Login
Blog
Lean Manufacturing Tools: 17 Tools, What Each Fixes and How OEE Tests It
In short: lean manufacturing tools are simple shop-floor methods, such as 5S, kanban, SMED and poka-yoke, and each one removes a certain kind of waste. Pick the tool by the biggest loss you can measure, then check OEE or lead time to see whether it worked.
- There are 17 common tools below, grouped by the job they do.
- Tools that cut stops, short stops or rejects show up in OEE.
- Tools that cut queues and stock show up in lead time instead.
- Test one change at a time, on the same line, with the same way of counting.
Most lists of lean tools stop at the name. A plant then runs a 5S campaign because it comes first on the list, and the line loses the same hours as before.
This page says what each tool fixes, shows it on a real floor, and names the number that should move if it worked.
For what lean is and where it came from, start with lean manufacturing principles. This page is about choosing and testing the tools.
17 lean manufacturing tools at a glance
OEE (overall equipment effectiveness) splits lost time into availability, performance and quality.
The last column says where each tool's result should appear. That column is our reading, and your line may differ.
| Lean tool | What it fixes | Where the result shows |
|---|---|---|
| 5S | Searching, clutter, hidden problems | Mostly indirect: faster changeovers and checks |
| Visual management | Problems nobody can see | Time until someone reacts |
| Andon | Slow reaction to a problem | OEE availability, shorter stops |
| Gemba walk | Decisions made from the office | Indirect: better choices of what to fix |
| Value stream mapping | Not knowing where the time goes, order to delivery | Lead time, stock |
| Standard work | Each shift doing the job differently | OEE performance and quality, less gap between shifts |
| TPM | Repeat faults and small stops | OEE availability and performance |
| Takt time | Running ahead of demand, or behind it | Output against demand |
| One-piece flow and cells | Batches waiting in queues | Lead time, work in progress |
| Kanban | Making too much, and running out | Stock, and OEE availability if shortages stop the line |
| Heijunka | Uneven load and long batches | Finished stock, lead time |
| SMED | Long changeovers | OEE availability, smaller batches |
| Poka-yoke | Mistakes that become defects | OEE quality |
| Jidoka | Defects passed down the line | OEE quality, fewer defects downstream |
| 5 Whys | Fixes that do not hold | The same stop reason shrinks and stays small |
| PDCA | Changes nobody checks | Result against the target you set |
| Kaizen | Waiting for one big project | The loss you aimed at, before and after |
No list is official. We left out just-in-time, which kanban and takt serve, and A3, which is a one-page PDCA report.
Start with the loss, not the tool
Here is one line for one shift. All numbers are example data, with the same ideal rate on every product.
| Example data | Value |
|---|---|
| Planned production time | 480 min |
| Ideal rate | 50 parts a minute |
| Stops: 2 changeovers of 30 min, no material 36 (counted as lost planned time here, some plants leave it out), machine faults 24 | 120 min |
| Run time | 360 min |
| Total parts | 15,300 |
| Good parts | 14,994 |
- Availability: 360 / 480 = 75.0%
- Performance: 15,300 / (360 x 50) = 85.0%
- Quality: 14,994 / 15,300 = 98.0%
- OEE: 75.0% x 85.0% x 98.0% = 62.5%. As a check, 14,994 / (480 x 50) gives the same.
The method is in our OEE calculation guide. To choose a tool, turn each loss into minutes at the ideal rate and sort them, as in a Pareto of downtime.
Look at the second bar. Nobody logged those 54 minutes, because they are jams cleared in seconds and a speed someone turned down.
So the first job on this line is changeovers, with short stops close behind. A 5S campaign on the whole plant would not be the obvious place to begin.
Minutes are a starting point, not the whole answer. Rejects also cost material, so 6 minutes understates them, and a gained minute only pays in full on your constraint.
Test one change at a time
Now run three separate tests on the same example line, with speed and rejects unchanged. Each test starts again from 62.5%.
- 5S in the stores: the area looks better and the audit score rises, but 120 minutes of stops stay 120. Done at the machine, the same tidy, marked tools help make the shorter changeover in test 2 possible.
- SMED: run time goes from 360 to 384 minutes, so availability is 80.0%. OEE is 80.0% x 85.0% x 98.0% = 66.6%, and good parts go from 14,994 to about 15,994.
- Kanban: run time goes from 360 to 378 minutes, so availability is 78.8%. OEE is 65.6%, and good parts go to about 15,744.
Three rules keep a test honest. Use the same line and products, count the same way over enough shifts, and agree the target before you start.
