enterprise analytical management s
TRANSCRIPT
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 1
Enterprise Analytical Management“Business Transformation in the New Era
of the Digital Revolution”
Madrid, November 13, 2019
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 2
Content
◼ Lessons learned from Evolution of Species
◼ Digital Transformation: Driving Forces and Value at Stake
◼ Enterprise Analytical Management - EAM
– Embedding analytics in the business value chain
– Key EAM principles
◼ Round Table
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 3AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
Universe Evolution & Age
-1,00,01,02,03,04,05,06,07,08,09,010,011,012,013,014,015,0
Universe Creation
– 13.8 bn years ago
Water Molecule
– 11.0 bn years ago
◼ Hominids
– 4 m years ago
◼ First Cell
– 3.8 bn years ago
◼ Earth Creation
– 4.5 bn years ago
bn years ago
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 4
Discontinuity, Driven by Disruption, is the Rule
Evolutionary development is marked by isolated episodes of rapid speciation
between long periods of little or no change
Equilibrium Equilibrium Equilibrium
Disruption Disruption
Rapid decline Rapid mass speciation
Source: “Punctuated Equilibrium”, Stephen J. Gould; Digital Transformation, Tom M. Siebel
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 5
Evolutionary Mass Extinction Events
Ordovician-
Silurian
Late Devonian Permian-Triassic Triassic-Jurassic
Millions Years Ago & Percentage of Species Extinct
-75%-86%
-75%-96%
-80% -76%
2000 m 445 m 340 m 250 m 200 m 65 m
Glaciation &
Falling Sea
Levels
Drop of CO2 and
Weather Cooling
Volcanic
Eruptions,
Global Warming
& Oceans
Acidification
Climate Change,
CO2 & CH4
Greenhouse
Effects
Cretaceous-
Tertiary
Asteroid Impact,
Volcanic Activity
& Climate
Change
Great
Oxydation Event
Oxygen
Atmosphere
Aerobic
Life
Pluricellular
Source: Digital Transformation, Tom M. Siebel
Shorten Change Cycles: In the past million years, the world
has experienced disrupted events 10 times
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 6
Same Principles Applies to Evolutionary Enterprise Extinction
Events
Massive
declining of
existing species
&
Rapid
speciation of
new species
External
Driving Forces
Shortening
Change Cycles
Accelerating
Speed Change
• Digital Transformation Forces
• Since 2000, 52% of the Fortune
500 have been either acquired
or bankrupt
• Digital disruption will wipe out
40% of Fortune 500 firms in
next 10 years, say c-suite
execs
• IT sector has grown from
$50bn in 1980 to $3.8tn in 2018
and expected $4.5tn by 2022
• New speciation of players with
different DNA Amazon, Google,
Facebook, Lyft, Zelle, Square,
Airbnb, Twilio, Shopify,
Zappos, Uber,…
As of 2019, the Fortune 500 companies represent approximately two-thirds of the United States's Gross Domestic Product with approximately $13.7 trillion in revenue,
$1.1 trillion in profits, and $22.6 trillion in total market value
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 7AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
Core of Digital Transformation
Confluence of Four Disruptive Technologies
Big Data Internet of Things
Cloud Computing Artificial Inteligence
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 8
Digital Transformation Will Create Trillions of Dollars of Value
Potential Economic Impact
Increase in Global Business and Social Value Timeframe Source
$100 Trillion 2016-2030 World Economic Forum, 2016
Increase in Annual Global GDP
$ 15.7 Trillion (driven by AI) By 2030 PwC, 2017
$ 13.0 Trillion (driven by AI) By 2030 McKinsey, 2018
$ 11.1 Trillion (driven by IoT) By 2025 McKinsey, 2015
$ 3.9 Trillion (driven by AI) By 2022 Gartner, 2018
Source: Digital Transformation, Tom M. Siebel
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 9
Many companies launch analytical initiatives but struggle to
capture tangible value
Pitfalls that we have found so far
Vision People Involvement Persistence
“Wrong vision on
value at stake and
change
management
needed”
“Not having the
necessary
capabilities and
talent”
“Analytics Talent
Data Scientist”
“Lack of
collaboration with
business users
during
development
phase”
“Moving on to
next use case
before value has
been captured”
Data
“Struggling to get
access to data or
use all the data
available”
“Map out a fully
reimagined 3-5
years vision for
the whole value
chain using
analytics rather
