Onesphore RAKOTOMALALA

Linkedin.com/in/-r-serzo-onesphore
 What is Demand Forecasting?

➝ Demand Forecasting is the process of predicting future customer demand for a

product or service based on historical data, market trends, and other relevant

information.

The primary objectives are to :

predict sales manage inventory plan production.

Why is it important ?

Demand forecasting is important because it directly influences :

Decision- Inventory Business Cost Customer

making management strategy efficiency satisfaction
 Factors to Consider in Demand Forecasting

➝ Several factors influence demand forecasting, including:

Internal Factors:

Historical Sales Data:

Past sales performance is a primary indicator of future demand.

Product Lifecycle:
The stage of the product lifecycle (introduction, growth, maturity, decline) affects

demand patterns.

Marketing Campaigns:
Promotions and advertising efforts can significantly impact demand.

External Factors:

Economic Conditions:
Economic trends and consumer confidence influence purchasing behavior.

Seasonality:
Seasonal variations impact demand for certain products.

Market Trends:
Shifts in consumer preferences and market trends can affect demand.
 Demand Forecasting M ethods

➝ Demand forecasting techniques can be categorized into two main types:

quantitative and qualitative methods.

1 Qu antitative Methods

Quantitative methods rely on historical data and mathematical models to predict future
demand.

➝ These methods are often more objective and data-driven:

Moving Averages

Exponential Smoothing
Time Series Analysis ARIMA (AutoRegressive Integrated Moving Average)

Regression Analysis

Econometric Models
Causal Models

Neural Networks

Random Forests

Machine Learning & AI Support Vector Machines

 Demand Forecasting M ethods

2 Qu alitative Methods

Qualitative methods rely on expert judgment and market research.

➝ These methods are more subjective and often used when historical data is limited or
unreliable:

Aforecasting technique that uses a panel of experts to predict future

events or trends.

im to combine diverse expert opinions and reach a more accurate

Delphi Method
A

forecast through structured feedback and revision.

Surveys and Questionnaires

Fo cus Groups
Market Research

Analogous Forecasting

Scenario Analysis
Historical Analogy
 S teps for Effective Demand Forecasting

➝ To create an effective demand forecasting model, follow these steps:

Define Objectives and Scope

1
Determine what you want to achieve with the forecast (e.g., inventory

management, capacity planning, financial planning).

Define the time horizon (short-term, medium-term, long-term) and the level of

detail required (product level, category level, market level).

G ather and Prepare Data

2 l vant historical sales data, market trends, and economic indicators.
Collect re e

Remove o utliers, correct errors, and handle missing values to ensure data quality.
Segment data based on different criteria such as product categories,
geographic regions, and customer segments.

Select Forecasting Methods

Quantitative Methods: Choose appropriate statistical and mathematical
3

models based on the availability and reliability of historical data.

Qualitative Methods: Use expert judgment, market research, and scenario

analysis when historical data is limited or when anticipating significant

market changes.
 Steps for Effective Demand Forecasting

I mplement the Forecasting Models

Build Models: D evelop and test
different forecasting models using selected
4 quantitative and qualitative methods.

Validate Models: Use backtesting and cross-validation to assess the accuracy

of the models. Compare forecasted values with actual historical data.
Refine Models: djust model parameters and improve algorithms based on
A

validation results.

I ntegrate External Factors

M ark t Tr n
e e ds: Incorporate current market trends, consumer behavior, and
5 competitive actions.
Economic Indicators: Include macroeconomic factors such as G P gro D wth,
inflation rates, and employment levels.
Seasonality and Events: ccount for seasonal variations and special events
A

that could impact demand.

Gener ate Forecasts

6 Run Mode s: l Use the finali ed models to generate demand forecasts for the
z

defined time horizon.

Aggr ge ate R
esults: Combine forecasts from different models and methods
to create a comprehensive demand forecast.
 S teps for Effective Demand Forecasting

Review and Adjust Forecasts

7
R gula R v w: Periodically revie
e r e ie w and update forecasts based on new data
and changing market conditions.

Adjustm ts: Make necessary adjustments to forecasts based on recent

en

sales data, expert insights, and market feedback.

