Your inventory isn't a safety net. It's evidence of a broken process.
Every pallet of WIP sitting between your processes, every buffer stock built "just in case" the upstream station is late, every finished goods inventory protecting you from your own unpredictable lead times — all of it is the physical manifestation of a system that doesn't flow.
Just-in-Time doesn't ask you to run with zero inventory and hope for the best. It asks a more fundamental question: why does the process need the buffer in the first place?
What JIT Actually Is — and What It Isn't
Just-in-Time is the manufacturing principle of producing the right part, in the right quantity, at the right time — and no earlier. It is one of the two pillars of the Toyota Production System alongside jidoka (built-in quality), and it is fundamentally a philosophy about flow, not a technique for reducing stock levels.
The inventory reduction that comes from JIT is a consequence of fixing the process, not a target to chase directly. Organisations that try to reduce inventory without addressing the process instability that caused it end up with stockouts, expediting, and frustrated customers. Organisations that fix the flow first find that inventory falls naturally as a result.
The opposite of JIT is push manufacturing — producing to a schedule or forecast regardless of what downstream processes actually need. Push systems create WIP buffers between every workstation because each is operating to its own schedule, not responding to actual downstream demand. The result is inventory that looks like protection but functions as a hiding place for problems: quality issues, downtime, changeover delays, and process imbalances that never get surfaced because the buffer absorbs them before anyone notices.
The Three Foundations of a JIT Pull System
JIT is implemented through three interlocking mechanisms. Understanding them separately matters — they're often conflated in a way that leads to partial implementations that don't deliver results.
Takt time establishes the pace. Takt time is the rate at which customers are buying: available production time divided by customer demand. If your facility has 450 minutes of available production time per shift and customers require 90 units per shift, your takt time is 5 minutes per unit. Every process in the value stream should be capable of producing one unit every 5 minutes. Processes faster than takt are overproducing. Processes slower than takt are the constraint.
Takt time gives you the standard. Without it, "flow" is just a word. With it, you have a precise target for every station's cycle time, a quantified definition of the constraint, and a rational basis for staffing and capacity decisions.
One-piece flow eliminates batch-and-queue. Batch production — running large lots through each process before moving them on — is the engine of inventory accumulation. A batch of 200 units entering a three-step process spends most of its time waiting: waiting to be processed, waiting for the full batch to complete, waiting in the queue at the next station. Shifting to one-piece flow, or the smallest practical lot size, keeps material moving continuously rather than accumulating in queues between stations.
Pull replaces push. In a pull system, no process produces until the downstream process signals that it needs material. The signal mechanism is the kanban — a physical card, bin, or electronic trigger that authorises production of a specific quantity of a specific part. Kanban doesn't eliminate inventory; it controls it precisely. A defined storage location called a supermarket holds only the quantity authorised by the kanban calculation, and it is replenished only when the downstream process withdraws from it. Nothing is produced speculatively.
[Important sidenote: if your team is struggling to see where your value stream is push-driven and where inventory is accumulating between processes, a current-state VSM will show you the full picture in a day. Download our free trial here and request a complimentary web meeting with one of our Lean experts. We've helped manufacturing teams across industries design JIT pull systems grounded in their actual demand and process data.]
How JIT Appears on a Value Stream Map
JIT and VSM are inseparable — the value stream map is both the diagnostic tool that reveals where JIT is missing and the design tool for building the future-state pull system.
On a current-state VSM, the absence of JIT is unmistakable. Inventory triangles sit between process steps, each labelled with the quantity of WIP in days or units. In a push system, those triangles typically account for the vast majority of total lead time — actual processing may take hours while the inventory between steps represents days of waiting. The push arrows (striped arrows in standard VSM notation) confirm that material is being scheduled forward rather than pulled by downstream demand.
