The Two-Bin (Kanban) Inventory System: A Practical Setup Guide
A two-bin (kanban) inventory system is one of the simplest reliable ways to keep parts on the shelf without babysitting a spreadsheet. You store each item in two containers. You pull from the first until it runs dry, that empty bin becomes your signal to reorder, and you pull from the second while the new stock arrives. No forecasting meeting, no daily count. The shelf tells you when to buy.
It sounds almost too basic, and that is the point. Toyota built a chunk of its production system on this idea, and the same logic scales down to a hardware distributor, a dental clinic, or a maker workshop. Here is how it actually works, how to size the bins with a real formula, and where it beats the min/max method most people default to.
How a two-bin (kanban) inventory system works
Every stocked item gets two bins, each holding a known quantity. The flow is a loop:
- You issue stock from Bin 1 only.
- Bin 1 empties. That empty bin (or the kanban card taped inside it) is the trigger. You place a replenishment order for exactly one bin's worth.
- While you wait, you issue from Bin 2. This bin has to cover you until the order lands.
- Stock arrives, you refill the empty bin, and it goes to the back of the line. Bin 2 is now the active bin, and the cycle repeats.
The "kanban" part is the visual signal. In a physical setup it is a literal card or the empty container itself. The rule the whole system rests on is this: the second bin must hold enough to cover demand for the entire lead time, plus a cushion. Size that wrong and you either stock out or drown in inventory.
The formula for sizing each bin
You do not guess bin sizes. You calculate them from two numbers you already have: how fast the item moves, and how long the supplier takes.
Bin quantity = (Average daily demand x Lead time in days) + Safety stock
Safety stock is your buffer against a slow supplier or a demand spike. A clean rule of thumb is one to two days of average demand, or if you want to be precise, use the statistical version: Z x standard deviation of demand over the lead time, where Z is 1.65 for a 95 percent service level.
Worked example
Say you stock M6 bolts. Pull the last 90 days of usage and the supplier's typical turnaround:
| Input | Value |
|---|---|
| Average daily demand | 40 bolts |
| Supplier lead time | 5 days |
| Safety stock (1.5 days) | 60 bolts |
| Bin quantity | (40 x 5) + 60 = 260, round to 300 |
So each bin holds 300 bolts. You keep two, meaning 600 on hand when both are full. When Bin 1 empties, you have consumed 300 and Bin 2's 300 must last the 5-day lead time. At 40 per day you burn 200 over those five days, leaving 100 as a real buffer if the truck is late or demand jumps. That headroom is the safety stock doing its job.
Notice the reorder point falls out of this for free: it is simply one bin's worth, 300 units. The empty bin is the reorder point made physical, which is why you never have to calculate it again once the bins are sized.
Two-bin vs min/max vs a plain reorder point
These methods solve the same problem with different amounts of overhead. Pick based on how much counting you want to do.
| Method | Reorder trigger | Best for | Weak spot |
|---|---|---|---|
| Two-bin (kanban) | Bin goes empty (visual) | Steady-demand, low-cost parts | Needs shelf space for two bins |
| Min/max | Count drops below min | Variable demand, tracked in software | Requires accurate perpetual counts |
| Reorder point | On-hand hits a number | High-value items worth watching | You have to keep counting to know |
Two-bin wins when demand is reasonably steady and the item is cheap enough that holding a full second bin is not painful. Min/max and reorder-point logic are stronger for expensive or lumpy items where you would rather not park a whole bin of capital on a shelf. Plenty of operations run all three at once, sorting items by an ABC analysis and putting the C-class fasteners, labels, and consumables on two-bin.
Setting it up in seven steps
- Pick the items. Start with steady movers that are cheap and annoying to stock out on. Consumables and fasteners are the classic first candidates.
- Get the numbers. Average daily demand and lead time per item. Ninety days of history is plenty.
- Size the bins with the formula above. Round up to a sensible pack or container size.
- Label everything. Each bin needs the SKU, the bin quantity, the supplier, and a reorder reference. This is your kanban card.
- Rack them in sequence so you always pull from the front bin and the empty one is obvious.
- Define the signal path. Who sees the empty bin, and how does that become a purchase order? A physical card board, a scan, or an API call all work.
- Review quarterly. Demand and lead times drift. Re-run the formula and resize.
Where two-bin breaks down
It is not universal. Skip or adapt it when:
- Demand is highly seasonal or spiky. Fixed bin sizes assume steady pull. A holiday spike empties both bins before your order lands.
- Items are perishable or serialized. FIFO and expiry tracking need date logic a plain bin cannot enforce.
- The item is expensive. Holding a full second bin of a high-value part ties up cash you could deploy elsewhere.
- Lead times swing wildly. If your supplier ranges from 3 to 20 days, the safety stock needed to stay covered makes the bins huge.
Running it digitally
The card-on-a-board version works great until you have hundreds of SKUs across two warehouses and nobody can see the empty bins from the purchasing desk. That is where a digital layer pays off: scan an empty bin, and the system raises the replenishment automatically, logs it, and shows on-hand across every location.
This is exactly the kind of workflow Inventoros handles out of the box. It is a free, open-source, self-hosted inventory system with multi-location stock, reorder rules, webhooks, and a full REST and GraphQL API, so you can wire an empty-bin scan straight to a purchase order without paying per seat or shipping your data to someone else's cloud. Set your bin quantities as reorder points, and the signal that used to be a card on a wall becomes an event your systems can act on. The documentation covers reorder configuration and the webhook payloads if you want to build the automation yourself.
FAQ
What is the difference between a two-bin system and kanban?
Kanban is the broader idea: a signal that authorizes replenishment when stock is consumed. Two-bin is one specific, physical way to run kanban, where the empty container is the signal. Every two-bin setup is kanban, but kanban also covers card systems, electronic signals, and multi-bin variants.
How many bins do you actually need, two or three?
Two is the standard. Some operations add a third bin (a "three-bin" system) when lead times are long or unreliable, giving an extra buffer while an order is in transit and reducing stockout risk. The tradeoff is more shelf space and more capital tied up, so only add the third bin when the numbers justify it.
How do you calculate the right bin size?
Multiply average daily demand by the lead time in days, then add safety stock. For the bolts example, 40 per day over a 5-day lead time plus a 60-unit buffer gives a 300-unit bin. Round up to a practical pack size and recheck the inputs each quarter.
Does two-bin work for perishable or high-value inventory?
Not well without changes. Perishable stock needs FIFO and expiry rules a plain bin does not enforce, and high-value items make holding a full second bin expensive. For those, a reorder-point or min/max method tracked in software usually fits better. Reserve two-bin for steady, low-cost, non-dated items.