Communicate and Collaborate

8 Share Forecasts: Distribute the demand forecasts to relevant stakeholders (e.g.,
supply chain, production, finance, marketing).

Collaborate: Foster collaboration between different departments to ensure

alignment and effective decision-making.

Monitor and Improve

P rf rm c
e o an r g Continuously monitor forecast accuracy and track
e Monito in :

9 deviations from actual demand.

Feed b ck L p Establish a feedback loop to learn from forecast errors and
a oo :

improve future forecasting processes.

Continuous Improvement: Regularly refine forecasting models and methods
based on performance analysis and evolving market conditions.
 L A L A L A L A LA LA LA
L A L A L A L A L A L A L A
M A M A M A M A M A M A M A
OTO OTO OTO OTO OTO O TO O TO
R AK R RA
K AK R AK RA
K
R AK R AK
s p h o r e
s p h o re
s p h o reDemand Forecasting KPI
s p h o re
s p h o rs
e
s p h o re
s p h o re
n e n e n e n e n e n e n e
O O A O A O A O A O A O
A L A L A L A L A L A LA A LA
A L A L A L A L A L A L A L
T OM T OM T OM T OM TOM T OM T OM
➝ ToKe Ovaluate theKOeffectiveneKss O of demand K Oforecasting,KOconsider the KOfollowing KKPOIs:
R A R A RA R A R A R A R A
o r e o re o re o re o re o re o re
s p h s p h s p h s p h s p h sp h s p h
n e oren ca e st Acc racy ne n e Bias ne n e n en n e tory Metrics
O O F

A O A
u
O A O A O A O I v

A L A L A L A L A L A LA A LA
Mean A sol te ercL enta e Error (MA EL Lias L @OnesphoreRkt Ln entory T rno erAL L
A A ) orecast B
A A A A
OMmagnitude of forecastTerrors
OMas a OMs M OM ➟ M su s Tf OnMinv n y is s ld nd plTOc Md v
b u P g P F I v u v

f c s s Oc nsis n ly ig
percentage ofKactual
T
➟ Measures the average
O values. O
➟ IT
lO
ndic
n c u l v lu s. KO
T ate
OTwhether ore a t
O
are o te t
O
h her or ea re how o te e tor o a re a e o er

A A K A K A ower tha
A K
a t a a p i d.
e
A K A K a er o

e R e R e R e R e R e R e R
o r
h Absolute Deviation o r
h (MAD) sph o r o r
h nal h o r o r
hDays o n entorysponh Hand o r
s pMean
s p
e ➟ C lcul s vneg bs lu diff nc s bne n
Trackin Si
s p s p g
s
g
p fI v

n ➟M n sue s f c s bi s v im ,nig e lig ing c nsis n ne ➟ Indic s numbne f d ys inv n y ill l s

O f c s d nd cO
a ate the a era e a

A O o te

A vO
ere e etwee

A O ea

A
re
Oore a t a o er t

A .O
e h h ht o te t ate the er o a e tor w a t at the

LA LA
u l v lu s. - und -f c s ing. cu n s l s
L L L L L
ore a te a a t a a e o er or er ore a t rre t a e rate
@OnesphoreRkt
L A L A L A L A L A L A L A
A A A A A A A
Mean SquaredOEM rror (MES) OM MeaO nMorecast Error (M O E)M OM Stocko t RatTeOM @OnesphoreRkt TOM
O T O T T T
OC lcul s v g KOin f c s v lu s, indic ingKO
F
T F

O ic inv n y l vKlsOf ll
u

➟
A K A K
Measures the average of the squared differences between
A K➟
A a
A ➟
ate the a era e error f uKncy
A ore a t a e
A at, The req e at wh h e tor e e a to zero

e R e R e R di c i n nd m egniRud f bi s. e R e
indic ing R p n i l f c s ing issue s.R
or or or or or or or
forecasted and actual values, emphasizing larger errors. the re t o a a t eo a at ote t a ore a t e

s p h s p h s p h s p h s p h s p h s p h
e e e e e e e
On O n ice Le el On
Ser O n Lead TO imne O n OCnost Metrics
LA LA LA LA LA LA LA
v v

