How to Calculate Safety Stock for

Inventory Management

An EazyStock Guide to
Optimized Inventory
Management

© EAZYSTOCK
 Contents
Page
3 The Basics of Safety Stock

4 The Importance of Safety Stock

5 The Challenges of Calculating Safety Stock

6 3 Common Methods for Calculating Safety Stock

9 Statistical Calculations for Safety Stock

12 Examples of Statistical Safety Stock Formulas

18 Dynamic Safety Stock

19 Summary

20 Appendix

2 © EAZYSTOCK
 The Basics of Safety Stock
Determining appropriate stock levels is one of the most important
and challenging tasks faced by inventory managers. If you carry too
much, you’ll tie up working capital. Carry too little and you’ll risk
potential stock-outs and unhappy customers.

There needs to be a balance between investing the right amount of capital in

inventory and achieving service level targets (a measure of stock availability).
One key challenge – and an important piece in the puzzle – is calculating safety
stock levels.

Safety stock, also called buffer stock, is important at all stages of the supply
chain. Safety stock is the extra layer of stock (raw materials or finished goods)
that businesses carry to mitigate the risk of run-out due to uncertainties in
supply or demand. Put simply, safety stock is used to prevent service levels
from being negatively impacted by unforeseen changes in demand or supplier
lead times.

In this guide we’ll look at various ways to calculate safety stock, starting simple
and then moving to more advanced, statistical formulas.

3 © EAZYSTOCK
 The Importance of Safety Stock

Safety stock is used to manage any demand or lead time uncertainty and
guard against supply chain or fulfillment disruptions. Deficiency situations
can arise because of fluctuating demand, forecast errors and variation in
suppliers' lead times.

Safety stock is intended to cover any shortfall in cycle stock during the lead
time period. It is an important element of the reorder point formula:

ROP = (avg. consumption × lead time) + safety stock

4
 The Challenges of Calculating Safety Stock
The goal of safety stock is to minimize disruption to order fulfillment, while
investing the lowest possible amount of capital in inventory.

When demand for inventory items is consistent and lead times are reliable, it’s
fairly easy to set safety stock levels that will achieve this. But, when demand and
supply fluctuate, many inventory planners find it much tougher to calculate
safety stock accurately. All too often they chose to use simple safety stock
formulas, which are insufficient to deal with the supply and demand challenges
they face.

Why carry safety stock?

• To protect against demand and

lead time variance

• To compensate for forecast

errors (when demand exceeds
forecast)

• To fulfill demand if there are

disruptions in production or
deliveries

• To avoid stock-outs and back

orders that can lead to lost sales
5
and poor customer service
 3 Common Methods for Calculating
Safety Stock
Here are three simple methods to calculate safety stock, used by many
inventory management teams.

01 Fixed Safety Stock

Many companies set a fixed level of safety stock for their inventory items
e.g. they add a ‘best-guess’ quantity to the reorder point to allow for any
issues.

This number is often set at item group level and based on the judgment or
assumptions of the inventory management team – there are no formal
calculations used. For example, an inventory planner may decide to carry a
week’s worth of safety stock and use last month’s highest week of sales as
the safety stock figure.

Pros/Cons:
While easy to set up and manage, this basic approach often leads to stock
imbalances. This could mean an unnecessarily large amount of capital
invested in excess stock, while stock-outs could be frequent on other items.

6 © EAZYSTOCK
 3 Methods for Calculating Safety Stock

02 Time-based Calculations

A time-based safety stock calculation finds the average sales/demand over a

fixed period of time and uses this value as the safety stock level. For example,
it’s simple to calculate average demand based on last month’s consumption.
In the case below it would be 200 units.

• Week one: 300

• Week two: 150
• Week three: 250
• Week four: 100

Safety stock for the forthcoming month would therefore be 200 units.

Pros/Cons:
Both fixed and time-based calculations take a “one size fits all” approach.
They assume that the forecasted demand will be accurate and that lead times
will remain consistent. In reality, of course, this rarely happens.

