---
title: "Day shift vs night shift on the same machine"
description: "Night shift often makes fewer parts for reasons that have little to do with speed. Normalize first, compare four numbers, then tell staffing from support."
category: "Downtime and utilization"
author: "Ulises Carrasco"
published: "2026-10-09"
updated: "2026-11-02"
source_url: "https://www.iotflows.com/blog/compare-shifts-on-one-machine/"
---
> Question: Why does night shift make fewer parts on the same machine?

## Key takeaways

- Remove differences in shift length, breaks, planned downtime and job mix before comparing a night shift's output with a day shift's.
- Compare uptime, stop count, stop reasons and parts per running hour, because a raw part count mixes four different causes.
- Fewer but longer stops on nights point to missing support, such as maintenance cover or material staged by day.
- Fewer parts per running hour at the same pace against ideal cycle time is a job-mix difference, not a slower crew.
- Use a shift comparison to fix how the night shift is supported and scheduled, never to rank the people on it.

Night shift makes fewer parts on the same machine mostly because it works a different shift, not a slower one: fewer planned hours, longer stops while it waits for help that went home, and the long-cycle jobs left for nights. Remove shift length, breaks, planned downtime and the job mix first. Then compare uptime, stop count, stop reasons and parts per running hour, and the real gap is smaller and easier to fix.

This guide is for plant managers who see a night-shift shortfall on a report and need to know what it means before acting on it. It works one machine’s week two ways and ends with a guide that separates a staffing problem, a support problem and a job-mix problem.

## Why is a raw part count an unfair comparison?

A raw part count mixes four things that differ between shifts: how long each shift is planned to run, how much of that time the machine ran, which jobs the shift was given and how fast those jobs run. Only one of them, how much of the planned time the machine ran, is mostly decided on the shift itself.

For example, a day shift with 44 planned hours a week and a night shift with 37.5 start 14.8% apart before anyone loads a part. A night crew that matches the day crew minute for minute still makes fewer parts.

> **Definition.** **Parts per running hour** is the number of parts a machine makes in a shift divided by the hours it actually ran, which takes stopped time out of the comparison and leaves the pace of the jobs that shift ran.

Its partner is *uptime*: running time divided by running plus stopped time inside the shift. The [uptime glossary entry](/docs/get-started/glossary/#uptime) defines it the same way. Uptime says how much the shift ran, and parts per running hour says how fast it ran.

## What should you normalize before comparing shifts?

Take out every difference the shift didn’t choose. Each row below changes the count without saying anything about how the shift ran.

| Difference | What it does to the count | How to remove it |
|---|---|---|
| Shift length | A longer shift makes more parts | Compare per planned hour |
| Breaks | Longer breaks shrink planned time | Take breaks out of planned time |
| Planned downtime | A PM window lands on one shift | Take scheduled-off time out |
| Job mix | Long cycles mean fewer parts | Compare earned hours |
| Midnight | A night lands on two dates | Report it on the day it began |
| Sensor offline | Missing data reads as a stop | Keep offline time out of both |

Planned time is the shift less its breaks and any window with the machine scheduled off.

*Earned hours* are the parts a shift made multiplied by the ideal cycle time of each part. They measure output in hours of work at the ideal pace, so a 6-minute housing earns more than a 3.6-minute bracket and the job mix drops out.

Roll several nights into one figure by adding hours and dividing once. Averaging nightly percentages gives a short night the same weight as a full one, and [pooling hours](/blog/what-is-machine-utilization/#roll-up) shows the difference on two shifts of unequal length.

Compare the same weeks for both shifts, at least four of them. One breakdown can swing a single week by several points, and it lands on whichever shift was running.

