2 jørgen steines platon
TRANSCRIPT
-
8/15/2019 2 Jørgen Steines Platon
1/40
WWW.PLATON.NET
Hvordan sikres (mere) værdi af
Business Intelligence projekter? [email protected]
1
-
8/15/2019 2 Jørgen Steines Platon
2/40
© Platon
● A leading Independent Information Management consulting company
● Headquarters in Copenhagen, Denmark
● 220+ employees in 9 offices
● 300+ clients in 8 countries
● Founded in 1999
● Employee-owned company
Platon – The Company
“Platon received good feedback in our satisfaction survey. Clientscited the following strengths: experience and skill of consultants,business focus and the ability to remain focused on the needs of
the client, and a strong methodological approach”
Gartner Jul y 2008
-
8/15/2019 2 Jørgen Steines Platon
3/40
© Platon
Side 3
• Fokus på Business Intelligence og Master Data Management• Unikke internationale eksperter• Netværksreception med underholdning af Jonatan Spang• Afsluttende netværksmiddag
Vi glæder os til at se dig og dine kollegaer d. 12. oktober 2011.
Nordens største Information Management konference: Keynote:
JAMES TAYLOR
Vi er stolte over at annoncere årets
keynote-taler: én af de største Business
Intelligence-guruer hele vejen fra San
Francisco i USA. James er en ledende
ekspert og forfatter indenfor regelbaseret
beslutningsstøtte (Decision Management
& Predective Analytics) og en anderkendt
keynote-taler ved diverse globale
konferencer.
28 unikke præsentationer, bl.a:
Book allerede datoen i din kalender i dag!
Du kan følge udviklingen af programmet påwww.IM2011.net.
-
8/15/2019 2 Jørgen Steines Platon
4/40
© Platon
Agenda
● What is BI● BI Governance
● BI Adoption
● BI requirement specification
● Summing up
4
-
8/15/2019 2 Jørgen Steines Platon
5/40
© Platon
Data Warehouse & Business
Intelligence
Page 5
Analyticalapplications
OLAP
DataMining
Enterprise
reporting
Business Intelligence
?!?
DataWarehouse
The term Business Intelligence (BI) covers
the use of information to drive businessinsight.
Basically it‟s about providing a betterfoundation for decision makers by providing
information in the right form, in the right
quality, at the right time.
The term Data Warehouse covers the management of data
Data is extracted from operational systems and integrated inthe Data Warehouse environment in order to provide an
enterprise wide perspective, one version of the truth.
-
8/15/2019 2 Jørgen Steines Platon
6/40
© Platon
Drivers for Business Intelligence
Page 6
Procurement
Production and logistics
Sales Service
HRMany types of employees
High employee turnoverBad employee satisfaction
Decreasing competencies
Need for collaboration
. . .
MarketingDecreasing market share
Missing cross/up-salesBad campaign response
Slow time to market
CRM aspirations
. . .
ITHeterogeneous infrastructure
Data quality issuesReporting back-log
Project delivery issues
. . .
FinanceCash flow problems
Low profitabilityLosses on debts receivable
Inflexible planning process
CPM aspirations. . .
CEOLow profitability
Decreasing market shareSlow reaction to threats and opportunities
Challenges implementing business strategyChallenges with mergers
. . .
Falling revenueMissing cross/up sales
Increasing COGS
Missed opportunities
Bad forecasting
Decreasing prices
Complex markets
. . .
Bad customer satisfaction
Increasing response time
More complaints
Random service levels
. . .
Quality issuesFalling service levels
Increasing lead time
Rising inventory levels
Resource bottlenecks
Increasing distribution costs
Inefficient processes
Extended value chain aspirations
Process outsourcing
Just-in-time aspirations
. . .
Unattractive pricesBad service levels
Lack of supplier insight
Lack of market insight
Rising stock levels
. . .
