15 KPIs Every CNC Machine Shop Should Track

You're staring at three red lights on your Mazak, your best operator just called in sick, and the customer wants their parts tomorrow. Sound familiar? Most shops run on gut feeling and prayer, tracking whatever their ERP system spits out. That's like navigating with a broken compass.

After fifteen years troubleshooting everything from ancient Bridgeports to brand new DMG Moris, I've seen shops burn cash because they weren't watching the right numbers. The difference between profitable shops and struggling ones isn't better machines or cheaper labor. It's knowing which metrics actually matter and acting on them before problems explode.

Here are the 15 KPIs that separate the pros from the wannabes.

Machine Availability and Reliability KPIs

1. Overall Equipment Effectiveness (OEE)

OEE combines availability, performance, and quality into one brutal number. World-class manufacturing aims for 85%, but most shops I visit run between 40-60%. That's money bleeding out the door.

Calculation: OEE = Availability × Performance × Quality

Break this down by machine and shift. Your Haas VF-2 might run 78% on day shift but drop to 52% on nights because the setup guy doesn't know the work offsets. Track it weekly, not monthly. Monthly averages hide the problems you need to fix.

Real-world example: A shop running five Mazak Integrex machines tracked OEE by individual spindle. They discovered Machine #3's B-axis was causing 15-minute delays every tool change due to a worn coupling. That single repair boosted their OEE from 61% to 74%.

2. Mean Time Between Failures (MTBF)

MTBF tells you how long your machines actually run before something breaks. Calculate this separately for critical systems: spindles, tool changers, coolant pumps, and axis drives.

Formula: MTBF = Total Operating Hours ÷ Number of Failures

Track failures by system, not just "machine down." A Fanuc Alpha spindle should run 8,000-12,000 hours between major issues. If you're seeing failures every 3,000 hours, you've got lubrication, contamination, or programming problems.

Pro tip: Don't count operator errors or normal tool wear as "failures." A broken drill bit isn't a machine failure. A seized Z-axis ball screw is.

3. Mean Time to Repair (MTTR)

MTTR measures how fast you fix problems. This separates good maintenance teams from great ones. Sub-30-minute repairs keep production flowing. Four-hour repairs kill schedules.

Calculation: MTTR = Total Downtime ÷ Number of Repairs

Break this down by failure type. Alarm code resolution should average under 10 minutes with proper documentation. Having quick access to alarm code databases cuts diagnosis time by 60-70%.

Hydraulic pump replacement on a Mazak HCN-5000 should take 45 minutes if you have the right pump in stock and procedures documented. If it's taking three hours, you need better preparation.

Production Performance KPIs

4. Spindle Utilization Rate

Most shops think they know how hard their spindles work. They're usually wrong by 30-40%. True spindle utilization measures actual cutting time versus available time.

Formula: (Actual Cutting Hours ÷ Available Hours) × 100

Track this via your machine's internal counters. Fanuc controls store this in parameter 3111-3116 (spindle load monitoring). Mazak Smooth controls show it in the maintenance screen under "Spindle Load History."

A $300,000 horizontal machining center running at 35% spindle utilization is a very expensive paperweight. Anything below 60% means you've got programming, setup, or scheduling problems.

5. Tool Life Performance

Tools that die early kill profits. Track actual tool life versus manufacturer specifications. A Sandvik 316 carbide end mill rated for 45 minutes of aluminum cutting should deliver 40-50 minutes in practice.

Metrics to track:

Tools lasting less than 80% of rated life
Premature tool failures by material type
Tool cost per part
Tool change frequency by operation

Real example: A shop making aluminum brackets noticed their 1/2" end mills were lasting 15 minutes instead of the rated 45. Root cause was excessive spindle runout (0.0008" vs. specification of 0.0002"). Spindle rebuild cost $4,500 but saved $800/month in tool costs.

6. Setup Time Variance

Setup time kills small-batch profitability. Track actual setup time versus standard for each job. Standard setup for a typical 4-axis part on a Haas UMC-750 should be 25-35 minutes with proper work holding and documented procedures.

What to measure:

Time from last part of previous job to first good part of current job
Setup time by operator (reveals training gaps)
Setup time by part complexity
Variance from standard setup time

Setups taking 50% longer than standard indicate missing tooling, poor documentation, or inadequate operator training.

Quality and Scrap KPIs

7. First Pass Yield

First Pass Yield measures parts that pass inspection without rework. Anything below 95% indicates process control problems. Track this by machine, operator, and part family.

Formula: (Good Parts on First Try ÷ Total Parts Produced) × 100

Break down failures by cause: dimensional errors, surface finish, programming mistakes, or tool problems. A pattern of +0.0005" errors on bore diameters might indicate systematic spindle growth issues, not random variation.