than to focus on
current process
pain points”
“Look for talent
beyond data
scientist and hire
translators,
DevOps experts,
Cloud specialist,
data
engineers,…”
“Involving actual
users in the
solution design
from planning to
implementation”
“Developing a set
of KPI’s that
measures
progress from
model
development to
value capture”
“Building a strong
pull process
based on
business needs,
technical solutions
and then the data
needed rather
than “data push”
approaches”
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 10
Embedding analytics management in the value chain is inspired
in key lessons learned from similar great transformations
Embedding Strategic Function into Value Chain
Total Quality Management
Total Quality Management (TQM)
From: Quality is a control function
To: Quality embedded into all value chain
Logistics SalesProcurement ManufacturingNew
Product
Quality
Control
Logistics SalesProcurement ManufacturingNew
Product
◼ Kaizen ◼ Heijunka ◼ Jidoka
Enterprise Risk Management
Enterprise Risk Management (ERM)
From: Risk is a central function
To: Risk embedded into all value chain
Channels SalesControl ProductsRisk IT
Channels SalesControl Products IT
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 11
EAM management is based on
Simultaneous Engineering
EAM is a Factory that integrates all
key elements
Analytical Vision & Strategy
Decision Making Processes
(Analytical Areas)
Analytical Areas & Cases
Development
Implementation and Value
Capture
Analytical
Vision
Business
Objectives &
Value Creation
People, Organization & Technology
Governance Model
Enterprise Analytical Management (EAM)
Data (Pull vs. Push)
Analytical
Knowledge
Enterprise Analytical Management (EAM) encompasses all
elements to catalyze analytical transformation
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 12
Adopting EAM across all lines of business and functions requires a
strong, coordinated strategy and a robust operating model
EAM Vision
Unified commitment from
all levels of management
◼ Leadership team is
completely aligned on an
analytics vision & strategy
◼ Vision to integrate
analytics across all
operations
◼ HR policy: making
analytics expertise a
requirement for leadership
position
◼ Securing buy-in further
down the organization
Embed & plug analytics
into the critical strategic
areas of the company
◼ Assessment of impact
within market positioning,
client loyalty and profit
impact
◼ Extract the value of
analytics
– Make analytics user-
friendly
– Embrace analytics as
essential tool that
challenges established
thinking
◼ Short & Long term plans
Spend more on analytics
and plan to increase
investment further
◼ Analytics budget has an
important participation
– Data
– Technology
– Analytics talent
– Embedding analytics
into business-process
workflows
◼ Increase analytics spend
in the next years
◼ Combine in-house spend
with third-party alliances
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 13
EAM is essentially seeing an evolution of the science of
Decision Making
EAM Decision Making Roadmap
In every organization, thousands of decisions affect business outcomes every
day and all of these could be informed by data insights, so we need to identify
those decisions that will drive the most value
Inventory of
Decisions
◼ List of key
decisions of
the business
◼ Decisions
definition:
– Frequency
– Number
– Impact
◼ Map of
decisions
◼ Selection &
prioritization
◼ Transform
decision into
Analytical
Areas
◼ Define
Analytical
Areas
◼ High-Level
business
case
◼ EAM
roadmap:
sizing and
planning each
Analytical
Area
Classify
Decisions
Set
Analytical
Areas
Petrochemical Company
(Analytical Areas)
Daily
Annual
Monthly
Weekly
Pricing
Low High
Predictive
Maintenance
People
Scheduling
Synchronous
ManufacturingProduct
Portfolio
Annual
Investments
Fre
qu
en
cy
Business impact
Size of circle show Number of Decisions / time
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 14
EAM needs to define data management approach
Develop a data strategy that supports the wider analytics strategy and avoid
investing heavily in collecting & cleaning data before having a clear strategy
Data Push vs. Pull as key approach in Analytics Strategy Definition
◼ With all data available,
what kind of business
issues can be solved?
Data ModelBusiness
impact
Trigger questions
◼ Knowing my business
needs, how analytics can
help and which data is
really needed?
