rosters and staffing by skill
OEE dashboards with Availability/Performance/Quality components by week
Downtime log and Pareto (top 3–5 loss modes) with notes/actions
Maintenance records: PM calendar and completion, recent work orders, MTBF/MTTR trend, critical spares list
Quality: FPY trend, scrap Pareto, last internal/external audit findings; calibration certificates for gauges on the line
Process control: routing, CT by station, changeover logs, setup verification checklists
People: training/competence matrix for line roles
Why they should have these
A certified quality system requires controlled production records and evidence of conformity. Even outside automotive, ISO 9001:2015 requires documented information for operations and product release; reputable clause summaries outline the expected documents buyers can request, as in the Advisera list of ISO 9001:2015 mandatory documents and records (2024). Automotive suppliers and similar factories often maintain additional rigor per IATF guidance and customer-specific requirements available at the IATF Global Oversight site (2025 FAQs/CSRs).
Red flags in the pre-screen
Only monthly summaries, no raw weekly/day-level data
Missing downtime Pareto or maintenance logs for the bottleneck machine
Changeover times claimed but not logged; no standard work for setup
Heavy subcontracting of critical steps with no visibility
Step 3: Run an on-site capacity audit (walk path and measurements)
Go see a live run of the same or an analogous product. Time it. Check that what’s on the wall matches what’s in the data.
Your walk path (1–2 days)
Start at the customer end (pack-out/label/QA gate)
Time a sample of 20–30 units for pick–pack–label–seal; compute UPH = 3,600 / avg seconds.
Watch for sustained queues; if end-of-line UPH < upstream assembly UPH, pack-out is your constraint.
Move upstream to the bottleneck candidate
Identify where WIP piles up; confirm by timing cycle time several times across an hour; log micro-stops.
Compare observed CT to claimed ideal cycle time; if you’re >15–20% slower, investigate immediately.
Verify staffing vs. roster
Count actual operators by station and skill; match to the roster; confirm breaks/relief coverage.
Check changeovers
Observe a real changeover end-to-end; time internal vs. external steps; verify pre-staging and standard work.
Maintenance and uptime discipline
Inspect PM completion boards, work orders, and critical spares cabinet; ask about last unplanned stop and fix time.
Quality and calibration
Review FPY trend at the line and final QA; confirm instruments are in calibration and accessible across shifts.
Upstream components and kitting
Visit molding/metal fab cells or the receiving/kitting area; verify kit completeness and feeder process capacity.
What your observations mean
Persistent WIP before a process usually indicates a downstream constraint; flow-time/WIP signals are staple indicators in lean practice and highlighted in resources like the NIST manufacturing flow time brief (2024–2025).
Step 4: Don’t miss the hidden bottlenecks
Common capacity killers aren’t on the main assembly line.
Probe these areas explicitly
Finishing/heat treatment/coating: batch ovens, furnaces, curing rooms with fixed cycles; compute hourly capacity = (batch size × yield) / cycle hours.
Curing/drying/aging: adhesives/paints with long dwell times and limited chamber capacity.
Packaging/labeling: manual pack lines, label print/apply, scan stations.
QC lab: test sample throughput, limited instruments/techs, long test cycles.
Toolroom: die/mold maintenance turnaround; spare cavity availability.
Outbound logistics: palletizing, labeling, staging bottlenecks.
Quick tests
Time 25 packs at end-of-line to see if pack-out keeps pace.
Ask for lab capacity math: stations × minutes/hour ÷ average test time; observe a shift for actuals.
Compare oven batch math to line UPH—if the oven’s net UPH is lower, you’ve found a gate.
Step 5: Run a short pilot build to prove rate and quality
Before you commit large POs, run a controlled pilot (sometimes called Run@Rate) on the target or equivalent line under normal conditions.
Pilot parameters (typical)
Duration/lot: 2 hours or ~300 pieces is a common demonstration size in production validation practices, adapted widely from automotive Run@Rate concepts. Automotive CSRs publicly reference such practices; for example, the FCA (Stellantis) customer-specific requirements (2020) reference a production demonstration run tied to PPAP and Run@Rate expectations, often characterized as “300 pieces or 2 hours,” as seen in the FCA US CSR to IATF 16949 (2020) PDF. Ford’s public PPAP specifics likewise outline sample sizes and capability evidence in validation contexts, as summarized in Ford Specifics for PPAP (2023) and the Ford IATF CSR (June 2025).
Conditions: standard staffing, standard materials and tooling, no extraordinary interventions
Measures: total count, good count, scrap/rework, cycle time distribution at the bottleneck, downtime with causes
Pass/fail gates (tune per product)
Rate: average hourly output ≥ your target at an acceptable OEE window (e.g., 60–75% at bottleneck during the run)
Quality: first-pass yield (FPY) ≥ 95% (or your industry threshold)
Stability: no single downtime cause dominating; capability or control evidence for key measurable features if relevant
Evidence package you keep
Raw counters and QA records; time-study sheets and videos; operator roster for the run; equipment list/settings; deviation log; supervisor sign-off.