8. Scrap Rate by Root Cause

Total scrap rate matters, but root cause analysis prevents repeat failures. Categories that matter:

Programming errors (wrong offsets, feed rates, speeds)
Setup mistakes (work holding, tool measurement, part orientation)
Material defects (hard spots, inclusions, wrong alloy)
Machine malfunctions (lost position, thermal growth, tool changer errors)
Operator errors (wrong tools, missed steps, measurement mistakes)

Track scrap costs, not just percentages. A 2% scrap rate on $10 titanium blanks hits differently than 2% on $500 forgings.

9. Process Capability Index (Cpk)

Cpk measures how well your process stays within specification limits. Values above 1.33 indicate good control. Below 1.0 means you're making scrap.

Calculate separately for:

Critical dimensions (bearing fits, sealing surfaces)
Each machine and operator combination
Different part materials and geometries

Use statistical process control (SPC) software, not Excel. Real-time Cpk monitoring catches process drift before you make scrap parts.

Cost and Efficiency KPIs

10. Cost per Operating Hour

This metric reveals true machine profitability. Include everything: depreciation, maintenance, tooling, energy, and operator wages.

Typical ranges:

Small 3-axis mills: $85-120/hour
4-axis horizontals: $150-220/hour
5-axis machines: $200-350/hour

Track actual costs monthly. Energy costs vary seasonally. Maintenance costs spike during spindle rebuilds or major component replacements. Factor these into your hourly rates.

11. Maintenance Cost as Percentage of Machine Value

Annual maintenance should run 8-12% of machine replacement value for machines under 10 years old. Above 15% indicates either abuse or end-of-life issues.

Break down by category:

Planned maintenance (PM, lubrication, calibration): 40-50%
Corrective maintenance (breakdowns, failures): 30-40%
Upgrades and improvements: 10-20%

A five-year-old $400,000 horizontal costing $80,000/year in maintenance has serious problems. Either fix the root causes or replace the machine.

12. Energy Cost per Part

Energy efficiency matters more with rising utility costs. Modern machines use 15-25 kW during heavy cutting, 8-12 kW during rapid moves, and 3-5 kW in standby.

Monitor:

kWh per part by material type
Peak demand charges (often 30-40% of energy bills)
Power factor penalties
Standby power consumption during nights/weekends

Leaving machines in ready mode overnight costs $50-80/month per machine. Proper shutdown procedures save real money.

Maintenance and Reliability KPIs

13. Planned vs. Emergency Maintenance Ratio

World-class operations run 80% planned maintenance, 20% emergency repairs. Most shops I visit run 50/50 or worse. Emergency repairs cost 3-5x more than planned maintenance and always happen at the worst time.

Track separately:

Preventive maintenance (scheduled intervals)
Predictive maintenance (condition-based)
Emergency repairs (unplanned downtime)
Corrective maintenance (known issues, scheduled repairs)

Use maintenance management software with proper work order tracking. Spreadsheets don't cut it past 5-6 machines.

14. Work Order Completion Rate

This measures maintenance team effectiveness. Target 95% completion within scheduled timeframes. Lower rates indicate poor planning, parts availability issues, or inadequate staffing.

What causes delays:

Parts not in stock (60% of delays)
Inadequate job planning (25% of delays)
Technical skill gaps (10% of delays)
Production schedule conflicts (5% of delays)

Track completion rates by maintenance type and technician. Some people are better at electrical troubleshooting, others excel at mechanical repairs. Match skills to tasks.

15. Spare Parts Inventory Turnover

Parts inventory should turn 3-4 times annually. Higher turnover risks stockouts. Lower turnover ties up cash in dead inventory.

Critical spares to monitor:

Spindle cartridges and seals
Ball screws and linear guides
Servo drives and amplifiers
Tool changer components
Coolant pumps and filters

Stock based on failure history, not manufacturer recommendations. A machine running aluminum all day needs different spare parts than one machining Inconel. Use maintenance tracking platforms to optimize inventory based on actual failure patterns.

Key Takeaways

Stop tracking vanity metrics that look good in reports but don't drive decisions. Focus on these 15 KPIs:

Immediate impact metrics: OEE, MTTR, and First Pass Yield. Fix these first.

Strategic planning metrics: MTBF, maintenance cost ratios, and energy consumption. These guide long-term decisions.

Operational efficiency metrics: Setup time, spindle utilization, and tool life. These determine daily profitability.

Financial health metrics: Cost per hour, spare parts turnover, and planned vs. emergency maintenance ratios. These predict future performance.

Track weekly, analyze monthly, act immediately. The best KPI system in the world won't help if you don't act on what the numbers tell you. Start with three metrics, get them right, then add more. A few KPIs tracked religiously beat dozens tracked poorly.

Your competition is already measuring these numbers. The question is whether you'll join them or keep flying blind.