## Which four numbers should you compare?

Compare uptime, stop count and average stop length, stop reasons and parts per running hour, shift against shift. Each answers a different question, and a night gap in each points to a different cause.

| Number | Question it answers | A night gap points to |
|---|---|---|
| Uptime | How much of planned time ran? | Stops, frequent or long |
| Stop count, average length | Many short stops or a few long? | Long: support. Many: staffing |
| Stop reasons | Where did the stopped hours go? | Waiting reasons: support |
| Parts per running hour | How fast did the jobs run? | Job mix, if pace to ideal holds |

Split uptime into stop count and stop length before reading it. For example, 810 stopped minutes can be 30 stops of 27 minutes or 81 stops of 10, and the two need different fixes.

Divide parts per running hour by the ideal rate for the job that ran. When both shifts run their own jobs at 90% of ideal, the pace gap is the job mix. The [guide to utilization, uptime, OEE and TEEP](/blog/utilization-uptime-oee-teep/) shows where uptime and that pace sit inside OEE.

## Work one week, day against night

Take one machine and one week, and run the comparison two ways.

#### Worked example: One week on a Haas VF-2, day against night

- Day shift: 06:00 to 16:00, Monday to Friday, with 60 minutes of breaks, so 9 planned hours a shift and 45 a week. A 1-hour preventive maintenance window on Wednesday, with the mill scheduled off, leaves 44 planned hours.
- Night shift: 20:00 to 04:00, Monday to Friday, with a 30-minute break, so 7.5 planned hours a shift and 37.5 a week.
- Day runs a bracket with an ideal cycle of 3.6 minutes. Night runs a housing with an ideal cycle of 6 minutes.
- Day: 35.2 running hours, 48 stops, 528 parts. Night: 24 running hours, 30 stops, 216 parts.

Night made 216 ÷ 528 = 40.9% of the day shift’s parts.

**Result:** Raw gap = 312 parts, 59.1% fewer on nights

### Way one: split the ratio into three factors

The night count divided by the day count is the product of three ratios: planned hours, uptime and parts per running hour.

**Night parts ÷ day parts** = Planned-hours ratio × uptime ratio × pace ratio

| Factor | Day | Night | Night ÷ day |
|---|---|---|---|
| Planned hours | 44 | 37.5 | 0.852 |
| Uptime | 80% | 64% | 0.800 |
| Parts per running hour | 15 | 9 | 0.600 |
| Parts | 528 | 216 | 0.409 |

Hypothetical machine. 0.852 × 0.800 × 0.600 = 0.409, the same as 216 ÷ 528.

Now read the pace against the ideal. The bracket’s ideal rate is 60 ÷ 3.6 = 16.7 an hour, and day ran 15, or 90%. The housing’s ideal rate is 10 an hour, and night ran 9, also 90%. The whole pace ratio is the job mix.

Figure: A waterfall from 528 day-shift parts to 216 night-shift parts: shorter planned time costs 78 parts, lower uptime 90 and the longer-cycle job 144. Only the 90-part uptime step is a difference in how the shift ran.

*Of the 312-part gap, 90 parts come from uptime; shift length and job mix explain the rest. Sample data.*

In parts, taken in that order, shift length costs 78, uptime costs 90 and job mix costs 144. Another order splits the parts a little differently, while the three ratios stay the same.

### Way two: compare earned hours

**Earned hours** = Parts × ideal cycle time

#### Worked example: Earned hours per planned hour

- Day: 528 × 3.6 minutes = 1,900.8 minutes, or 31.68 earned hours, over 44 planned hours = 72.0%.
- Night: 216 × 6 minutes = 1,296 minutes, or 21.6 earned hours, over 37.5 planned hours = 57.6%.

Night earns 57.6 ÷ 72.0 = 0.80 of the day shift’s output per planned hour, the same as the uptime ratio.

**Result:** Fair gap = 80% uptime against 64%, a ratio of 0.80

Both ways agree. Once shift length and job mix are out, the night shift runs its job as fast as the day shift runs its own, and the whole difference is 16 points of uptime.