For starters calculating demand based on historical sales alone cannot

account for demand variance, particularly for items with erratic or intermittent
demand. At the same time, lead times can be inconsistent due to a range of
issues, such as production downtime or delivery delays.

Using a basic safety stock model will no doubt result in getting the right levels
of stock for some items, but others will see levels too high and others will be
too low.

7
 3 Methods for Calculating Safety Stock

03 Average/Max. Calculation

A more prudent safety stock calculation is to use an average/max. formula,

which accounts for when lead times rise and sales max out.

In our previous example, average sales were 200 units a week but rose to 300
and let’s say that the lead time is 1 week but can be as high as 1.5 weeks.
Using the formula below, we can work out safety stock:

(max. sales × max. lead time) – (avg. sales × avg. lead time)
(300 × 1.5) – (200 × 1)
= 250 units

Pros/Cons:
Problems arise with this formula if the maximum lead time and sales are
considerably higher than the average, which results in the levels of safety
stock being significantly inflated.

While simple to use, a key drawback of all the formulas so far is that they fail
to link back to service levels. Service levels are important as they are
connected to how well you can serve your customers with on-time deliveries.
Any out-of-stock item, even if it’s just one SKU, will lead to an incomplete order
which can be detrimental to customer satisfaction. The service level KPI
therefore closely correlates with customer service, acquisition, loyalty and

8
retention.
 Statistical Calculations for Safety Stock
Statistical safety stock calculations overcome many drawbacks we’ve just
discussed. Despite being more complicated to use, they are much more
accurate. This is because they use probability distributions to model demand
and account for variance.

A probabilistic approach accepts that there is uncertainty when predicting

future events, such as demand volume and frequency, and accommodates for
this by covering for a percentage of all possible inventory requirements.

Statistical safety stock formulas are based on achieving a desired service level.
Service level is the expected probability of being able to satisfy all possible
demand scenarios within a particular period of time.

For example, if you set a service level target of 99%, this means your safety
stock levels will cover for 99% of all probable requests; in other words, you can
give your customers what they want, when they want it, 99% of the time.

9 © EAZYSTOCK
 Examples of Statistical Safety Stock
Formulas
Safety stock formula with demand uncertainty
Let's start with a basic statistical safety stock calculation. This formula
assumes that there are no supplier lead time variations. It only takes into
account variation in demand.

𝑺𝑺 = 𝒁 ∗ 𝝈𝒅 ∗ 𝑳𝑻
Key:
• SS = Safety stock level
• Z = Z-score
• σd = Standard deviation of demand
• LT = Average lead time

This formula introduces two statistical elements – the Z-score and standard
deviation. Let’s break down each element in turn.

Z-score
The service factor, or Z-score is based on your service level target – balancing
inventory costs with the risk of a stock-out. The higher the desired service level,
the more safety stock is required.

Once you set you service level target you can work out your Z-score using a
normal distribution table.
10
 Examples of Statistical Safety Stock Formulas

Normal distribution is used to allow you to predict the probability of selling a

certain quantity of stock. With normal distribution the probability of an
occurence, if graphed, is a bell shaped curve, where all data is equally spread
out around a central value with no bias left or right.

For example, if you sell an average of 1000 units, you are as likely to sell more
than 1000 next month as you are to sell less than 1000. The probability
decreases the further from the center of the curve e.g. the probability of selling
2000 is lower than selling 1500, as is the probability of selling 500 over 750.

As the chart on the next page shows, if you decide to set your service level
target at 50% you don’t need safety stock because you have a 50/50 chance of
selling more or less than the average next month.

11
 Examples of Statistical Safety Stock Formulas

However, if you want a service level of 95%, the normal distribution will give you
a Z-score (a multiplier coefficient) of 1.64.

Probability of no
stock-outs)

5% probability
Safety stock of stock-outs

Average demand
0 1.64 Z-score
Service level Service level
target of 50% target of 95%

The Z-score can easily be determined using a service factor table (as shown on
page 20), or you can use Excel function NORMSINV to convert a service level
percentage to a service factor.