Pro tip
If the factory claims surge capacity via a second shift, include a weekend or night trial run to validate staffing, QA, and maintenance coverage off-shift. OSHA emphasizes the importance of coverage across shifts for safe, compliant operations; consult practical guidance such as the OSHA multi-shift inspection guidance (PDF, 2017) when evaluating off-shift readiness.
Step 6: Decide with thresholds and a simple go/no-go matrix
Use the same math and evidence to make a clear decision.
Suggested thresholds (adjust per category)
Sustained output evidence: ≥ X units/week demonstrated for 4 consecutive recent weeks on the target or equivalent line
OEE at bottleneck: ≥ 60–75% range under normal conditions
Lead time: ≤ Z weeks from firm PO to ship for MOQ, with a documented critical path
Surge capacity: ≥ 20% above steady-state via additional shift(s) or parallel lines within 2 weeks’ notice
Pilot FPY: ≥ 95%; defects within agreed limits and stable
Second shift: documented and test-run (named roles, QA coverage, maintenance plan)
Go/no-go matrix (use your findings)
Greenlight: Meets rate and FPY in pilot; 8–12 weeks of data corroborate; surge plan proven; contract terms accepted
Conditional: Misses one area but has a credible, time-bound corrective plan (e.g., SMED event, added pack line) with a re-pilot scheduled and interim mitigations (split POs/dual-source)
No-go: Multiple critical misses (rate, FPY, unaddressed bottlenecks, or unverifiable data); no credible corrections in timeline
Step 7: Lock it in with contractual capacity and visibility
Protect your ramp with enforceable terms once the factory passes validation.
What to include
Reserved capacity: a specific weekly/monthly unit volume or % of line time that is not reallocated without consent
Surge capacity: +20% (or defined units) available with N days’ notice; require proof of staffing/QA/maintenance coverage
Allocation priority: define who gets priority if constraints occur
Reporting: weekly capacity/OEE and shortage/downtime summaries; monthly review
Remedies: service credits/liquidated damages calibrated to impact; step-in rights for outsourced steps if applicable
Audit rights: to review schedules, capacity bookings, and critical inventories
Why this matters
Contract management bodies emphasize clarity and measurable commitments for risk allocation. See principles and standard considerations summarized by the WorldCC (IACCM) Contracting Principles (2022). For sample phrasing of allocation/priority constructs used in practice, you can study public clause repositories like the Law Insider “Priority Customer” clause collection to inform your legal counsel’s drafting.
Step 8: If shortfalls appear, use targeted mitigations
Fix the constraint, don’t spread effort evenly.
When cycle time > takt or OEE is low
Run a downtime Pareto on the bottleneck and attack the top three losses with a focused kaizen.
Improve changeover: apply SMED (separate internal/external setup, pre-stage tools, standardize connectors). Case discussions regularly show 50–90% reduction in changeover time from focused SMED programs; for practitioner-oriented guidance and examples, see resources from the Lean Enterprise Institute’s The Lean Post and principle overviews from the Shingo Institute.
Add hours/shift(s) temporarily while improvements land; protect maintenance windows.
When pack-out or QC limits flow
Add a parallel pack cell, pre-print labels, or rebalance tasks to match assembly UPH.
In QC, add instruments/techs or move to in-line checks; validate measurement system before capability claims (AIAG’s Core Tools explain MSA/SPC practice; see the AIAG Core Tools manuals page).
When upstream components starve the line
Implement kanban/VMI for packaging and fasteners; qualify a backup for critical components; reserve capacity with key tier-2s.
When seasonal labor threatens stability
Cross-train, lock staffing rosters early, and validate off-shift runs; avoid quality drift by ensuring QA coverage and supervisor presence.
Step 9: Keep score weekly with a simple supplier dashboard
Ask for a one-page weekly snapshot and review it jointly.
KPIs to include
OEE trend by bottleneck and key lines; FPY and scrap Pareto
Throughput vs. plan; capacity bookings vs. available hours
On-time delivery; lead time vs. SLA; backlog aging; shortages/expedites
Good practice
Trend charts with red/yellow/green thresholds and a short root-cause/action log for any red KPI. This kind of lightweight, visible cadence is echoed in supply chain improvement guidance such as the ASCM Insights coverage on KPI-driven performance and visibility (2024–2025) and public-sector examples of supplier dashboards like the GSA Supplier Base Dashboard.