Push
Business
needsData ModelPull
Business
Value
Captured
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 15
In the last 15 years Advanced Analytics have move from the back office
to the board of the top tier companies around the world
Advanced Analytics EvolutionC
om
pe
titi
ve
Ad
va
nta
ge
Sophistication of Inteligence
Standard Reports
“What Happened”
Add hoc reports
“How Many, How Often, Where?”
Query Drill down
“What exactly is the problem?”
Alerts
“What actions are needed?”
Statistical Analysis
“Why is this happening”
Forecasting/extrapolation
“What if this trends continue?”
Predictive Modeling
“What will happend next?”
Experimental Design
“What happend if we try this”
Optimization
“What is the best that
can happend?”
Machine Learning
“What can we learn
from the data?”
“Competing On Analytics”, Thomas H. Davenport
Autonomous
Analytics
Optimization
Predictions
Segmentation
& Classification
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 16
EAM develops a professional team with a combination of the key
skills required
EAM Main Analytics Roles
Business
Translator
Model
Strategist
Analytics
Analyst
Data
Engineer
Translate
business
problems
Scientific
supervisor
Model and
tools
development
Extract,
clean and
prepare the
data
◼ Focuses on understanding business requirements
◼ Serves as a link between a client and the analytics team
◼ Has experience managing analytical/quantitative
projects and good understanding of
assumptions/limitations of the employed models
◼ Has advanced knowledge of mathematics and statistics
◼ Experience building and testing statistical models
◼ Experience with statistical packages and tools
◼ Has an advanced knowledge of coding and statistics
◼ Advanced programming skills in statistical/programming
languages like R, Python, Scala, SAS (or generic
purpose like C++, Java)
◼ Knowledge of visualization techniques and libraries
◼ IT or Computer Science Background
◼ Advanced knowledge of data management and
architectures (Structured, non-structured, real time)
◼ Experience in ETL
◼ Project, team
and customer
relationship
management
◼ Define modeling
approach
◼ Ensure model
quality
◼ Model
development
◼ Unitary testing
and validation
◼ Provide quality
data to the
Analytics
modeling team
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 17
EAM team is organized as a shop floor which enables collaborative
work
Analytical U-Cell
Client
Mo
de
l D
ev
elo
pm
en
t
Data CurationEnterprise Analytical
Management
Model RunBusiness
improvements
Analytics
Analyst
Data Engineer
Model
Strategist
Business
Translator
Generally, we organize as many U-Cells as Analytical Areas
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 18
EAM has to embed analytics inside the company organization
EAM Organizational Interaction Development Best Practices
Analytical transformation provided by EAM has to be manage as a top
strategic topic of the top management agenda
◼ Embed critical EAM-CDO capabilities across
the whole company organization, not just in
an analytical center of excellence
◼ Secure senior management commitment
and appoint the right leader to act as a
bridge (first level function)
◼ Establish the right culture with CEO and top
executives emphasizing the importance of
analytics
– i.e. Ask top executives to come up with at
least 3 ideas about how analytics would
improve their business areas
◼ Implement a digital and analytics
organization that fits the company’s:
governance model, maturity, potential for
standardization and best-practice sharing
Centralized1 Decentralized2 Hybrid3
CEO
BU3BU1 BU2EAM-CDO
AALAAL AAL
1
3
2
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 19
EAM Governance is based on developing several lines of action
Analytical Areas Scope Lines of Action Governance & Roles
Attack
(Efficacy)
Defense
(Efficiency)
An
aly
tical A
RE
AS
◼ Revenue Growth
◼ Demand Forecast
◼ Customer Experience
◼ Product Development
◼ Pricing Optimization
◼ Loyalty Increase
◼ Cost Optimization
◼ Risk Management
◼ Resource Allocation
◼ Regulatory &
Compliance
◼ Capital Management
1st Line
of Action
◼ EAM Model Owners & Model
Users that identify cases and