The graph below shows the relationship between target service level and the
Z-score. The relationship is non-linear e.g. higher service levels require
disproportionately higher Z-scores and therefore disproportionately higher
safety stock levels.
Z-score

12 Desired service level

 Examples of Statistical Safety Stock Formulas

Standard deviation of demand

Standard deviation is used in statistics to measure how spread out a certain
set of numbers is from the average. This allows the safety stock calculation
to take into account demand variance.

Standard deviation of demand is calculated like this:

1. Determine the mean (average) of a set of numbers
2. Determine the difference between each number and the mean
3. Square each difference
4. Calculate the average of the squares
5. Calculate the square root of the average

Let’s take a working example. Mary works at The Plumbing Superstore. Below
you can see her forecasted demand for a line of showerheads. The average
weekly demand is 525 units.

Difference between
Forecasted demand
Period actual and forecasted Difference squared
shower head A
demand
Week one 550 25 625

Week two 500 -25 625

Week three 650 125 15625

Week four 400 -125 15625

32500

The total of the squared differences is 32,500 units

The average of 32,500 is 8125 units
The square root of 8125 is 90
90 units is the standard deviation of demand per week

13
 Examples of Statistical Safety Stock Formulas

You can also use Excel function STDEVPA to calculate standard deviation. In
safety stock calculations, the forecast quantity is often used instead of the
mean in determining standard deviation.

Always ensure that you use the same unit of measurement for both the
demand and lead time calculations e.g. if your standard deviation of demand is
calculated in days, then calculate your average lead time in days too!

Let’s now put all these elements together, using The Plumbing Superstore as
an example. Mary wants to achieve a service level for her showerheads of 95%.
In the service factor table on page 20, this gives a Z-score of 1.64. The
workings on the previous page show the standard deviation of demand is 90
units per week and average lead time is 3.5 weeks.

𝑺𝑺 = 𝒁 ∗ 𝝈𝒅 ∗ 𝑳𝑻

𝑺𝑺 = 𝟏. 𝟔𝟒 ∗ 90 ∗ 𝟏.87
𝑺𝑺 = 𝟐𝟕𝟔 𝒖𝒏𝒊𝒕𝒔

Pros/Cons:
If your supplier lead times are fairly stable and you rarely have delivery delays,
this formula will work well. But in practice, this is often unlikely and you need to
add factors to compensate for variance in your supply chain.

14
 Examples of Statistical Safety Stock Formulas

Safety stock formula with demand & lead time

uncertainty

In the previous equation, safety stock was only used to mitigate demand
variability. When variation in lead time is also a concern, the equation must
be expanded. The formula below does just that:

𝑺𝑺 = 𝒁 ∗ 𝝈𝟐𝒅 ∗ 𝑳𝑻 + 𝒙𝟐 ∗ 𝝈𝟐𝑳𝑻

Key:
• SS = Safety stock level
• Z = Z-score
• σd = Standard deviation of demand
• LT = Average supplier lead time
• x = Average demand
• σLT = Standard deviation of supplier lead time

Let’s now look at a working example of this formula, using The Plumbing
Superstore.

15
 Examples of Statistical Safety Stock Formulas

Mary still wants to provide a 95% service level for the showerheads.

▪ Average demand is 525 units per week

▪ The standard deviation from demand is 90 units
▪ The average supplier lead time is 3.5 weeks, but is as fast as 2.5
weeks and as slow as 4.5 weeks
▪ The standard deviation in lead time is 0.79 weeks (approx. 5.5 days
stock)

𝑺𝑺 = 𝒁 ∗ 𝝈𝟐𝒅 ∗ 𝑳𝑻 + 𝒙𝟐 ∗ 𝝈𝟐𝑳𝑻

𝑺𝑺 = 𝟏. 𝟔𝟒 ∗ 𝟗𝟎𝟐 ∗ 𝟑. 𝟓 + 𝟓𝟐𝟓𝟐 ∗ 𝟎. 𝟕𝟗𝟐

𝑺𝑺 = 𝟏. 𝟔𝟒 ∗ 𝟐𝟖𝟑𝟓𝟎 + 𝟏𝟕𝟐𝟎𝟏𝟖

𝑺𝑺 = 𝟏. 𝟔𝟒 ∗ 447.6

𝑺𝑺 = 𝟕𝟑4 units

16
 Examples of Statistical Safety Stock Formulas

As you can see, the safety stock level is now much higher, as lead time variance
is also being taken into account.