L A L A L A L A L A L A L A
Ser ice Le el erMcA A A A A A A
v

T O v enta eP

T OM g ore
T OM
ca s t Le ad T im e A cc
T OMracy
F

TOM Ca rr yin C os
T
t
OM
uo n en t ory
T OM g fI v

p c n Og f cus m d m nd O OM su s ccu cy Kf fO c s s b s d n imKO l cO O s g,

➟
s ck u s. A
The
K er e ta e o
A K m i u
to er e
K➟
A ui d fulfill d s.RA
a et w tho t ea re
A
the a ➟ ra o
K s f
insu nc ,And bs l sc nc c s s.
ore a t a e
K
lding inv n y, including
A
o the t e The tota o t o ho e tor tora e
to
e R
o t
e R e R e
req
e
re
Rto or er
e R e R ra e a o oe e e o t

h o r h o r h o r h o r h o r h o r h o r
e s p e s p e s p Ordere s p e s p e s p ocko t Costse s p @OnesphoreRkt
On ➟ p c n gOnf cus m Ad s Onfulfill d A ➟ On v g imA b nOnd pl c m An nd On ➟ C sAs ss ci Od ni unning Au f s ck, suc s l s A
Fill Rate Lead Time St u

The er e ta e o to er or er that are e The a era e t e etwee or er a e e t a o t a o ate th r o to to ha o t

nd. L L L L s lL s nd xp di d s ipping. L L
w

c mpl ly f m s ck
o ete ro
L
to
A at ha
L A d liv y.
A
L @OnesphoreRkt AL A
e er
L A L A L A ae a e e te h

M A M A M A M M A M A M A
T O T O T O T O TO T O T O
Order O ent Cycle Time KO
l illm
K
Fu f
KSO pplier Lead Time O
K aria ility u
K O Cost o oo V
KOds Sold (CO S) KO
b fG G

➟ R
im A k n fulfill cus m RAd s, f m d R A ➟ c nsis ncy f l RdAim s f m suppli s, impRcAing ➟ diRA
c c ss ibu bl Rp Aduc i n f g ds
pl cre e @OnesphoreRkt ore f c s li bili oy. re e e c mp ny. e
The t e ta e to to er or er ro or er The o te o ea t e ro er a t The re t o t attr ta e to the ro to o oo

h o a em n d liv y.
e t to e
h o
er r h h h o r
ore a t re a t
h
s ldrby
o h o r o a o a

e s p e s p e s p e s p e s p e s p e s p
O n O n O n O n O n O n O n
Sales MeLtArics L A Oper L A
ati on al E ici en cy L A C o L A
lla or ati on and C LoAmm nicationAL A
L A L A L A L A L A ff
L A L
b u

M A A A A A A A
Sales per C T stO
omer T OM T OectMOrder Rate TOM TOM orecast alTOe M Added ( A) T OM
O
v gK s l s g n
u

d p cus K
O K Oer
K O P f

K O KO KO F V u FV

➟
d m ndR
A ns.
The a era e ae
R A m , indic ing
e erate er
R A ➟
to er
p c n g
at
f
R d As d liv d n im , in
The
R
full, And
er e ta e o or er ➟ M su A
R s e imp v m n in f c A
ere o t
R s ccu cy du
e a ea re the ro e e t ore a t a ra e to
e

o r ea p atter

o re o re i u s.
o re o re
tho t error c
oll re
b iv ff s nd p c
o
ss re m n s.
n nc o a orat e e ort a ro e e ha e e t

h h h h h h h
w

s p
e Sales ariancene @OnesphoreRkt nes p s p s p
eCycle Time e s p s p
e lan Adherencnee s p
O n O O Or O
d n
er O n O n O @OnesphoreRkt
A L A ➟ v g AimLfAm d pl c m n fulfillm LA n . A degree to which the demand A plan is followed, L A
V P

n f cAsL d nd c u l s l s. A A A L L
Lplanning and execution. AL A
A
➟ The
➟ Thediff nc b ere e

