Lesson 7: Make-to-stock under DBR

When you must make to stock

Make-to-order should be the natural way to manufacture. Variety keeps growing, models change faster, and everyone wants less inventory. Yet sometimes producing to stock is unavoidable:

when the production lead time is longer than the customer is willing to wait;
when a huge peak in sales is expected and capacity can’t follow it.

Make-to-stock by forecast has its own damage. Nobody knows when the work in process will be sold, so precious constraint capacity goes to products that aren’t needed now while others are truly urgent. The result is the familiar pair: too much stock of some products and shortages of others.

Two habits make it worse. Batching: large stocks and long forecasts make large batches look harmless, and a min-max rule imposes a minimum batch, which adds units nobody needs, lengthens the lead time and lengthens the cycle. Forecasting: a forecast seems to answer “how many will we sell next period?”, but it is never accurate, and what matters is not the average but how many we might sell. The longer the horizon, the worse it gets, and frequent re-forecasting adds so much noise that nobody can tell whether the forecast was valid.

The direction of the solution

Hold low finished stock at the plant warehouse, so practically all of it is sold soon.
Replenish fast: every day, make what was consumed. No batching unless absolutely necessary.
Use buffer management for priorities.

Each day, for every product: production order = replenishment level − stock − work in process. One order per product, released by priority and by the capacity of the constraint, which is loaded up to its buffer time and no further.

The replenishment level

The main planning decision is the replenishment level (the norm). It is the make-to-stock equivalent of the shipping buffer: it protects against peaks of demand and against delays, and it sets the total stock in the system. Five things decide its size:

the rate of consumption;
the variability of consumption;
the average replenishment time;
the variability of the replenishment time;
the required service level (critical, high, or not truly critical).

The generic rule:

Replenishment level = “maximum” expected consumption within the average replenishment time, factored up for the unreliability of the replenishment time.

A plant that implemented this for its standard products wrote it as maximum sales × production lead time × a reliability factor, with the production lead time taken as half the current one, since choking the release halves it.

Buffer status for stock

Buffer status = (replenishment level − on hand) ÷ replenishment level

The emergency level is where the stock is so low that the product must be expedited. It must leave enough time to expedite an order, and the replenishment level must be high enough that the stock doesn’t cross it too often. The default is a red zone of a third of the level.

Buffer management also tells you whether the level is right. A product that penetrates the red zone again and again has a level that is too low: demand is growing, replenishment takes longer, or fluctuations are bigger than assumed. A product that never leaves green for a long time has a level that is too high. Record every penetration into red, and review them periodically.

Stock and time: when to use which

TOC thinks in time buffers, and stock can be translated into time: finished stock as days of sales, stock at the constraint as hours of its work. But the translation rests on an average, and an average of past sales is just a forecast.

Suppose 400 units sit in stock, four weeks of sales at 100 a week. Then comes a week with no sales. The average over the last four weeks is now 75, and the same 400 units are suddenly 5.3 weeks. Nothing was produced, yet the report says stock went up.

So use each where it fits:

Planning (how much stock should we hold?) is a time question: how much can we sell before replenishment arrives, plus possible delays.
Execution (what is urgent today?) uses units: the share of the replenishment level that is missing.

Pause and think: during a seasonal peak, the plant finds its replenishment batches are larger than usual. Is something wrong?

Yes: batches should always be one day of sales, and anything larger is a planning error
No: more is consumed between releases at the peak, so orders are bigger
Yes: it means the forecast was too high, and the levels should be cut immediately

At peak times the constraint approaches being a bottleneck and the stock is expected to go down. Off-peak, batches are small, often a day or two of sales, and the constraint’s load must be watched so the extra set-ups don’t eat its spare capacity.

Firm future orders on stocked products

A stocked product sometimes also has firm orders for a future date. Ignoring them is unreasonable; releasing them the day they arrive, perhaps months early, is too. And buffer management can be thrown off either way.

An example. Item X has a replenishment level of 100 and a demand horizon of three weeks (its average lead time). An order for 75 units, to ship in three weeks, arrives today. On hand: 85. An order for 12 was released three days ago. So a new work order of 100 − 85 − 12 + 75 = 78 is released.

If buffer management subtracts the 75 at once, effective stock falls to 10. The older order jumps from 15% to 90%, and the new one starts at 78%: urgent priority for demand three weeks away.
If it ignores the 75 until the ship date, then two and a half weeks later, with 54 units on hand, the order carrying the 75 shows only 46%: no rush, three days before the shipment.

The proposition: release the work order when the firm demand comes within the demand horizon; let buffer management subtract it from stock only when it is within a third of the horizon, the equivalent of crossing into red.

What to stock, and how to switch

Stock a product or component only when there is a clear gain: more sales or a better price because the shorter lead time really matters to customers; an investment that isn’t too large; demand variability that isn’t so high that the stock may never sell. Sometimes the gain is internal: stocking components common to many end items, typical in V and T plants, frees a long-set-up resource that would otherwise be a bottleneck.

Switching from make-to-order to make-to-stock needs care. The stock has to be built on top of regular demand, which needs excess capacity at the time of the switch. Don’t change the rule before the stock is in place.

Check your understanding

Answer all five. Get four right to complete the lesson.

1. Make-to-stock is unavoidable when…

The forecast is accurate enough to plan production a month ahead
The constraint has long set-ups
The production lead time is longer than the customer will wait
Sales asks for it

Or when a peak is expected that capacity can’t follow.

2. Each day, the production order for a stocked product is…

Replenishment level − stock − work in process
The forecast for next month, divided by the working days
The economic batch quantity
What the min-max rule says

Make what was consumed, daily.

3. A product keeps penetrating the red zone. This means…

The plant should expedite harder
The forecast was wrong last month
The product should move to make-to-order
The replenishment level is probably too low

Demand may be growing, replenishment slower, or fluctuations bigger.

4. 400 units were four weeks of stock; after a week with no sales they are “5.3 weeks”. The lesson is…

Stock should always be reported in weeks
Plan in time, but set daily priorities in units
The average should use more weeks
The product is obsolete

Converting stock to time daily creates false signals.

5. A firm future order on a stocked item is deducted from stock in buffer management…

As soon as the order is booked
Only on its ship date
When it comes within a third of the demand horizon
When the work order for it is released to the floor by planning

Release within the horizon; make it urgent only within its last third.