The
MultidimensionalManager:
”24 Ways to Impact yourBottom Line in 90 days”
-
8/15/2019 2 Jørgen Steines Platon
7/40© Platon
An example
Page 7
-
8/15/2019 2 Jørgen Steines Platon
8/40© Platon
● It is estimated that 10% of all insurance claims are attempts to fraud
● For Codan this equals 400 mill. DKR per year
Predictive analytics
Codan - Fraud
Insurance claim- collect information
Standard caseLoss consultant
investigates
??
Insurance claim- collect information
Standard case
Loss consultantinvestigates
Risk of fraud is predictedthrough a
data mining tool
Page 8
-
8/15/2019 2 Jørgen Steines Platon
9/40© Platon
Predictive analytics
Codan - Pricing
Old model – postal codes New model – 100 x 100 meter cells
Low risk
High risk
• Several parameters to determine the risk• Only a few from the customer• The rest is based on data
Page 9
-
8/15/2019 2 Jørgen Steines Platon
10/40© Platon
Agenda
● What is BI● BI Governance
● BI Adoption
● BI requirement specification
● Summing up
10
-
8/15/2019 2 Jørgen Steines Platon
11/40© Platon
IT Governance
11
IT Governance: Specifying the decision rights and
accountability framework to encourage desirable
behavior in the use of IT
Governance Corporate governance
● The opposite of Governance: Anarchy (from Greek: ἀναρχίᾱ anarchíā,"without ruler“)
● "No rulership or enforced authority.” ● "Absence or non-recognition of authority and order in any given sphere.”
● "Act[ing] without waiting for instructions or official permission... The root ofanarchism is the single impulse to do it yourself: everything else follows fromthis.”
-
8/15/2019 2 Jørgen Steines Platon
12/40© Platon
BI Governance
12
BI Governance is the framework and processes fordetermining the priorities, deployment practices, and
business value of enterprise business intelligence initiatives.
How do we get exe-
cutive level awareness
and support?
How do we resolve
conflicting interests?
Who decides what to
work on next?
How can we be more
proactive and
anticipate changing
business needs?
How do we quantify
and track the values of
our BI investments?
-
8/15/2019 2 Jørgen Steines Platon
13/40© Platon
BI Governance
- Business and IT standpoints
13
BusinessInnovation
Flexibility
Responsiveness
Train
Users
Recommend
Actions
Analyze
informationDBA
Develop
ETL
Datamodelling
RequirementSpecs
Design
Front endDevelop
reports
User
support
Execute
Bus. Proc.
Standards
for reporting
IT
Cost effectiveness
Operational efficiency
ReliabilityScalability
BICC
IT
DW Business
unit
Business
unit
Business
unit
-
8/15/2019 2 Jørgen Steines Platon
14/40© Platon
BI Governance
- Organisational structure
14
ProgramBoard
Coordinate &
prioritize
Coordinate& prioritize
Program level
Project C
SteeringCommittee
Project B
SteeringCommittee
Project A
SteeringCommittee
Project level
Operation level
BICC DW
-
8/15/2019 2 Jørgen Steines Platon
15/40© Platon Page 15
IT Governance
Infrastructure and
operational applications
BI Governance
Business performance
and decision support
Governance relationships
Data Governance
Information quality
and processes
● The purist would claim they are independent
-
8/15/2019 2 Jørgen Steines Platon
16/40© Platon Page 16
IT Governance
Data Governance BI Governance
Infrastructure
and operational applications
Information qualityand processes
Business performance
and decision support
Business strategy alignment
Legal compliance
Knowledge management
Project portfolio management
Service Level Agreements
Governance relationships
Business value tracking
…
-
8/15/2019 2 Jørgen Steines Platon
17/40© Platon
Step 1: Define the governance
level of the BI Program
17
BI Methodology
BI Policies
BI Organisation
Common Data Definitions
BI Tools & Systems
One way Ad hoc Degree of federation
BI Architectures
BI Project prioritization
Step 1: Define the governance level of the BI Program
Step 2: Identify decision making ‟bodies‟ Step 3: Define decision areas and decision rights
Step 4: Design and implement governance processes
-
8/15/2019 2 Jørgen Steines Platon
18/40
© Platon
Agenda
● What is BI● BI Governance
● BI Adoption
● BI requirement specification
● Summing up
18
-
8/15/2019 2 Jørgen Steines Platon
19/40
© Platon
What can drive better
deployment and adoption
Better BIadoption
Strategyclarification
Focus on usage
OrganisationalChange
Management
Communication,marketing and
branding
Other
19
-
8/15/2019 2 Jørgen Steines Platon
20/40
© Platon Page 20
Change Management
I can not live
without my
Excel sheets.