### Where the stopped hours went

Night stopped for 13.5 hours in 30 stops, an average of 27 minutes. Day stopped for 8.8 hours in 48 stops, an average of 11 minutes. Fewer, longer stops are the first clue.

| Reason | Day (h) | Night (h) | Night − day (h) |
|---|---|---|---|
| Setup | 3.6 | 3.0 | −0.6 |
| Tool change | 2.2 | 1.5 | −0.7 |
| Waiting for maintenance | 0.8 | 5.0 | 4.2 |
| Waiting for material | 0.5 | 3.0 | 2.5 |
| Other | 1.7 | 1.0 | −0.7 |
| Total | 8.8 | 13.5 | 4.7 |

Figure: Stopped hours by reason for one week. Night waits 5.0 hours for maintenance against 0.8 on days, and 3.0 hours for material against 0.5. Setup, tool change and other stops are lower on nights.

*Two waiting reasons hold 6.7 extra hours on nights, more than the whole gap. Sample data.*

Bring the two waiting reasons down to the day shift’s level and night stops for 6.8 hours instead of 13.5. Uptime reaches 30.7 ÷ 37.5 = 81.9%, and the mill makes about 60 more housings a week at 9 an hour.

There is a second, smaller signal. Night’s 4 setups averaged 45 minutes against 27 for the day shift’s 8, so the night shift also sets up more slowly.

## What usually makes night shift slower?

Night is a minority schedule. In 2024–25, 15% of US wage and salary workers usually worked a non-daytime schedule and 3% worked nights; the release doesn’t break manufacturing out.[1] In many plants, the people a stopped machine waits for work days.

Each common cause leaves its own trace in the four numbers.

- **No maintenance cover.** A fault the day shift clears in 15 minutes waits for an on-call technician or for the morning. Look for few stops, long ones and a maintenance reason, often on a stop that runs to the end of the shift.
- **Material and tooling staged by day.** When bar stock, inserts or fixtures for the next job weren’t staged before day shift left, night waits or improvises. Look for waiting-for-material stops near the start of the shift or at a job change.
- **Fewer setters.** One setter covering six machines at night turns a 27-minute setup into a 45-minute one, and machines queue for them. Look for longer setups and several machines stopped at once.
- **Approvals only days can give.** A first-article sign-off, a program edit or a quality hold that needs an engineer. Look for stops coded waiting for quality or programming.
- **Longer-cycle jobs left for nights.** Long, stable runs go to nights so the crew has fewer setups. Look for fewer parts per running hour, uptime the same or higher and pace against ideal unchanged.

The minutes lost at the handover itself are a separate loss with their own fixes, covered in [shift changeover losses](/blog/shift-changeover-losses/).

## Is it a staffing, support or job-mix problem?

Read the four numbers against the columns below. Most night gaps match one column clearly, and some match two.

| Signal on nights | Staffing | Support | Job mix |
|---|---|---|---|
| Uptime | Lower | Lower | Same or higher |
| Stop count | Higher | Same or lower | Same or lower |
| Average stop length | Long setups | Much longer | Same |
| Top added reasons | Setup, no operator | Waiting: maintenance, material | None |
| Parts per running hour | Same | Same | Lower |
| First fix | Setter, cross-training | On-call cover, staging | Earned hours |

Compare one machine over the same four or more weeks, after normalizing.

> **Which problem do you have?**
>
> Treat it as support when night stops are fewer but longer and the added hours sit under waiting reasons. Treat it as staffing when stops are more frequent or setups run longer. Treat it as job mix when uptime matches and only parts per running hour differs.
>
> You don’t need to act when the gap is inside the normal week-to-week swing.

Each problem has a different first fix.