jointly develop or acquire,
operate and maintain model
2nd Line
of Action
◼ EAM Central Group provides
view, knowledge, support,
resources & challenge to the
first line of action. Also
provides strategic analytics
guidance
3rd Line
of Action
◼ EAM Auditors: review the
quality & risk of the models
developed according to
company EAM program
requirements
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 20
EAM Auditing is a key role because they have to assess the risk
of managing each model
All models will be assigned a risk rating reflecting their risk to the company
taking into account model’s business use and materiality
Sources of Analytical Model Risk
(EAM Auditing Team)
Data
◼ Data quality
◼ Data completeness
◼ Data availability
Method
◼ Proper statistical method
◼ Estimators confidence
intervals
◼ Hypothesis,
approximations and
simplifications
◼ Validation methodology
Usage
◼ Continuous calibration
◼ Lack of update
◼ Usage outside current
data range
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 21
The “build it yourself” approach requires numerous integrations
resulting in a high degree of complexity
Batch
Processing
Azure Analysis Services
Amazon Sage Maker
Amazon DynamoDB
Amazon RDS
Stream
Processing
Machine
Learning
Framework
and Services
UI and Data
Visualization Tools
Data Exploration
Tools
Platform
Management
Services
Data
Integration
No SQL
Storage
Data Connectors, APIs,
Enterprise Application
Infrastructure
Relational
Database
Cloud Object
Storage
Databricks
Amazon
DynamoDB
Stream
Analytics
Azure Events Hubs
Amazon
Lambda
Azure Data
Lake
Amazon S3
SQL Data
Warehouse
Amazon
Aurora
Non-Relational
Database
Hadoop
Storage
Application
Development
Tools
Azure Data
Explorer
Trusted
Advisor
Amazon API Gateway
Amazon CloudFront
Amazon CloudWatch
Elastic Load Balancing
AWS CloudTrail
Azure Data Factory
Azure Data Catalog
Amazon
DynamoDB
AWS IoT
Amazon Kinesis
HDInsight
AWS Labs Formation
AWS IoT Events
Amazon SNS
Amazon SQS
Amazon Firehose
AWS CloudFormation
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 22
Building the Platform: Model-Driven Al Architecture…
A model-driven architecture provides an abstraction layer that vastly simplifies
and accelerates the development and deployment of AI and IoT applications
Logging
Auto Scaling
Integrated
Development Tools
Microservices &
Applications
Profiling
Device IntegrationIntegration Services
Stream Services
Encryption
Authentication
Message Queue
RDBMs APIs
Authorization
Monitoring
Multi-Dimensional
Distributed
In-Memory Service
AI / Machine
Learning
SchedulingTime Series
Service
Key-Value
Store
M2M
& IoT
End
User
Business
Analyst
Application
Developer
Data
Scientist
Data
Engineer
Distributed File System
Source: Digital Transformation, Tom M. Siebel
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 23
… and integrating it: Multi-Cloud Deployment
Organizations require an AL architecture that enables them to deploy
applications on multiple public cloud platforms as well as on bare metal behind
the firewall in a private cloud or data center
Pa
aS
Iaa
S
Amazon
SQS
Amazon
S3
Amazon
Redshift
Amazon
Kinesis
Amazon
DynamoDB
Azure
Stack
Azure
Blob
Storage
Azure
Event
Monitor
Azure
Database for
PostgreSQL
Azure
Event
Hub
Mpas
Spanner
BigQuery
Cloud
Speech
Cloud
Translation
Intel
Nervana
Intel Deep
Learning
System
Intel
Computer
Vision
Intel
Movidius
Intel
GNA
Predictive
Maintenance
Inventory
Optimization
Energy
Management
Precision
Health
Anti-Money
Laundering
Mfg. Quality
OptimizationCRM AI
Sa
aS
Cu
sto
mer
Exte
nsio
n
SaaS APPLICATIONS CUSTOMER APPLICATIONS
Logging
Metadata Management
MapReduce
Continuous Analytic
Processing
Queue
Stream
Batch
Access Control
Al
Suite
Integrated
Development Tools
Microservices &
Applications
Source: Digital Transformation, Tom M. Siebel
AL
TA
IRM
an
ag
em
ent C
on
su
lta
nts
C3AID_2018 / 24
Closing Summary
Context Leadership Set Goals Organization
◼ Disruptive
Change
◼ Frequency
◼ Velocity
◼ Agenda
◼ Resources &
Investments
◼ Decision
Making
Process
◼ Short term &
Long Term
◼ Bet on
growing
knowledge
on the
business
◼ Move on
Technologies
◼ Short term
lab
◼ Make vs.
buy