Let‟s buildit and they
will come.
We earn
money
anyway.
The users
The managers
The BI people
I need my own
definitions.I don‟t want my
results to be
visible for all.
We know
what they
need.
Similar to ERP implementations?
The successful companies focuses 70 % of
the implementation resources on processes,
education and other soft aspects and only 30
% on technology
-
8/15/2019 2 Jørgen Steines Platon
21/40
© Platon
● Branding…
● Provides a single identity when communicating about your BI Program
● Differentiates your „product‟ from other choices
● Create a logo
● Use it on reports, the intranet and all communications like newsletters,
status reports, presentations etc.
● Extend your brand through report certification
● A process of promoting a report to a mass audience
● Further drives the data integrity of your BI program and builds userconfidence
● Creates a adoption effect as management only wants to view reportsthat have been branded and/or certified
Communication, marketing and
branding
Page 21
-
8/15/2019 2 Jørgen Steines Platon
22/40
© Platon
Agenda
● What is BI● BI Governance
● BI Adoption
● BI requirement specification
● Summing up
22
-
8/15/2019 2 Jørgen Steines Platon
23/40
© Platon
Does this look familiar?
23
Analysis Design Development Implementation
Increasing costs to fix defects
discovered later due to
incorrect requirements
-
8/15/2019 2 Jørgen Steines Platon
24/40
© Platon
BI solution types
24
Dashboards / cockpits
Predictive analytics / data mining
Ad hoc analytics / OLAP
Reporting
Alerts and exception
GIS and other visualization
Balanced scorecard
Performance management
Analytical CRM
http://images.google.dk/imgres?imgurl=http://www.miamidolphins.com/email5.gif&imgrefurl=http://www.miamidolphins.com/dolphinstonight.asp&usg=__bZ48XzbMPTLVU40NQs5LJsMxMS4=&h=350&w=315&sz=85&hl=da&start=3&tbnid=-z2A-JbUsqleuM:&tbnh=120&tbnw=108&prev=/images?q=email&ndsp=18&hl=da&sa=Nhttp://www-01.ibm.com/software/data/cognos/images/screenshots/c8bi-analysis-screenshot1.jpghttp://www.cognos.com/products/business_intelligence/applications/Analytics_Financial.jpg
-
8/15/2019 2 Jørgen Steines Platon
25/40
© Platon
Types of (BI) requirements
● Business requirements
● Information requirements
● Functional requirements
● Detailed report / usage requirements
● Other requirements
● How about defining the business processes that apply the newinformation to managerial actions?
25
What is the business
need, pain or problem?
What business questions
do we need to answer?
What data is necessary to
answer those questions?
How do we need to use the
resulting information to
answer those questions?
All the other stuff – AKA nonfunctional requirements
Detailed layout etc
-
8/15/2019 2 Jørgen Steines Platon
26/40
© Platon
David McCandless: The beauty of data visualization
Design inspiration
http://www.ted.com/talks/david_mccandless_the_beauty_of_data_visualization.htmlhttp://localhost/var/www/apps/conversion/tmp/scratch_6/catalog/0636920000617/previewhttp://www.ted.com/talks/david_mccandless_the_beauty_of_data_visualization.htmlhttp://www.ted.com/talks/david_mccandless_the_beauty_of_data_visualization.htmlhttp://www.ted.com/talks/david_mccandless_the_beauty_of_data_visualization.html
-
8/15/2019 2 Jørgen Steines Platon
27/40
© Platon
The requirement specification
document – The simple version
27
● Introduction
● Business requirements
● Business process requirements
● Information requirements
● Functional requirements
● Detailed report / usage req.
