Skip to content

Method guide · Value stream mapping and flow

Just-in-time (JIT) manufacturing: what it is, how it works and how to start

Short answer

Just-in-time (JIT) manufacturing is a production system in which each process makes and delivers only what the next process needs, when it needs it, in the quantity it needs. Production is pulled by actual use instead of pushed by a forecast, so inventory stays low and problems show up instead of hiding in stock.

Just-in-time manufacturing, usually shortened to JIT, is a way of running production so that parts arrive at each step exactly as that step needs them: not earlier, which creates inventory, and not later, which stops the line. It was built at Toyota and is one of the two pillars of the Toyota Production System, next to jidoka. In practice it means setting a pace from customer demand (takt time), making parts flow one at a time or in small lots, letting the next process pull from the one before it with kanban signals, and levelling the schedule so the pull is steady. The point is not zero inventory. The point is that every piece of inventory has a reason and a limit, so that when something goes wrong you see it the same day.

Last reviewed:

What is just-in-time manufacturing?

Just-in-time manufacturing is a production system that makes and moves only what is needed, only when it is needed, and only in the amount needed. The customer can be the person who buys the product or the next workstation on the line. Either way, nothing is made until that customer has used what it already had.

The same idea goes by several names. Just-in-time production, JIT production system and JIT manufacturing system all mean the production side: pacing, flow and pull on the shop floor. Just-in-time inventory and JIT purchasing mean the supply side: ordering so that material arrives shortly before it is used. This guide is about the production system. The supply side depends on it, because a supplier can only deliver to a schedule the plant can keep.

JIT is often described as a philosophy rather than a technique, and that is fair: the techniques (kanban, takt time, small lots, quick changeovers) are the means. The aim is to remove the waste that comes from making things before they are needed: the stock itself, the space it takes, the handling, the parts that go obsolete, and the problems that stay hidden because there is always another pallet to work from.

Our post Just-in-Time (JIT) Manufacturing: What It Is and How It Works goes deeper into what JIT eliminates and what it needs before it can function. Short definitions of each term are in the lean glossary.

Where JIT came from

Just-in-time was developed at Toyota in Japan in the decades after the Second World War. Toyota's own account credits Kiichiro Toyoda with the phrase and Taiichi Ohno with building the system on the shop floor. Ohno described it in his book Toyota Production System: Beyond Large-Scale Production, which our book summary covers. He called just-in-time and jidoka the two pillars of the system.

The reasoning behind it was practical. Toyota made small volumes of many models and could not afford the large stocks that mass production in the United States relied on. It needed a way to make a mixed set of cars in small quantities without building inventory, and the answer was to let each process pull exactly what it needed from the one before it. Ohno has said he took the pull idea from the way an American supermarket restocks its shelves as customers take goods.

The two pillars depend on each other. Just-in-time removes the inventory that hides problems. Jidoka, the practice of stopping at the first abnormality and fixing it, is what makes that safe: with little stock, a defect has to be caught where it happens or it stops the next process. Our jidoka guide covers that side, and the free TPS house template shows how the pieces fit together on one page.

How a JIT production system works

A JIT production system has four moving parts: a pace set by demand, parts that flow, a pull signal that authorises each move, and a levelled schedule that keeps the pull steady. Each one relies on the one before it, so this is also the order to put them in.

How a JIT system is built, in six parts

  1. 1

    Set the pace from demand: takt time

  2. 2

    Make parts flow: one-piece flow and cells

  3. 3

    Pull where flow is not possible: kanban and supermarkets

  4. 4

    Level the schedule: heijunka

  5. 5

    Shorten changeovers so small lots pay off

  6. 6

    Standardise and stop at problems

Takt sets the pace, flow and pull carry it, leveling and short changeovers make small lots possible, and stopping at problems keeps the whole thing honest.
  1. 1

    Set the pace from demand: takt time

    Takt time is the available production time divided by customer demand for that time. If customers need 400 pieces a day and the line runs 900 minutes, takt is 2.25 minutes a piece. Every step is then designed to finish one piece inside that time, no faster and no slower. The takt time calculator does the sum, and Takt Time in JIT explains how it paces a line.