The future-state VSM shows what JIT looks like when designed correctly. Push arrows are replaced by pull loops — withdrawal kanbans authorising the downstream process to take from the supermarket, and production kanbans authorising the upstream process to replenish what was withdrawn. Supermarket icons appear between processes where one-piece flow isn't practical: shift boundaries, supplier delivery cycles, or stations with significant changeover time. The inventory triangles shrink to the calculated kanban quantity. Total lead time on the timeline falls accordingly.
The most powerful moment in a JIT implementation is often when the team draws that current-state timeline and realises that 85–90% of their lead time isn't processing time — it's inventory sitting between stations waiting its turn. That gap is the improvement opportunity, and the map makes it impossible to ignore.
A Real Example: From Push to Pull in an Automotive Components Facility
A Tier 2 automotive components supplier was running a classic push operation. Production scheduling drove every workstation independently to a weekly plan, with no signal mechanism between processes. The result was predictable: large WIP buffers at every handoff point, chronic floor space constraints, and a lead time of 31 days on a product family whose actual processing time totalled less than 3 days.
On-time delivery had fallen to 71% despite — and partly because of — the excess inventory. The buffers were obscuring where the real capacity and quality problems sat. Nothing was urgent because nothing was visibly starved.
A cross-functional team mapped the current-state value stream and found four major inventory accumulation points: between press forming and sub-assembly (9 days), between sub-assembly and final assembly (8 days), between final assembly and surface treatment (6 days), and in finished goods awaiting dispatch (5 days). Processing time across all four operations totalled 2.8 days. Twenty-eight of the 31 days of lead time were inventory wait.
The future-state design introduced pull loops at each accumulation point, with supermarkets sized to cover the replenishment lead time plus a calculated safety quantity. Kanban quantities were set from actual daily demand rates and measured process cycle times. Batch sizes were reduced at two stations where long changeover times had historically driven large-lot production — freeing up the flexibility the pull system required to respond to daily demand variation.
Results after eight months:
- Lead time: 31 days → 11 days
- WIP inventory value: $1.8M → $490K
- On-time delivery: 71% → 94%
- Floor space recovered: 2,100 sq ft
The inventory reduction was real — but the more significant outcome was what the lower inventory exposed. Two quality escape points and one capacity constraint that the buffers had been concealing for years became immediately visible once the cushion was gone. Resolving those issues drove the on-time delivery improvement from 71% to 94%.
Getting Started: Four Steps to Building a Pull System
1. Map the current state before touching the process. Walk the value stream, draw the map, and calculate total lead time. Then separate processing time from inventory wait time at each step. That ratio — often ten to one or worse — is your baseline and your business case for the pull system redesign.
2. Set the takt time for your primary product family. Divide available production time by customer demand. Compare every station's cycle time against takt. Processes above takt are the constraint. Processes significantly below takt have idle capacity — and in a push system, that idle capacity often drives overproduction as operators run to schedule rather than to demand.
3. Design the future-state pull system on paper. Identify where one-piece flow is feasible — connected processes with similar cycle times and minimal changeover. Where it isn't practical, place supermarkets and design the kanban loops. Calculate kanban quantities from replenishment lead time, daily demand, and demand variation — and factor in pitch (takt time × pack-out quantity) to align kanban sizing with how material actually moves in your facility. Gut-feel safety stock numbers are not a substitute for this calculation.
4. Implement one pull loop at a time, starting at the pacemaker. The pacemaker process is where you introduce the customer demand signal — typically the last process before finished goods or shipping. Build and stabilise the pull loop between the pacemaker and its immediate upstream supplier first. Then extend the signal one step further upstream. Attempting to implement the full future state simultaneously is one of the most common — and expensive — JIT implementation failures.
The goal of JIT isn't empty shelves. It's a production system responsive enough that you don't need the shelves in the first place.
Ready to map your current-state value stream and see exactly how much of your lead time is inventory wait rather than processing? Download a free 30-day trial of eVSM and build your first pull system design directly on the map. Or book a complimentary meeting with one of our Lean experts — we can help you calculate kanban quantities and design a pull system grounded in your actual demand and process data.