More parts only pay if someone needs them. Our page on OEE in lean manufacturing shows when a rising number hides waste.
Tools that help you see the problem
The plant results in the next sections are reported by the plants or the programs that helped them, read 9 November 2026. None has a control group, so treat them as what can happen, not as a forecast.
5S
5S stands for sort, set in order, shine, standardize and sustain. NIST says it arranges tools and parts so "required tools are visible with their proper placement self-explanatory".
On the floor: a shadow board at the machine, with one spot for each tool. Current Applications, a motor maker, worked on 5S together with cell design and standard work before it tested single-piece flow.
Visual management
The Lean Enterprise Institute (LEI) describes tools, parts and indicators placed in plain view, so the status of the system "can be understood at a glance by everyone involved".
On the floor: a board at the line showing plan against actual, by hour. NIST's story on Lynn Welding quotes operations manager Ibrahim Ibrahim: "It's become a habit to check the scheduling boards to help keep the jobs flowing."
Andon
An andon is a signal, such as a lamp or a board, that calls for help when something is wrong. LEI says it lights when a machine sensor detects a problem, or when "an operator pulls a cord or pushes a button".
On the floor: at NUMMI, the Toyota and GM joint venture, GM colleagues asked if workers would get the right to stop the line. Toyota's answer, in John Shook's account: no, "the obligation to stop whenever they find a problem".
Gemba walk
Gemba means the place where the work happens. A walk means going there to look and ask before you decide.
On the floor: stand at the line during a changeover with a stopwatch. LEI gives the reason: standardized work "cannot be written at a desk in the engineering office".
Value stream mapping
A value stream map draws every step and every wait, from order to delivery. NIST calls it "Often the first step in applying lean process improvements".
On the floor: With CONNSTEP, its local MEP center, Lynn Welding mapped one fabricated product, and the map showed issues as basic as incomplete order information. Use it first when customers complain about lead time, not about one machine.
Tools that keep the work stable
Standard work
Standard work sets the job sequence, the pace and the stock needed at a station. It is the baseline, because you cannot tell whether a change helped if each shift works differently.
On the floor: Lynn Welding wrote standard procedures for one welding booth, then applied them to the other booths.
TPM
TPM (total productive maintenance) makes machine losses everyone's job, not one department's. LEI gives the daily version: "operators perform daily activities such as lubricating, cleaning, tightening, and inspecting equipment".
On the floor: a five-minute check at shift start, on the machine with the most repeat faults. Then see whether fault minutes on that machine fall.
Tools that make work flow
Takt time
Takt is the pace demand asks for: available time divided by demand. LEI's example: 480 minutes a day and 240 widgets a day give a takt time of two minutes.
On the floor: a counter at the line showing where it should be by now. Running well ahead of takt is not a win, because the extra parts go to stock.
One-piece flow and cells
Instead of moving big batches between departments, you place the steps close together and pass one piece at a time. NIST lists reduced lead time and inventory among the benefits.
On the floor: Current Applications, a motor maker of about 60 people, improved cell design, standard work and 5S, then moved two product lines to single-piece flow. NIST reports lead time down 50% and work in progress down from 105 units to 5.
Kanban
A kanban is a signal, often a card, that says make or move more of this part. NIST sums it up: it "pulls inventory when needed" and "replenishes what has been consumed".
On the floor: an empty bin at the line goes back to the supplying step, and that bin is the order. If the supply step keeps failing, fix that first, or cards only show the shortage faster.
Heijunka
Heijunka levels the type and amount of production across a period. LEI's example: orders of 500 a week arrive unevenly, so the plant keeps a small finished stock and makes 100 a day.
On the floor: it mixes products in short repeats instead of long runs. That means more changeovers, so SMED usually comes first.
SMED
SMED (single minute exchange of die) shortens changeovers by moving work out of the stop, then trimming what is left. NIST measures it from "the last good piece off the current run" to the first good piece off the next.
On the floor: at Accurate Manufactured Products Group, a quick changeover project took average setup from 2 hours 15 minutes to 40 minutes, about 70% shorter. Purdue MEP says this came from "procedural changes and not due to any monetary changes".
Modelcraft mapped its changeovers and found 12 no-cost or low-cost fixes. With those and its other changes, NIST reports 15% more output from the milling department.
Tools that stop defects
Poka-yoke
Poka-yoke means error-proofing with simple, cheap devices. LEI's strongest type is a shutdown: "a machine will not start if a work piece is set incorrectly".
