Safety Stock Isn't Just a Number: How MRP and Demand Volatility Should Work Together
Updated: 6d
Safety stock is often treated as a number in an ERP system.
A minimum quantity.
A parameter.
Something that gets reviewed occasionally when inventory is either too high or a material runs out.
But safety stock has a much more important role than that.
Used properly, it should provide a deliberate buffer against the uncertainty between ordering something and receiving it.
And that means safety stock shouldn't really be considered in isolation.
It needs to work as part of the wider MRP and inventory planning process.
What is safety stock actually there for?
The simplest way to think about safety stock is as a buffer.
Demand isn't always predictable.
Suppliers don't always deliver exactly when expected.
And the business doesn't have perfect information about either.
There is therefore a period where the business is exposed to uncertainty.
The original methodology I developed for a customer defined this as the period between placing an order and receiving the goods. Safety stock provides the minimum quantity intended to remain available during that period.
That distinction is important.
Safety stock isn't there to satisfy normal demand.
Normal demand should be covered by the replenishment cycle.
Safety stock is there to provide a buffer against the variation around that expected demand and supply.
Where does MRP fit?
MRP should bring several pieces of information together.
In the methodology I developed, MRP works from four key datasets:
Current Inventory
Inventory Policy
Demand
Supply Parameters
It then looks forward across the demand horizon and uses information such as lead times and minimum order quantities to generate replenishment recommendations.
That is important because MRP isn't simply answering:
“What do I need to buy?”
It is effectively asking:
“Given what I have, what I expect to need, what I have already ordered and how long supply takes, when do I need to replenish?”
That is a much more useful question.
Safety stock creates the trigger
Imagine inventory gradually reducing as demand consumes it.
At some point, projected inventory reaches the safety-stock level.
MRP then needs to recommend replenishment sufficiently early to allow for the supplier lead time.
The original model illustrates this relationship between:
projected inventory
safety stock
lead time
replenishment quantity
cycle stock
reorder point
the period of uncertainty.
The principle is relatively simple:
Don't wait until you have run out before reacting.
But that only works if the safety-stock level is meaningful.
And that's where things become more interesting.
A common approach is to calculate safety stock using three factors:
Demand volatility
How variable has demand been?
The methodology uses standard deviation of demand over the chosen historical period as the measure of volatility.
Lead time
How long is the period of uncertainty?
The longer the lead time, the greater the exposure to demand variation before the next replenishment arrives.
Service factor
How much of the demand variation do you want the safety stock to cover?
The greater the service factor, the greater the amount of variation being protected against.
The calculation used was:
Safety Stock = STDEV(Demand) × √Lead Time × Service Factor
This produces a static safety-stock quantity that can then be used as an MRP replenishment parameter.
But there is a problem with static safety stock
A static safety-stock figure assumes that the future demand profile is reasonably represented by the historical calculation.
But what happens when demand changes significantly?
Imagine an item normally has relatively low demand.
Its calculated safety stock might therefore be relatively low.
Then you know that demand is expected to increase significantly over the next few months.
Should the business continue to hold exactly the same safety-stock quantity?
Not necessarily.
The original methodology therefore also considered dynamic safety stock, where the required safety-stock level changes according to the forward demand profile.
The example in the original document shows how a period of higher demand can require a higher safety-stock level, while a lower-demand period can require less.
That is a useful distinction:
Historical demand can tell you how variable demand has been. Forward demand can tell you when that variability might matter most.
Where does XYZ analysis come in?
This is where the methodology becomes particularly useful.
Not every SKU behaves in the same way.
Some items have relatively stable demand.
Others have highly variable demand.
And treating them all in the same way can create problems.
XYZ analysis provides a way of classifying items according to demand volatility.
In simple terms:
X items – relatively stable demand
Y items – some variation in demand
Z items – highly variable or less predictable demand
The exact classification thresholds need to be appropriate to the business and the data being analysed.
The important point isn't the label.
It's the information the classification provides.
Why does demand volatility matter?
Consider two materials with the same average monthly demand.
One is consumed relatively consistently.
The other has significant peaks and troughs.
Their average demand may look similar.
Their risk profile doesn't.
If the business applies the same approach to both, it may either:
hold too much stock of the stable item
or
hold too little stock of the volatile item.
XYZ analysis helps highlight that difference.
It therefore provides another useful input when reviewing safety-stock policies.
Safety stock should support the items that matter
This becomes particularly important when looking at critical materials.
A business doesn't necessarily need to treat every SKU in exactly the same way.
For example, an item might be:
critical to production
difficult or slow to source
highly variable in demand
subject to a long supplier lead time
That combination presents a very different supply risk from an item that is readily available, stable in demand and easy to replenish.
XYZ analysis can therefore help identify where demand volatility should influence the review of safety-stock levels, particularly when combined with the item's operational importance.
The objective isn't simply to increase safety stock.
It's to put the right level of protection in the right place.
More safety stock isn't necessarily the answer
This is where inventory management can easily go wrong.
If a material keeps running out, the obvious response can be:
“Increase the safety stock.”
But why is it running out?
Perhaps:
demand volatility has changed
lead times are inaccurate
supplier performance has deteriorated
MRP parameters are wrong
minimum order quantities are inappropriate
the forecast isn't reflecting the demand profile
inventory transactions aren't accurate
the material isn't actually being consumed as expected
Increasing safety stock might hide the problem rather than solve it.
And it creates another consequence:
more working capital tied up in inventory.
The MRP parameter is only as good as the information behind it
This is perhaps the most important point.
MRP can be very effective at generating replenishment recommendations.
But it relies on the information it is given.
If lead times are wrong, the recommendation will be wrong.
If inventory records are wrong, the recommendation will be wrong.
If demand is poorly represented, the recommendation will be wrong.
If safety stock is simply an arbitrary number, the protection it provides may be arbitrary too.
So the question shouldn't simply be:
“Is MRP working?”
It should be:
“Are we giving MRP the right information to make sensible decisions?”
A better way to think about inventory protection
Rather than looking at safety stock as an isolated number, consider the relationship:
Demand
↓
Demand volatility
↓
Lead time & supply risk
↓
Inventory policy
↓
Safety Stock
↓
MRP replenishment
↓
Purchasing / Supply
↓
Inventory availability
Each part influences the next.
Change one parameter and you may create a consequence somewhere else.
Which brings us back to a principle I often use when looking at supply chains:
Every decision in the supply chain creates a consequence — often somewhere else.
The objective isn't maximum inventory
The objective is appropriate inventory.
Enough protection to manage genuine uncertainty.
Enough forward visibility to anticipate changes.
Enough planning discipline to allow MRP to do what it is supposed to do.
And enough understanding of demand behaviour to recognise that not every SKU carries the same level of risk.
That's where safety stock becomes more than an ERP parameter.
It becomes part of a deliberate supply strategy.
A final thought
I've seen inventory discussions become focused on one question:
“How much stock should we hold?”
I think a better starting point is:
“What are we trying to protect ourselves against?”
Once that is understood, we can start considering demand volatility, lead time, service requirements, supply risk and the role MRP should play.
And that can lead to a much more informed inventory policy.
Better information. Better parameters. Better decisions.
About HMS
At HMS, I help manufacturing businesses develop practical approaches to inventory management, MRP, supply planning and demand forecasting.
The objective isn't simply to reduce inventory.
It's to understand why inventory is where it is, what it is protecting, and whether it is actually doing the job it is supposed to do.
Comments