Capacity Verification Checklist (printable)
Documents (last 8–12 weeks)
Line schedules (plan vs actual) by week
Production reports (counts, scrap)
Shift rosters and headcount by skill
OEE dashboard (A/P/Q components) by workcenter
Downtime log and Pareto with actions
PM calendar and completion records; last 5 work orders; critical spares list
FPY trend and scrap Pareto; latest audit findings
Routing; CT by station; changeover logs; setup checklists
Training/competence matrix; calibration certificates
On-site observations
Time 3–5 cycles/hour at the bottleneck; record micro-stops
Pack-out UPH vs assembly UPH; end-of-line queues
Real changeover observed and timed; pre-staging present
Staffing matches roster; QA and maintenance presence
Upstream feeder process capacity checked (mold/metal/kitting)
Pilot run evidence
2-hour or ~300-piece run completed under normal conditions
Average hourly rate ≥ target at acceptable OEE window
FPY ≥ threshold; defects logged
Downtime causes logged; no dominant failure mode
Supervisor sign-off; photos/logs archived
Takt/Cycle/OEE Quick Calculator (example)
Example scenario
Demand: 10,000 units/week
Available time: 5 days × 2 shifts × 8 hours × 3,600 = 288,000 s
Takt = 288,000 / 10,000 = 28.8 s/unit
Observed bottleneck CT = 25.0 s → Raw UPH = 3,600 / 25.0 = 144 UPH
OEE (during observation): Availability 0.85 × Performance 0.92 × Quality 0.98 ≈ 0.767
Adjusted UPH = 144 × 0.767 ≈ 110.5
Hours planned for your SKU/week: 80 → Estimated weekly output = 110.5 × 80 ≈ 8,840 units
Gap vs. demand: 1,160 units (13.2% short) → Options: add 10.5 hours/week, improve OEE to ~0.97 (unrealistic), or reduce CT by ~2.8 s via SMED/rebalance, or add a parallel resource.
Common pitfalls and red flags (and what to do)
Seasonal labor reliance with no cross-training → Require skill matrices by station; validate off-season pilot; stage POs; consider dual-source.
Heavy subcontracting of critical processes without visibility → Approve subs; audit their capacity; add step-in rights.
Optimistic changeover times with no logs → Observe changeover; run a SMED event; set max lot sizes based on verified setup time.
Deferred preventive maintenance → Review PM completion; require uptime targets and grace for maintenance windows; confirm critical spares.
Packaging slower than assembly → Add pack cell(s), pre-stage consumables, assign a dedicated QA gate to avoid end-of-line clogs.
QC lab backlogs → Add instruments/techs; shift tests nearer to line; allocate test slots for your product; verify MSA first.
Contract clause pointers (non-legal, for counsel to adapt)
Reserved Capacity: “Supplier shall reserve not less than [X units/week] of Line [ID] for Buyer’s Products during the Term. Reserved Capacity shall not be reallocated without Buyer’s prior written consent.”
Surge: “Supplier shall provide up to [20%] additional capacity with [14] days’ notice. Supplier shall maintain trained staffing, QA coverage, and maintenance plans to support surge.”
Reporting: “Supplier shall provide a weekly report including OEE by workcenter, throughput vs plan, top downtime causes, FPY, shortages, and capacity bookings.”
Remedies: “Failure to provide Reserved Capacity or Surge within agreed notice shall trigger service credits of [$ or %] per [unit/day], capped at [X%] of monthly invoice.”
Audit: “Buyer may audit capacity planning records, production schedules, and inventory related to Reserved Capacity upon [reasonable] notice.”
For broader contracting best-practices and risk allocation considerations, review the WorldCC Contracting Principles (2022).
Ongoing monitoring cadence (post-contract)
Weekly: supplier sends dashboard; 30-minute review of reds and actions; confirm next week’s capacity booking.
Monthly: trend review of OEE/FPY/downtime; corrective actions; confirm surge readiness (rosters, QA coverage, PM windows).
Quarterly: pilot-like spot check on a complex SKU or off-shift; audit critical subs.
Why stay disciplined
Visibility and planning integration reduce surprises. Industry guidance frequently stresses proactive monitoring; for example, NIST’s 2025 visibility reports underline the value of consistent KPI cadence to anticipate risk in ramp scenarios, as seen in the NIST reshoring and visibility report (2025).
Glossary (quick)
Bottleneck: The slowest step that limits throughput.
CT (Cycle Time): Time to complete one unit at a process.
FPY (First-Pass Yield): % of units passing without rework.
MTBF/MTTR: Mean Time Between Failures/Mean Time To Repair.
OEE: Availability × Performance × Quality; fraction of ideal productive time.
SMED: Single-Minute Exchange of Dies; structured method to cut changeover time.
Takt Time: Available time per period divided by demand; the pace required to meet customer demand.
UPH/UPM: Units per hour/minute.
VMI/Kanban: Vendor-managed inventory; visual replenishment signals.
What to do next
Book your pre-screen and request the 8–12 week evidence.
Schedule the on-site audit and identify the bottleneck with real timing.
Plan and run the pilot build; judge against your thresholds.
If green/conditional, negotiate reserved and surge capacity with clear reporting and audit rights.
Stand up a weekly dashboard and never skip the review.
If you follow this sequence—pre-screen, on-site, pilot, contract, monitor—you’ll avoid the classic “paper capacity” trap and scale with confidence.