A L etwee ore a te

A L a a t a

A L ae

A L The a era e t

A Lindicating alignment between

e ro or er a e

A
e t to e t

T OM T OM T OM T OM TOM T OM T OM
K Oth Rate KO K Oackorder Rate KO @OnesphoreRkt KO KO orecastin Acc Kracy O
Sales ro
RA G w

➟ re p c n g inc s inres l s v
RA e R A B
p c n g ref d s
R A
c nn b fillre
R A C onsens
R A s
eaccuracy of forecasts generated R A
e through collaborative
u F g u

h o The er e ta e

h o rea ep i d,
h o r ae ➟ o er a

h o er o

h o Thed er e ta e o or er

h
➟
oTher that a

h o
ot
r e e at the

e s p fl c
re e ting d m nde nds.
a

e s p tre

e s p im f
e
d
s p pl c m n .
e s
t
p e o or er a e

e s p input
e t from various
e s p
departments.

O n O n O n O n O n O n O n
L A L A L A L A L A LA LA
L A L A L A L A L A L A L A
A A A A A A A
T OM T OM T OM T OM TOM T OM T OM
K O KO K O K O K O KO KO
R A R A R A R A R A R A R A
o r e o re o re o re o re o re o re
s p h s p h s p h s p h s p h s p h s p h
n e n e n e n e n e n e n e
O O O O O O O
 Tools and Software for Demand Forecasting

➝ To support effective demand forecasting, consider these tools and software options:

R -
eal time data p rocessin g, advanced anal ytics , A I and machine learning capabilities, robust
scenario p lannin g, and collaborative p lannin g. analytics, and collaborative planning tools.

dvanced statistical models, real-time data

A Comprehensive demand forecasting tools, I-driven insights, &
A

analysis, and what-if scenario planning. seamless integration with other Microsoft applications.

Machine learning algorithms, predictive analytics, Concurrent planning, real-time simulations, and
and integration ith other Oracle SCM modules.
w end-to-end supply chain visibility.
 Benefits of Demand Forecasting

➝ The benefits of demand forecasting include:

Optimized Inventory Levels

Maintain the right stock to reduce costs and prevent stockouts.

Enhanced Supply Chain Efficiency

Improve production planning and reduce lead times.

Cost Reduction
Lower operational and carrying costs.

Increased Customer Satisfaction

Ensure product availability and timely deliveries.

Better Financial Planning

Enable accurate budgeting and cash flow management.

Competitive Advantage
Respond quickly to market changes and stay ahead of competitors.
 Common Pitfalls in Demand Forecasting

➝ Avoid these common pitfalls in demand forecasting and learn how to address them:

Overreliance on Historical Data

➟ Combine historical data w ith current market trends and expert insights.

Ignoring External Factors

➟ Incorporate economic indicators, competitive actions, and seasonal variations.

Lack of Collaboration
➟ Engage cross-functional teams from sales, marketing, and supply
chain for comprehensive input.

Infrequent Updates
➟ Regularly update forecasts with the latest data and market information.

Neglecting Forecast Accuracy Metrics

➟ Monitor KPIs like M PE and bias to continuously improve forecasting models.
A

Failing to Plan for Variability

➟ Use scenario planning to prepare for different demand fluctuations and
uncertainties.
 Best Practices for Demand Forecasting

➝ Follow these best practices for effective demand forecasting:

Use Multiple Methods

Combine uantitative and ualitative methods for more accurate forecasts.
q q

Regularly Update Data

Continuously update and clean data to ensure forecast accuracy.

Incorporate Market Trends

Include current market trends and external factors in your models.

Collaborate Across Departments

Ensure input from sales, marketing, and supply chain teams.

Monitor Forecast Accuracy

Regularly review and adjust forecasts based on performance metrics.

Scenario Planning
Prepare for various market conditions with scenario analysis.
 + Follow

Found this post useful ?