  2. 2

    Make parts flow: one-piece flow and cells

    Where steps can sit next to each other, arrange them in process order so a part moves straight from one to the next instead of waiting in a batch. This is one-piece flow, and the U-shaped work cell is its usual layout. See One-Piece Flow in JIT, Cellular Manufacturing in JIT and the step-by-step cell design guide.

  3. 3

    Pull where flow is not possible: kanban and supermarkets

    Between steps that cannot be joined (a shared machine, a supplier, a different building) hold a small, capped stock called a supermarket. The downstream process takes what it needs; the kanban card or empty container that comes back is the only authority for the upstream process to make more. Pull System in Lean and Supermarket Pull System cover the design, and the kanban sizing calculator sets the number of cards.

  4. 4

    Level the schedule: heijunka

    Pull only works if demand on the upstream process is steady. Heijunka levels the volume and the mix over the day, so a cell makes a little of each product in a repeating pattern instead of a big run of one product followed by a big run of another. The heijunka box planner builds the pattern, and Heijunka in JIT explains why it protects takt.

  5. 5

    Shorten changeovers so small lots pay off

    Small lots mean more changeovers. If a changeover takes an hour, nobody will run a lot of twenty. SMED separates the setup work that can be done while the machine runs from the work that needs it stopped, then shortens both. See SMED in JIT Manufacturing and the SMED changeover calculator.

  6. 6

    Standardise and stop at problems

    Standard work fixes the sequence and the standard work in process at each station, so the pace holds from shift to shift. Andon and jidoka mean an operator stops and calls at the first defect or shortage, because with little inventory a problem passed on becomes a line stop downstream. Pitch in JIT Manufacturing shows how a regular pitch, takt times container quantity, gives the team a check every few minutes that the pace is holding.

The tools of a JIT production system and what each one does
ToolWhat it doesFree resource
Takt timeSets the pace of every step from customer demandTakt time calculator
One-piece flow and cellsMoves a part straight to the next step with no batch queueCell design post
KanbanAuthorises each make or move; caps the inventory in the loopKanban card template, kanban sizing calculator, kanban rules poster
SupermarketA small, capped stock between steps that cannot be joinedSupermarket pull post
HeijunkaLevels volume and mix so the pull is steadyHeijunka box planner
SMEDShortens changeovers so small lots are practicalSMED changeover worksheet and calculator
Standard work and pitchHolds the pace and shows within minutes when it slipsStandard work guide
Jidoka and andonStops at the first problem so defects are not passed onJidoka guide, andon guide

The resources are linked in the steps above and listed at the end of this page. The [kanban rules poster](/posters/kanban-rules-poster) puts Ohno's six rules of kanban on the wall where the cards are used.

JIT vs push production, lean and just-in-case

Three comparisons come up in almost every conversation about JIT. The first is the one that matters on the floor.

Push (traditional batch) production compared with just-in-time

  • Push, batch and queue

    What starts production
    A schedule built from a forecast
    Lot size
    Large, to spread the changeover cost
    Inventory between steps
    Whatever builds up
    Where a part waits
    In a pile in front of each machine
    When a defect is found
    Days later, when the batch is used
    Lead time
    Long, mostly waiting
    What a problem looks like
    Hidden, there is always more stock

  • Just-in-time

    What starts production
    A pull signal from the next process
    Lot size
    Small, made practical by short changeovers
    Inventory between steps
    Capped by the number of kanban
    Where a part waits
    Briefly, in a sized supermarket, or not at all
    When a defect is found
    Within minutes, at the next step
    Lead time
    Short, close to the work content
    What a problem looks like
    Visible, the line slows or stops
The same plant run two ways. Under push, stock hides problems; under JIT, a problem slows or stops the line, so it gets fixed. Batch Production: Meaning, Advantages and Disadvantages gives the case for batches where they still make sense.