- **Support.** Put a technician on call with a response target, run a staging check before the day shift leaves (material, inserts, fixtures and programs for the next two jobs), and write down which first-article sign-offs night may make. On the example mill, that is the 81.9% uptime above.
- **Staffing.** Move setups into the day shift where the schedule allows, put a setter on nights where it doesn’t, and cross-train operators on routine changeovers. Coaching closes a skill gap; discipline doesn’t.
- **Job mix.** Keep the long runs on nights if they suit the shift, and judge each shift against a goal built from its own job. NIOSH suggests scheduling heavy or demanding work at times when workers are most alert.[2] A long, stable run suits a night shift; a string of setups and first-article checks suits the day. A per-shift goal is what makes [the mid-shift check](/blog/production-monitoring-shift-goal/) fair.

## Compare shifts automatically

IoTFlows sorts every stop, cycle and part count into the shift and day it fell in, using the [shifts and timezone](/docs/admin/shifts-and-timezone/) you set once. The Advanced Report’s **Shifts** filter counts only time inside the shifts you tick, and [Shift islands](/docs/monitoring/advanced-report/#shift-islands) reports a shift that crosses midnight whole on the day it began. The Downtimes report filters stop reasons by shift the same way.

Operator Performance totals a machine’s uptime for each clocked work session, and is meant for [finding sessions nobody closed](/docs/monitoring/operator-performance/#reading-fairly), not for ranking people. [Criollos Foods](/customers/criollos-foods/) tracked bagging-line rates across products and shifts, and reports a 35% productivity increase “through identifying and training operators running below the 30 bags/min target rate”. See [machine utilization tracking](/solutions/machine-utilization/).

## Before you present a shift comparison

Run this list before a day-against-night number goes into a meeting.

- **Same machine, same weeks,** at least four of them.
- **Planned hours per shift,** with breaks and scheduled-off windows removed.
- **The overnight shift reported whole** on the day it began.
- **Offline sensor time kept out** of both shifts.
- **Hours pooled** across nights, never nightly percentages averaged.
- **Four numbers side by side:** uptime, stop count with average length, stop reasons, parts per running hour.
- **Pace against each job’s ideal,** so job mix shows as job mix.
- **One named problem,** staffing, support or job mix, and one fix.

> **A shift gap is not a crew score**
>
> The night crew inherits the job mix, the staging and the maintenance cover. Present the comparison as how the shift is supported and scheduled, and bring the night lead into the review, because they know which waits the stop reasons don’t name.

## Questions

### Should night shift have the same production goal as day shift?

Use the same method and different numbers. Build each shift's goal from its own planned time, the downtime that shift is expected to have and the ideal cycle time of the job it runs. A night shift with fewer planned hours and a longer-cycle job has a lower goal, and it can meet that goal as often as the day shift meets its own.

### How many weeks of data do you need to compare two shifts?

Use at least four full weeks, the same weeks for both shifts. One breakdown can swing a single week by several points of uptime, and it can land on either shift. Over four weeks a one-off stop shows up as one bad week among four, while a gap that comes from how the shift is supported shows up in every week.

### How do you compare shifts when crews rotate?

Compare the time slot over a full rotation, so every crew has worked nights and days the same number of times. Any gap that remains belongs to the slot, meaning its support, staging and job mix, rather than to a crew. A gap that follows one crew around the clock is a training question, handled with coaching rather than a report.

### How do you report a night shift that crosses midnight?

Report the whole shift on the day it began. Split at midnight, a Friday night shift puts two hours on Friday and six on Saturday, so Monday reads short and Saturday shows output on a day nobody worked. Check which convention a report uses before comparing it with a day shift.

## Sources

1. [Job Flexibilities and Work Schedules Summary - 2025 A01 Results (2024-25 data)](https://www.bls.gov/news.release/flex2.nr0.htm), U.S. Bureau of Labor Statistics, accessed Oct 30, 2026
2. [NIOSH Suggests Approaches on Shiftwork to Reduce Worker Fatigue, Stress](https://archive.cdc.gov/www_cdc_gov/niosh/updates/shift.html), National Institute for Occupational Safety and Health (CDC), accessed Oct 30, 2026