If you followed the purest interpretation of Lean, you would have no warehouse, stock or raw materials. In reality you probably will, as we live in an imperfect world and sometimes you need to be pragmatic – but this does not mean opportunities to improve are written off to pragmatism.
Six Sigma has been applied to manufacturing for years of course. Whether your company uses Six Sigma or not, you could borrow a simple principle, to release a significant amount of cash in finished goods, WIP and raw materials. This could in turn reduce the stock footprint and the risk of obsolescence.
Sound too good to be true?
This principle applies to ‘make to stock’ arrangements. ‘Make to order’ works differently but has similar opportunities. That is for another day.
Most operational planning will schedule manufacturing based on the number of weeks of stock or similar. Let’s illustrate with an example. These figures are not realistic numbers, they are here for illustrating a principle only.
A factory produces three products imaginatively named A, B and C. The stock looks like this:
Reviewing this, a planner would possibly decide to make product C as it has the lowest weeks of stock available and the stock level is below average. On the surface of it, nothing wrong with that.
Product C has annual sales of £100,000. What if those sales comprise entirely of orders from one customer who takes £1,923 each week? It never varies. In that case you would only need to make precisely that amount and ship it. There would be no need for more than a week’s worth of stock.
Similarly, what if product B varies by 50% week to week. You would want to carry more stock to protect customer service levels.
Taking other factors to one side for a moment, we can base a stock management on variability rather than average volume.
We should first consider what we mean by variability. Variation to what? You could base the calculation on average sales volume, but better still, to forecast. This would allow the additional measure of MAPE to the forecast accuracy, which is related to the variability we need to measure. MAPE (Mean Absolute Percentage Error) is a very good way to drive improvement in forecast accuracy, which has a direct effect on stock (as we will discover) and various other factors which impact manufacturing costs.
If you measure the variability of a good sample size of orders, versus the corresponding forecast value, you can derive a value of standard deviation (Sigma). The higher the number, the greater the variation. We now have a quantifiable understanding of the variation of product orders against forecast.
Next, we must explain that the maths we are using is based on a normal distribution to describe the variation. Mathematically the tail of a normal distribution never reaches zero. This means that to accommodate all possible variation, stock levels would be infinite. Clearly this is not sensible.
But we can use the maths to target a service level and return a target stock level. For example, a failure rate of 0.5% means you must accommodate 2.6 standard deviations of variability.
Say product A had a variation described by a standard deviation of £18.30 per week in order value, you would have to hold the average order quantity plus 2.6 time the standard deviation to attain a service level of 0.5%. This would be £240 of stock as opposed to the current £600.
Doing this across all three products would reduce the stock from £9,100 to £3,619 – 60% less stock and 60% of ‘cash in stock’ is released.
You are not forced to treat all products the same. You can set a tolerable failure rate by product if you wanted to. Extrapolate this further, you could set it by customer and product, to return a total stock holding. Suddenly you have your hands on the controls, balancing off service levels and ‘cash in stock’.
There are other factors to consider of course, such as MOQs, economic purchase and manufacture quantities, storage costs, changeover times, etc. All these factors (and others) are quantifiable and could be integrated into a planning formula.
The principle can be applied across finished goods as well as raw materials. Simply push the same steps through the BOM to raw materials from finished goods.
These are not real numbers of course, but the principle stands, and the savings are there.
Following this principle can create new improvement drivers. We touched on MAPE. There is now a direct correlation between forecast accuracy, cash in stock and customer service level driving improvement. Not only that, you can quickly isolate the problem products and accounts.
Many Lean programmes will target the speed of changeovers in manufacturing. Many will experience reducing 60 minutes plus changeovers to under 10 for example, but this stock management principle can eliminate them. Conversely, if the changeovers are fast enough, stock can be further reduced having more changeovers and reduced run lengths. You just need to balance off which is the most economic and agile route. This can be done mathematically; we can cover this another time.
There is the principle. Of course, deploying this will have some challenges. You could use a spreadsheet to demonstrate the principle or to act as a proof of concept. However, to deploy something like this sustainably, a database system would be required. Vensis has helped clients manage stock more efficiently by building database systems that link to other existing systems, such as ERP/MRP, performance measures, S&OP, etc. If a computer can be doing it, it should be doing it. After all, this is just maths and following a set of pre-determined rules. Computers are pretty good at that; allowing staff to focus on the next area for improvement.