JIT vs lean manufacturing

Lean manufacturing is the whole approach: identify value, remove waste, build flow, pull, and improve continuously. Just-in-time is the part of lean that deals with flow and pull. The term lean was coined later, by researchers describing what Toyota did, and it covers things JIT does not, such as problem solving, standard work, leadership routines and supplier development. If someone says they are doing JIT, ask whether they also have the quality and problem-solving side, because JIT without jidoka is a line that stops a lot.

JIT vs just-in-case

Just-in-case is the name given to the opposite policy: hold extra stock of everything so that a late supplier, a breakdown or a rush order never stops production. It feels safe and it is expensive, because the stock covers every problem equally whether or not the problem is likely. JIT does not say hold nothing. It says decide how much to hold and why, cap it, and work on the reasons it is needed. A safety stock set from a measured lead time and demand variation, as the safety stock calculator does, is JIT thinking. A pallet kept because it has always been there is just-in-case.

A worked example: one bracket cell moves from push to pull

This example is illustrative; the plant and the numbers are made up to show the arithmetic. A plant makes a steel bracket in four steps: laser cutting on a shared machine, then bending, welding and powder coating in a cell. Customer demand is 400 brackets a day. The cell runs two shifts of 450 minutes each, 900 minutes a day.

Before: push

The laser runs brackets once a week in a lot of 2,000, because its changeover takes 60 minutes and the planner wants to use the machine well. 2,000 brackets at 400 a day is five days of stock, sitting on pallets in front of the bending press. A cutting fault found at welding on Thursday has by then been copied into every bracket cut on Monday. The bending press works in lots of 200 for the same reason. Lead time from cut blank to coated bracket is about a week, almost all of it waiting.

Step 1: takt time

Takt time is available time divided by demand: 900 minutes divided by 400 brackets is 2.25 minutes, or 135 seconds, per bracket. Check it the other way: 400 brackets at 135 seconds is 54,000 seconds, which is 900 minutes. Every step in the cell has to finish a bracket inside 135 seconds.

Step 2: flow inside the cell

Bending, welding and coating are moved next to each other and a bracket passes straight from one to the next. The lot of 200 at the bending press disappears, and a bending fault is seen at the welding station a few minutes later instead of a day later. Our cell design guide walks through the station count and the shape of the cell.

Step 3: a kanban loop to the shared laser

The laser cannot join the cell, because it cuts parts for three other product families. So a supermarket of cut blanks is placed at the start of the cell and the laser only cuts brackets when a kanban comes back from it. The loop is sized from daily demand (400), the replenishment lead time of the laser (half a day, measured), a safety allowance of 10 percent and a container of 20 blanks.

Number of kanban = demand during the replenishment time, plus the safety allowance, divided by the container quantity. Demand during half a day is 400 times 0.5, which is 200 blanks. Plus 10 percent is 220 blanks. Divided by 20 per container is 11 kanban. Check it from the other end: 11 containers of 20 hold 220 blanks, and 220 divided by 400 a day is 0.55 of a day, a little over half a day of stock. That is the most that can ever sit in front of the cell, and it replaces five days of pallets.

Step 4: the changeover has to shrink

The laser now has to cut brackets every half day in small lots, with three other families in between, instead of once a week. Say it cuts a bracket blank in 90 seconds: 400 blanks a day is 36,000 seconds, or 600 minutes, of the laser's 900 available minutes. With four changeovers a day at 60 minutes each, that is another 240 minutes, 840 in total, which fits but leaves no room for a breakdown. After a SMED project brings the changeover to 12 minutes, four changeovers take 48 minutes and the laser needs 648 of its 900 minutes. Check: 600 plus 48 is 648, and 900 minus 648 leaves 252 minutes of capacity for the other families and for problems. The SMED changeover worksheet is the form for that project.