On the floor: a fixture shape that only accepts the part the right way round. Then count that one defect type, not total scrap.
Jidoka
Jidoka gives machines and operators the means to detect a problem and stop work at once. LEI traces it to Sakichi Toyoda's loom, which "stopped automatically when any thread broke".
On the floor: a sensor stops the press when a part is missing, so the fault stays at the step that caused it. Expect more visible stops at first, so availability can dip.
Judge it by fewer defects further down the line.
Tools that fix problems for good
5 Whys
Keep asking why until you reach the cause you can remove. In Taiichi Ohno's example, a blown fuse traces back to a pump with no strainer, which let metal scraps in.
On the floor: take the top stop reason from last week, and ask on the day it happens. If the same reason is back next month, the answer stopped too early.
PDCA
Plan, do, check, act is the loop behind every tool here. LEI says "If the team has achieved a measurable gain, it standardizes" the new method by updating standard work.
On the floor: many plants skip check. Write the target on the board before the change, in minutes, with a date to look again.
Kaizen
Kaizen means continuous improvement: many small changes, made by the people who do the work. It can be a daily habit or a focused event.
On the floor: a small team takes one named loss for a few days and changes what it can. Unilux, a boiler maker of about 40 people, used kaizen events after a value stream map, and delivery went from 12 to 14 weeks down to 6 to 10.
What OEE will not show
OEE looks at a machine or a line, and it does not see queues between steps. So a value stream map or a cell can be a big win with little change in OEE.
Heijunka adds changeovers, so OEE can even fall.
So give each tool the right test. Read lead time, stock and on-time delivery next to OEE, as we explain in OEE in lean manufacturing.
Common mistakes
- Running tools as a campaign. A 5S month with no named loss behind it fades when the month ends.
- Copying the order from a list. A plant losing hours to faults gains little from starting with heijunka.
- Trying three tools at once. The number moves, and nobody knows why.
- Counting activity. Events held and cards printed are not results. Minutes and lead time are.
- Skipping the baseline. Without a true count from before, every result is an opinion.
How Fabrico fits
Fabrico is a cloud-based OEE and light MES platform. It does not run your lean tools, but it gives you the baseline and the check for the ones that aim at line losses.
- Stops: they come from the machine signal, through a PLC or a Fabrico IoT device, including stops too short to log by hand.
- Reasons: the operator adds the reason on a tablet or phone, so your stops sort by minutes. Slow running and rejects show in performance and quality.
- OEE: availability, performance and quality by line and shift, so a before and after uses one way of counting.
- Parts: AI cameras (an add-on) count parts.
- Video: when you review a stop, you can open the camera video for that minute. In the standard setup it is streamed, not stored in Fabrico.
- Maintenance: built in on the same machine data, as a complement.
Your lean board and your ERP stay where they are, and production planning and scheduling are not part of Fabrico today.
Ficosota (also a Fabrico investor) raised OEE from 53.1% to 59.6% on 16 lines, June to September 2026, with monthly good output up 23.2%.
New to the measure? Start with what OEE is.
If cameras fit your line, ask us to show them in the demo.
Want to see stops, reasons and OEE on lines like yours? Book a demo or Contact us.
What people ask next
What are the main tools of lean manufacturing?
The ones most lists share are 5S, value stream mapping, kanban, SMED, poka-yoke, jidoka, andon, standard work, TPM and kaizen.
No list is official, and no plant needs them all at once.
Which lean tool should we start with?
If customers complain about lead time, start with a value stream map. If the line loses hours, start with the biggest named loss and take the tool that fits it.
When you have no numbers at all, a gemba walk and a simple stop log come first.
What is the difference between lean tools and lean principles?
Principles say what you are trying to do, such as create flow. Tools are how you do it, such as a cell or SMED.
Our page on lean manufacturing principles covers the five.
Is Six Sigma a lean tool?
Not usually. Six Sigma is a separate method that uses statistics to reduce defects and variation, and many plants run both.
Our page on lean vs Six Sigma compares the two.
Do lean tools work in a small plant?
Yes. Two of the plants on this page have about 40 and about 60 people.
Small plants often start with low-cost changes, such as the procedure changes behind the shorter setups above.
Do we need software to use lean tools?
No. Most of them started with cards, boards and paper, and a walk costs nothing.
Software helps with counting and follow-up. Our list of lean manufacturing software compares tools for that job.
Curious which loss to go after first on lines like yours? Book a demo or Contact us.
Related articles
Real Fabrico tiles, sample data.