Pros/Cons:
This formula is based on the assumption that the difference between the
Assumptions:
forecast and the actual demand follows a normal probability distribution. If the
• You
forecast want
error tonot
does provide
have aanormal
95% service levelthe
distribution, toquality
your customers
of the result will
The demand for your product is 10 units per day
be• diminished.
• Your standard deviation in demand is 2
Yourthat
• items
For supplier has an average
have intermittent demandlead timesuch
patterns, of 15
as days
slow and lumpy
• Theyou’ll
demand, standard
achievedeviation in leadusing
higher accuracy timePoisson
is 3 days
distribution or negative
binomial distribution.

Choosing the right safety stock model

Whichever safety stock calculation you use, it’s very important to carefully test
it prior to final implementation. This is so you can ensure that it works correctly
and to analyze what impact it will have on inventory levels and cashflow.

While more sophisticated models lead to more accurate safety stock levels,
they also take time and resources to implement. Manual calculations are also
frozen in time and virtually impossible to implement at SKU level.

A solution is to use software that will automatically do this for you. A good
inventory optimization tool will reduce the time involved in undertaking manual
calculations and avoid the risk of human error and the associated costs.

17
 Dynamic Safety Stock
An inventory optimization tool such as EazyStock takes statistical safety stock
calculations to the next level.

Safety stock levels are set at SKU level and consider a number of key elements:

Demand type: EazyStock uses a range of statistical algorithms and probability

distributions e.g. normal distribution, Poisson distribution, empirical distribution
and negative binomial distribution. The most appropriate algorithm is chosen
based on each item’s demand type. So whether an item’s demand is fast, slow,
lumpy or erratic, EazyStock will calculate safety stock as accurately as possible.

Supplier lead times: EazyStock’s dynamic lead time feature reviews purchase
order history and alerts you when supplier lead times deviate from the norm.
You can then request the system automatically increases safety stock levels, to
help achieve target service levels.

Safety stock calculations are automatically reviewed and updated daily (when
required) based on service level targets, demand variance, forecast accuracy,
supplier lead times and order frequency. When any of these factors change,
EazyStock dynamically updates safety stock algorithms and adjusts final safety
stock levels accordingly.

With a tool such as EazyStock, you no longer need to calculate and update
safety stock manually and enter this into your ERP. Instead EazyStock does all
the math for you.

18
 Summary
Today’s marketplaces and supply chains are more dynamic and uncertain than
ever. Demand and supply volatility is almost inevitable, making safety stock a
necessity for many businesses.

It’s important to calculate safety stock levels as accurately as possible, so you

have the right balance between inventory investment and stock availability e.g.
hitting your service levels.

Statistical safety stock calculations are much more accurate than fixed safety
stock or time-based methods, as they allow for more variance in demand and
lead times. However, using these more complex formulas in manual
spreadsheets is time-consuming and can lead to errors – especially if you
manage a large number of SKUs.

Companies who really want to optimize their safety stock and inventory levels
should consider an inventory optimization software solution.

This will plug into existing ERP or stock control systems and automatically
calculate forecasts, safety stock, service levels and reorder points, giving
operations teams the data and the time to make informed, strategic supply
chain decisions.

19 © EAZYSTOCK
 Appendix:
Service factor (Z-score) table

Target Service Service Factor Target Service Service Factor

Level (Z-score) Level (Z-score)

20 © EAZYSTOCK
 Powerful Inventory
Optimization
MADE SIMPLE
Book a free demo and learn how to
optimize your safety stock

https://info.eazystock.com/request-a-demo

21 © EAZYSTOCK