What changed

Illustrative before and after for the bracket cell
MeasureBefore (push)After (JIT)
Cut blanks waiting in front of the cellUp to 2,000 (five days)At most 220 (about half a day)
Lot size at the bending press2001 (one-piece flow)
Time for a cutting fault to be foundUp to a weekWithin the half-day loop
Laser changeover60 minutes, once a week12 minutes, four times a day
Lead time, cut blank to coated bracketAbout a weekUnder a day

The inventory fell from five days to about half a day: 2,000 to 220 blanks. Nothing was made faster. The waiting was removed. Use the [lead time calculator](/tools/lead-time-calculator) to check a loop like this with Little's Law: 220 blanks divided by 400 a day is the 0.55 of a day above.

Benefits and risks of just-in-time

The benefits follow from the example. Less inventory means less cash tied up, less floor space, less handling, and fewer parts that become obsolete when a design changes. Shorter lead time means a plant can promise faster and change the mix sooner. The least obvious benefit is the one Ohno cared most about: with little stock, problems show up quickly, so they get fixed instead of worked around.

The risks are the mirror image. A line with half a day of stock stops if a supplier is a day late, a machine breaks down, or a quality problem takes out a lot. JIT is fragile when it is applied to the supply chain before the plant can run it, when it depends on a single distant source with no agreed buffer, or when it is used as an excuse to cut inventory without fixing the reasons the inventory was there. The right answer in those cases is a sized buffer and a plan to remove its cause, not a return to pallets everywhere.

  • Stable quality at each step. A defect rate that was tolerable with a week of stock stops a line that holds half a day. Start with the quality at the source practices before cutting inventory.
  • Reliable equipment. Breakdowns that a buffer used to hide now stop the next step. Our maintenance guides cover the autonomous and preventive maintenance that JIT depends on.
  • Quick changeovers. Without them, small lots cost more than they save.
  • Suppliers who can deliver to the pull, or a sized buffer where they cannot yet.
  • People who will stop and call when something is wrong, and leaders who respond instead of blaming. That is the jidoka side, and it is covered in the andon guide.

How to start JIT in one area

Do not start with the whole plant. Pick one product family with steady demand and one cell or line, and run the sequence from the example. A model area proves the method and trains the people who will spread it.

  1. 1

    Map the current state

    Draw the value stream for the family: each step, its cycle time, the inventory in front of it in days, and the lead time. The waiting is where the gains are. Our value stream mapping guide covers the map.

  2. 2

    Calculate takt and check every step against it

    Any step slower than takt is the bottleneck and must be fixed or doubled before pull can work. Any step much faster than takt will overproduce unless it is capped.

  3. 3

    Join the steps that can be joined

    Move the steps that serve only this family into a cell and run one piece at a time. Agree the standard work in process between stations.

  4. 4

    Put a supermarket and kanban where steps cannot be joined

    Size the loop from measured replenishment time and demand, print the cards from the kanban card template, and put the kanban rules on the wall.

  5. 5

    Level the pull

    Build a repeating pattern of the family's products across the day with the heijunka box, and release work to the cell at that pitch.

  6. 6

    Fix what the lower inventory exposes

    Changeovers, breakdowns and defects will now stop the cell. That is the method working. Run each one as an improvement project, and when the cell is stable, take a kanban out of the loop and see what appears next.

Expect the first weeks to be uncomfortable. The cell will stop for reasons the old pallets used to hide, and people will ask for the stock back. Hold the inventory cap, fix the causes, and measure lead time and inventory every week so the gain is visible.

Common mistakes

  • Treating JIT as an inventory target. Cutting stock without fixing changeovers, breakdowns and defects produces line stops, not flow.
  • Starting with suppliers. Asking a supplier for daily deliveries before the plant can consume them daily moves the pallets to the dock.
  • Skipping takt time. Without a pace from demand, a cell runs as fast as it can and overproduces.
  • Sizing kanban loops by guess. Use measured replenishment time and real demand, and recalculate when either changes.
  • Running pull with a lumpy schedule. Without heijunka, the upstream process sees a surge of one product and starves on the next.
  • Leaving the rules unwritten. If parts are made without a card, the cap is gone. Put the rules where the cards are and hold to them.
  • Doing JIT without jidoka. Low inventory and tolerated defects together mean a line that stops all day.