● Other requirements
The requirement specification
-
8/15/2019 2 Jørgen Steines Platon
28/40
© Platon
The requirement specification
document – The really simple
version
28
-
8/15/2019 2 Jørgen Steines Platon
29/40
© Platon
The requirement specification
document – The expanded version
29
● Executive summary
● Introduction
● Business requirements
● Business process requirements
● Information requirements
● Functional requirements
● Detailed report / usage req.
● Security requirements
● Performance requirements
● Operational requirements● Migration requirements
● User doc. and training requirements
● Other requirements http://www.volere.co.uk/template.htm
-
8/15/2019 2 Jørgen Steines Platon
30/40
© Platon
Business process requirements
● “Change” is the keyword
● Textual description is ok
● Or use a swim lane design where the workflow or supportinginstructions, procedures or use cases are changed
30
Procedure
–Prioritize order based oncustomer rating by…
Use Case
–When the sales rep enters… The system shows…
-
8/15/2019 2 Jørgen Steines Platon
31/40
© Platon
Cover all information
requirements
● Ask, ask, ask…
● Explain and exemplify - with all stakeholders
● Facts
● Business rules
● Dimensions and hierarchies
● Value sets
● Timeliness
● How „fresh‟ should the data be (update frequency)
● Specific dates the new data is needed
● History
● How much calendar time should be covered
● How about changes in hierarchies - program requirement could betype 2 SCD and project requirement could be type 1 SCD
31
-
8/15/2019 2 Jørgen Steines Platon
32/40
© Platon
Does this look familiar?
32
Perhaps some more structuredtechniques are needed?
-
8/15/2019 2 Jørgen Steines Platon
33/40
© Platon
The process & methods
33
Identifystakeholders
Clarify methodof collecting
requirements
Plan and invitefor meetings
Prepare andsend materialor mindset at
meeting
Conduct /collect
Consolidate/ document Validate/prioritize
Update
requirementspec.
Send forreview Verify andsign off
● The sub activities for specification process is outlined in the figure
below.
-
8/15/2019 2 Jørgen Steines Platon
34/40
© Platon
The process & methods
34
Identifystakeholders
Clarify methodof collecting
requirements
Plan and invitefor meetings
Prepare andsend materialor mindset at
meeting
Conduct /collect
-
8/15/2019 2 Jørgen Steines Platon
35/40
© Platon
BI and Agile development
-
8/15/2019 2 Jørgen Steines Platon
36/40
© Platon
The effect of initial roll-out times
on project success
36
-
8/15/2019 2 Jørgen Steines Platon
37/40
© Platon
BI Requirements
- Business and IT standpoints
37
Innovation
Flexibility
Ease of use
Reliability Scalability
Accuracy
CorrectnessSpeed
User Experience
BICC ?
-
8/15/2019 2 Jørgen Steines Platon
38/40
© Platon
Pay attention to data quality
● Poor data quality is the second most common reason for BI
failure
● Data quality is a big risk
● Get a clear picture on data quality issues as early aspossible - during analysis or even before
● Don‟t wait until the development takes place
38
-
8/15/2019 2 Jørgen Steines Platon
39/40
© Platon
Agenda
● What is BI● BI Governance
● BI Adoption
● BI requirement specification
● Summing up
39
-
8/15/2019 2 Jørgen Steines Platon
40/40
Summing up
● Value comes from decisions and changed behavior – notfrom providing reports
● Information requirements are key in all BI projects
● Make the right balance between time and perfection
● Expect change and expect to manage change