Work in Progress (WIP): Why Starting More Work Makes Everything Later

Picture a supermarket where every cashier, to look busy, starts scanning the next customer's basket before finishing the current one. Each till is now "working on" three customers at once. Nobody leaves the store any faster. Everybody waits longer. Factories do exactly this every day, and call it keeping the machines busy.

Work in progress (WIP) is everything that has been released to production but isn't finished yet: material issued, orders started, parts sitting between operations. It's money you have already spent and can't yet sell. And there's a simple law that links it directly to how long your customers wait.

Little's law: the formula behind WIP

In 1961, John D. C. Little, an operations researcher who later spent most of his career at MIT, published a proof of a relationship that queueing experts had been using for years. For any stable system, over the long run:

WIP = throughput × lead time

Or, rearranged the way planners use it most:

Lead time = WIP ÷ throughput

It works for supermarket queues, airport security and production lines. It doesn't care about product type, machine brand or ERP. The only conditions are that you measure long-run averages and use the same units on both sides.

A worked example

Your line finishes 10 units a day. That's its throughput, set by the slowest step. On average, 40 units are somewhere between release and finish.

Lead time = 40 ÷ 10 = 4 days.

Now sales pushes for more orders to be started "to get a head start". WIP rises to 60 units. The bottleneck still finishes 10 a day, because releasing more work doesn't make the slowest machine faster.

Lead time = 60 ÷ 10 = 6 days.

You started more and everything got two days later. Nothing is produced sooner. The extra 20 units just queue.

Go the other way and cap WIP at 30 units. As long as the bottleneck never runs out of work, throughput stays at 10 a day.

Lead time = 30 ÷ 10 = 3 days.

Same machines, same people, same output, one day faster than where you started. That's the whole argument for controlling manufacturing lead time through WIP rather than through overtime.

What WIP really costs

Say each unit in progress carries $250 of material and labor.

• At 40 units: 40 × $250 = $10,000 tied up on the floor.
• At 60 units: 60 × $250 = $15,000.
• At 30 units: 30 × $250 = $7,500.

With a typical carrying cost of 20% a year, the difference between 60 and 30 units is $7,500 × 20% = $1,500 a year for one line. That's the visible part. The hidden costs are bigger: late engineering changes that hit 60 half-built units instead of 30, parts damaged while waiting, and planners hunting for "where is order 118?" The same logic applies to inventory management in general. WIP is just inventory that happens to be in the way.

Myth: "An idle machine is lost output"

Only at the bottleneck. A machine upstream of the bottleneck that runs flat out doesn't add output. It adds WIP in front of the bottleneck, and by Little's law, lead time. The goal is a bottleneck that never starves, not every machine running all the time. Manufacturing optimization starts at the bottleneck explains how to find yours.

Release control and CONWIP, simply

If WIP drives lead time, the lever is the moment you release work. That's where CONWIP comes in (short for constant work in process).

Think of the traffic lights on a highway on-ramp. They let one car in every few seconds. It feels slower at the ramp, but the highway keeps flowing, and everyone gets home sooner than if all the cars squeezed in at once.

CONWIP does the same on the shop floor:

1. Set a cap. For example, 30 units for the line, based on Little's law and the lead time you want.
2. Release one, finish one. A new order goes to the floor only when a finished one leaves.
3. Keep the queue in the office, not on the floor. Orders waiting for release are just paperwork. They can still be re-sequenced, changed or canceled at no cost.

Start the cap a bit high and lower it step by step. If the bottleneck starts running out of work, you went too far.

How to count your WIP

You can't cap what you don't count. A quick way:

• List released orders without a final confirmation. Those are your WIP orders.
• Convert to one unit. Pieces, meters or hours at the bottleneck, whichever fits your product.
• Measure throughput the same way. Average completed units per day over the last four to six weeks.
• Divide. Compare the result to your actual lead time. If they're close, the law is working and you now have a lever.

How this looks in factory.online

In factory.online, WIP is visible as released job orders that don't yet have a final output confirmation, with the batches and quantities they carry. The Gantt shows where those orders queue in front of each resource. A planner who sees 60 units waiting ahead of a line that finishes 10 a day can hold back new releases and watch the queue shrink, instead of adding more.

If your lead times keep growing while output stays flat, it's worth counting your WIP together. Book a demo and bring last month's numbers.