Morsa publishes this guide. Morsa is software with an AI agent that works in your ERP, email and Teams; it writes the morning plan-versus-actual brief with the reason behind each miss. Benchmarks are from the named sources; the worked day is ours.
Quick answer
Twelve KPIs, five groups. Start with 12; each extra number takes airtime from the red ones.
Three that matter most for a job shop or assembly plant: schedule attainment, on-time delivery and open material shortages, because together they say whether customers get their orders.
Published benchmarks: manufacturing’s recordable injury rate was 2.7 per 100 full-time workers in 2024 [1]; measured OEE (overall equipment effectiveness) averages 55 to 65% [3][4]; manufacturing absence ran 2.9% in 2025 [5]. Everything else is measured against your own last 13 weeks.
Cadence: yesterday’s 12 daily, 13-week trends and cost dollars weekly, injury rate and OEE by machine monthly.
The 12 KPIs: formula, data source and target
First pass yield is the share of parts good the first time, with no rework.
1. Injuries and near misses
Formula: Count yesterday. Monthly rate = recordable cases x 200,000 / hours worked [2]
Data source: Safety log
Target: Zero a day; manufacturing’s 2024 rate was 2.7 [1]
2. First pass yield
Formula: Units good the first time / units started
Data source: ERP inspection results or the nonconformance report (NCR) log
Target: Your 13-week average, then up
3. Scrap rate
Formula: Scrap cost / production cost
Data source: ERP scrap transactions
Target: Your 13-week average, then down
4. Schedule attainment
Formula: Operations finished by their scheduled date / operations scheduled
Data source: ERP operation completions
Target: 90% or better (our rule of thumb)
5. On-time delivery
Formula: Order lines shipped on time and in full / lines due
Data source: ERP shipments against promise date
Target: Your customers’ stated requirement, or 95% of lines to start (our rule of thumb). How retailers score it: OTIF guide
6. Past-due orders
Formula: Count and dollars of open lines past promise date
Data source: ERP open order report
Target: Falling every week
7. OEE on the bottleneck
Formula: Availability x performance x quality
Data source: Machine monitoring or the operator’s shift log
Target: Beat your own last quarter; measured averages are 55 to 65% [3][4]
8. Downtime by reason
Formula: Minutes stopped, by reason code
Data source: Downtime log or monitoring
Target: Top reason smaller than last week
9. Labor efficiency
Formula: Standard hours earned (the hours the routing says the finished work should take) / hours worked
Data source: ERP labor tickets
Target: Look into anything under 85% (our rule of thumb)
10. Overtime
Formula: Overtime hours / hours worked
Data source: Timeclock or payroll
Target: Your budget
11. Absenteeism
Formula: Hours absent / hours scheduled
Data source: Timeclock
Target: Against the 2.9% manufacturing average [5]
12. Open shortages
Formula: Jobs due in 5 days missing a component
Data source: ERP shortage report plus the buyer’s notes
Target: None without an owner and a date
Schedule attainment says whether the floor did what the schedule asked; on-time delivery says whether the customer got the order. A plant can hit 95% attainment and still ship late if the schedule was wrong.
A worked day at a 100-person plant
The plant: metal fabrication and assembly, 100 people over two shifts, 70 of them direct operators, 14 work centers. The laser is the bottleneck. Set up the board with these columns: KPI, yesterday, 13-week average, target, owner, due date. Here is Tuesday’s, showing Monday, with the meeting’s action in place of the last three columns.
KPI | Monday | What the meeting does |
|---|---|---|
Safety | 0 recordables, 1 near miss at dock 2 | Maintenance lead adds a guard rail by Friday |
First pass yield | 1,890 of 2,000 welds = 94.5% (13-week average 96.0%) | Quality checks the fixture on cell 3 today |
Scrap | $1,840 on $92,000 = 2.0% | Within average; no action |
Schedule attainment | 55 of 64 operations = 85.9% | 5 of the 9 misses were at the laser; see OEE |
On-time delivery | 38 of 41 lines = 92.7% | Planner calls the 3 customers before 10 AM |
Past-due orders | 9 lines, $62,400 | Down from 11 last Tuesday |
Laser OEE | 82% x 90% x 97% = 71.6% | 95 of 173 lost minutes were waiting for sheet steel |
Downtime, plant | 410 minutes: changeover 180, material 120, breakdown 110 | Material wait goes to purchasing |
Labor efficiency | 506 of 562 hours = 90.0% | No action |
Overtime | 42 of 562 hours = 7.5% | Saturday shift only if the laser stays behind |
Absenteeism | 40 of 560 scheduled hours = 7.1% | HR checks the week’s trend; 7.1% in one day is high against a 2.9% yearly average |
Open shortages | 3 jobs due this week; 1 PO has no supplier date | Buyer gets a confirmed date by noon |
The hours reconcile for the 70 direct operators: 70 x 8 = 560 scheduled, minus 40 absent, is 520 regular hours, plus 42 overtime, so 562 worked. Their finished work earned 506 standard hours. Laser availability is 787 minutes run of 960 planned across two 8-hour shifts, which is 82.0%.