AI Value Stream Mapping

Map your flow for JIT in LeanSuite

AI value stream mapping from video: LeanSuite turns phone footage of each process step into a current-state value stream map with measured cycle times, counted inventory, the lead-time ladder and process cycle efficiency

See AI Value Stream MappingBest value stream mapping software

Go deeper

This page is the short version. These guides cover each part in full.

All our articles on this subject are in the value stream mapping and flow topic.

Free templates and tools

Key terms

Lean glossary

Doing this in LeanSuite

See it on your own lines

Bring the problem you are working on to a 45-minute demo. Or look around with sample data first, or pilot LeanSuite on your own lines for 90 days, with the setup done by our team.

Book a 45-minute demoClick through the interactive demoStart a 90-day pilot

FAQ

Questions about just-in-time manufacturing

  1. Just-in-time manufacturing means making and delivering only what the next process needs, when it needs it, in the quantity it needs. Production is pulled by actual use instead of pushed by a forecast, so inventory stays small and problems are seen quickly instead of being hidden by stock.

  2. Toyota, where the system was built, is the standard example: each process makes parts only when the next one signals for them with a kanban. On a smaller scale, a cell that holds half a day of cut blanks in a sized supermarket, and whose supplying machine only cuts more when an empty container comes back, is running just-in-time.

  3. Lean manufacturing is the whole approach: define value, remove waste, create flow, pull, and improve continuously. Just-in-time is the part of lean that deals with flow and pull on the shop floor. JIT came first, at Toyota; lean is the later name researchers gave to the full system, which also includes problem solving, standard work and leadership routines.

  4. Just-in-time is the goal: each process gets what it needs when it needs it. Kanban is one of the tools that achieves it: a card or container that authorises the upstream process to make or move a set quantity, and whose count caps the inventory in the loop. You can have kanban without the rest of JIT, but you cannot run JIT between separated processes without some pull signal.

  5. Takt time to set the pace from demand, one-piece flow and cells so parts do not wait in batches, kanban and supermarkets to pull between steps that cannot be joined, heijunka to level volume and mix, quick changeovers (SMED) so small lots are practical, standard work to hold the pace, and jidoka so problems stop the process instead of being passed on.

  6. Lower inventory, so less cash, space and handling tied up in stock; shorter lead times, so faster promises and quicker changes of mix; fewer obsolete parts; and faster detection of defects and breakdowns, because there is no stock to hide them. The last one is the point: problems get fixed instead of worked around.

  7. With little stock, a late supplier, a breakdown or a bad lot stops the line quickly. JIT needs stable quality, reliable equipment, short changeovers and dependable suppliers, and it is fragile when inventory is cut before those are in place or when it depends on a single distant source with no agreed buffer. The fix is a sized buffer plus work on its cause, not pallets everywhere.

  8. No. Just-in-time means every piece of inventory has a reason and a limit. Supermarkets between processes, safety stock sized from measured lead time and demand variation, and the standard work in process inside a cell are all inventory, and all deliberate. What JIT removes is the stock that exists only because a batch was made before it was needed.

  9. It removes waiting rather than speeding up work. Parts move straight to the next step or wait briefly in a capped supermarket, so lead time falls to something close to the work content, and the space, handling and counting that piles of stock need disappear. Because defects are found within minutes, less work is repeated or scrapped.

  10. They are two sides of one idea. Just-in-time production is the shop-floor system: takt, flow, pull and levelling. Just-in-time inventory or purchasing is the supply side: ordering so material arrives shortly before it is used. The supply side depends on the production side, because a supplier can only deliver to a schedule the plant actually keeps.

Related guides and tools

More method guides

Or browse the topics, the full library and the lean glossary.

Rather see it on your own problem?

Bring the problem you are stuck on, and we will show you on a call how LeanSuite handles it.

Schedule Demo

Pass it on

We want LeanSuite to be the best place on the internet for lean help. If this was useful, send it to someone on your team or in your network who needs it.