How to Improve OEE: 12 Strategies and a 90-Day Plan

Key takeaways

• To improve OEE, convert each factor into lost minutes first. The lowest percentage is not always the biggest loss.
• Start on the bottleneck. An hour won there is an hour won for the whole line.
• Each OEE factor has its own levers: prevention and SMED for availability, short-stop fixes and speed restoration for performance, standard startup settings for quality.
• OEE improves when each top loss has an owner, an action and a date. A dashboard alone changes nothing.
• In the example below, a 10-point OEE gain gives 14.8% more good output from the same planned hours.

Why OEE improvement stalls

Many plants that start measuring OEE see a quick gain, then a plateau.

The reasons are usually the same:

• The data is not trusted, so every meeting turns into an argument about the numbers.
• Losses are reviewed monthly, when nobody remembers what caused them.
• Nobody owns the fix. The loss is visible on a chart, but no task is ever created.
• Effort goes to the wrong machine, one that is not limiting output.

The 12 OEE improvement strategies below remove those four blockers. Steps 1 to 3 fix the data and the focus, and step 12 fixes the review and the ownership.

If you need a refresher on the metric itself, start with our guide to overall equipment effectiveness.

Think in minutes, not percentages

OEE is a product of three percentages, so the lowest factor can mislead you.

Take a line with 84% availability, 83% performance and 97% quality. Its OEE is 67.6%.

Here is where the time goes on a shift with 450 minutes of planned production time:

Performance has the lowest percentage, but stops cost more time: 72 minutes against 64.

Performance losses only apply to the time the machine ran, so each point is worth fewer minutes. For the full method, see our OEE calculation guide.

How to improve OEE: build the foundation

1. Capture data automatically

Manual logs miss most short stops and tend to round durations to the nearest five or ten minutes.

Take stops and counts directly from the machine, through the PLC, a retrofit sensor or a camera, so the numbers no longer depend on memory.

2. Agree the definitions before the targets

Write down what counts as planned production time, the threshold for a short stop, and the ideal cycle time for each product.

If these change from week to week, your trend will move without anything changing on the floor.

3. Start on the bottleneck

Every line has one step that limits its output.

Improving a machine that is already waiting for material adds nothing. Our guide to bottleneck analysis shows how to find it.

Improve availability

4. Prevent the failures that repeat

Rank breakdowns by minutes lost and look for the machines and components that come back again and again.

Put those on a preventive schedule based on run hours or cycles, and train operators in basic care through autonomous maintenance.

5. Shorten changeovers with SMED

Changeovers are planned stops, but they still reduce availability.

Separate the steps that need the machine stopped from the ones that can be done in advance, then move as many as possible outside the stop. See our SMED guide.

6. Cut the time from stop to restart

Every stop has at least two parts: the time until someone responds, and the time to repair.

Measure both. Clear escalation rules shorten the first, and spare parts plus machine history shorten the second.

Our guide to mean time to repair covers the repair side in detail.

For 15 ways to win back availability, read our playbook on how to reduce machine downtime.

Improve performance

7. Hunt down short stops

Short stops rarely appear in manual logs, yet they can cost as much as a breakdown.

Count them by station and fix the top three: a guide rail, a sensor position, a product that is slightly out of spec. Our micro stops guide goes deeper.

8. Find out why the speed was lowered

Machines often run below their ideal speed because someone once reduced it "to be safe".

Find the original reason. If the problem behind it is fixed, restore the speed and watch quality closely.

9. Stop starving and blocking the bottleneck

A machine waiting for material, or waiting for the next machine to take its output, loses time through no fault of its own.

Short waits show up as short stops, and longer ones as unplanned stops, so the loss can land in performance or availability.

Look at material flow, buffers and the speed of the machines around it.

Improve quality

10. Remove startup scrap

Many defects happen right after a changeover or restart, while settings are adjusted by trial and error.

Record the proven settings for each product and use them as the starting point every time.

11. Fix defects at the source

Check the process where the defect is created, not only at final inspection.

Simple control charts help you catch drift before it produces scrap. See statistical process control.

Make it stick

12. Review losses every day and turn them into tasks

Hold a short daily review at the line: yesterday's OEE, the top three losses, and one action for each.

Every action gets an owner and a due date, and is checked at every meeting until it is closed. Our daily OEE meeting guide gives a simple agenda.

Match the loss to the lever

For the full breakdown of each loss, read the six big losses.

For a side-by-side list of methods such as Pareto charts, 5 whys and TPM, see our guide to OEE improvement tools.

Worked example: what a 10-point gain is worth

The line above runs 240 shifts a year, or 1,800 planned hours. Over 90 days, the team works on its top losses.

• Fully productive time rises from 1,217 to 1,397 hours a year.
• That is about 180 extra productive hours from the same planned time.
• Good output rises by 14.8% (77.6 ÷ 67.6 = 1.148), with no new equipment and no extra shifts.

To put a value on hours like these, see our guide to the hidden factory.

An OEE improvement action plan template

Keep the plan on one page. One row per loss, reviewed every day.

The figures below are an example from a different line.

The minutes column keeps the list in the right order. The status column is what the daily meeting checks.

A 90-day OEE improvement plan

1. Days 1 to 30: connect the bottleneck, agree the definitions, and collect a trusted baseline. Fix only what is obviously broken.
2. Days 31 to 60: convert losses into minutes, pick the top three, and assign owners with due dates.
3. Days 61 to 90: turn every fix into a standard, such as a preventive task, a checklist or a setting sheet, then pick the next three losses.

After 90 days, extend the method to the next machine once the first one's gains have held for several weeks.

Common OEE improvement mistakes

• Chasing the lowest percentage instead of the biggest loss in minutes.
• Setting a world-class target on day one. The widely quoted 85% is a reference point, not a starting target (see world-class OEE).
• Improving OEE by changing the definitions, for example by moving changeovers out of planned time.
• Using OEE to rate operators. It leads to better-looking numbers, not better machines.
• Starting on ten lines at once. Prove the method on one, then spread it.
• Fixing without writing it down. A fix that is not turned into a standard comes back as a loss.

How Fabrico helps

Fabrico is a cloud-based MES and OEE platform for manufacturers, with maintenance management (CMMS) built in.

It collects stops and counts from PLCs, IoT sensors and AI cameras, calculates OEE in real time, and detects the micro stops manual logs miss.

Because the CMMS is part of the same platform, your team can turn a loss into a work order or a preventive maintenance plan, and track it until it is done.

The AI assistant answers questions about a machine's losses in plain language, and an Operational Assessment is available as an add-on.

See where your OEE is really lost: book a demo. For related metrics, read our guide to production efficiency.