Two red items, the laser and the shortage, share one cause: material that did not arrive. That is why shortages sit on a production board. The meeting runs 15 minutes, standing. Green rows get no airtime. Every red row leaves with one owner and one due date, written in the sheet.
How to build it from the ERP and one spreadsheet
You do not need new software for the first version. Any ERP that can save a report can feed this.
Save five ERP reports: operations completed yesterday against schedule, shipments against promise date, open order lines past due, scrap and labor transactions, and the shortage report. Schedule them to land in a shared folder at 5 AM; if your ERP cannot, the first-shift lead runs them at 6:30.
Make one workbook, with one tab per report and a Today tab laid out like the board above.
Add three hand-kept tabs: safety and near misses, downtime with a reason code (unless machines report it), and absences. Our machine downtime log has a starter layout, and shop floor data collection covers what operators should record.
Color by rule, not mood: red when yesterday is worse than both target and the 13-week average; amber when it misses only one.
No history yet? Export the last 13 weeks of the five reports on day one as the baseline for the ERP rows. The hand-kept tabs start empty: use the published benchmark where one exists and set the rest after four weeks.
No machine monitoring? Use a shift log. The bottleneck operator notes each stop over 5 minutes with its reason and length, plus good and total parts at shift end. That gives availability, performance and quality for the day; the OEE guide has the method.
Cadence | What you look at |
|---|---|
Daily, 15 minutes | Yesterday’s 12 against target; red rows get an owner |
Weekly, 30 minutes | 13-week trends; scrap and overtime dollars; the top downtime reason and the top scrap part |
Monthly | Recordable injury rate; OEE by machine |
Getting the reason behind each miss before 7 AM
A board shows the laser lost 173 minutes, not why. Morsa writes the plan-versus-actual brief from whatever ERP you run, whether Epicor Kinetic, JobBOSS2, Global Shop, Business Central or another.
It reads yesterday’s completions, shipments and shortages and fills in the board.
For each red row it adds the reason from shift notes, Teams messages and the buyer’s email: “sheet steel PO 4471 arrived 2 days late”.
It proposes an owner and due date for each; the plant manager approves or changes them in the meeting, and Morsa follows up until each is done.
At J4S, a 120-person glass plant, on-time completion of operational commitments (the owner-and-date items a meeting like this produces) rose from about 30% to about 75% across about 900 commitments in four weeks (J4S customer story). To see the brief on your own plant’s data, book a 30-minute walkthrough.
What to do this week
Schedule the five ERP reports to land by 6:30 AM, and export the last 13 weeks of each for the baseline.
Build the workbook: the Today tab plus the three hand-kept tabs.
Run the 15-minute meeting for five days, and change only the targets that are clearly wrong.
Sources
US Bureau of Labor Statistics, Table 1, Incidence rates of nonfatal occupational injuries and illnesses by industry and case types, 2024 (manufacturing, NAICS 31-33: 2.7 total recordable cases per 100 full-time workers): https://www.bls.gov/web/osh/table-1-industry-rates-national.htm (read 8 November 2026)
US Bureau of Labor Statistics, How to compute a firm’s incidence rate (“200,000 hours … represents the equivalent of 100 employees working 40 hours per week, 50 weeks per year”): https://www.bls.gov/iif/overview/compute-nonfatal-incidence-rates.htm (read 8 November 2026)
Hedman, Subramaniyan and Almström, Analysis of Critical Factors for Automatic Measurement of OEE, Procedia CIRP 57, 2016 (average 65% across 884 machines at 23 companies): https://doi.org/10.1016/j.procir.2016.11.023 (read 8 November 2026)
Evocon, World-class OEE: industry benchmarks from more than 50 countries (3,500+ machines, May 2023 to June 2024; most plants “closer to 55-60%”; vendor data): https://evocon.com/articles/world-class-oee-industry-benchmarks-from-more-than-50-countries/ (read 8 November 2026)
US Bureau of Labor Statistics, Current Population Survey Table 47, Absences from work of employed full-time wage and salary workers by occupation and industry, 2025 (manufacturing absence rate 2.9%; 11-month average): https://www.bls.gov/cps/cpsaat47.htm (read 8 November 2026)
Morsa, J4S customer story: https://morsa.ai/customers/j4s (read 8 November 2026)
