Le 5 componenti essenziali di una Data Strategy

Company Conf idential – For Internal Use OnlyCopyr ight © SAS Inst i tute Inc. Al l r ights reserved.

The 5 Essential Components of a Data Strategy

Evan Levy - Vice President, Data Management Services, SAS

#DataStrategy


Most organizations have no data strategy

The Reality of Applications and DataApplications Data

IT strategy defines tools, platforms, development and approach

Rigorous development process

Standard tools & platforms

Economies-of-scale methods

Processing is a duty; data a courtesy

Data is a “from scratch” activity

Data details rely on tribal knowledge

Data standards are often point solutions

Data sharing is a production need

1339F3D


What is a Strategy?

“…a careful plan or method for achieving a particular goal usually over a long period of time…”

A plan designed to improve all of the ways you acquire, store, manage, share, and use data

What is a Data Strategy?

“…a method or plan chosen to bring about a desired future, such as achievement of a goal or solution to a problem…”

“…strategy is different from vision, mission, goals… It is the result of choices on where to play and how to win, to maximize long-term value…”

1339F3D


Data Strategy

The 5 Essential Components

1339F3D


Data Strategy

Data Strategy is not “One Size Fits All”

Host owned & Aligned

Online/bulk w/change control

Source/Appl defined

Multi-Org/App Support

Store

Integrate

Govern

Application or team defined

Business-defined

Identify

Internal/External Access

Bulk/File Oriented

Application specific

Cross-Functional Teams & Process

Provision

Organization Enterprise

• Each component is independent and can evolve as needed

• Each component is specific to an individual set of skills and capabilities

• Not all organizations need an “enterprise-class” data strategy

• Complexity (and maturity) increase with organizational scope

• The Data Strategy should establish goals for each component

1339F3D


The Core Components

IdentifyThe ability to identify data and understand its meaning regardless of its structure, origin, or location

Challenges

• Data Sources aren’t documented; existence is based on tribal knowledge

• There is no list of data elements, their names, and their meanings

• Terminology and meanings can vary across applications (and business areas)

• Groupings and hierarchy varies across business areas

• The “best source” is based on current views, opinions, and convenience

Terminology & Meaning

Application Aligned Business Defined

Grouping / Hierarchy Support

Metadata

System Adoption

Organization Scope

Application Aligned Business Defined

Static Document Actively Managed

Single Organization Enterprise

Single System All Systems

Strategy Options

133C5DD


The Core Components

ProvisionEnabling data to be packaged and made available while respecting all rules and access guidelines

• Once created, data is typically required by 10+ systems

• Sharing data is a courtesy (at the convenience of the source not the consumer)

• Source selection is based on availability without attention to accuracy

• Every data source has a unique set of methods for sharing and distribution

• No one is responsible for sharing data

Source Formatted

Consumer Packaged

Platform Access

Packaging

Tracking

Data Audience

Acceptance Checking

Bulk/Flat Extract

Online & Bulk

Internal storage Delivery Tracking

None Actively Monitored

Developers Users & Partners

Challenges Strategy Options

133C5DD


The Core Components

GovernEstablishing and communicating information policies and mechanisms to ensure effective data usage

Org Aligned Cross Functional

Data Security

Policy and Rule Definition

Data Quality

Mechanisms

Data Management Oversight

Application Specific Business Defined

Application-centric Stakeholder Defined

None Centrally Audited

Application-centric Process Defined

..


• DG often focuses on metadata not policy or process definition

• Committees forget their focus is decision and delegation (they become a working team)

• Data is assumed to be accurate; there is no discussion about acceptance

• Mature governance should reduce committee activities not increase them

• Measuring policy/rule adoption is critical – and often ignored

133C5DD


Company:

Consumer Packaged Goods Key Facts:

Consumer Packaged Goods

• $14 Billion in revenue

• Cookies, crackers, breakfast, frozen foods

• 33,000 employees

• Manufactures: 20 countries

• Sells: 180 countries

.

Business Details • Product details contained within SAP and several

other (internal and external) systems

• Product information is created (and updated) by different business teams

• Business terminology is inconsistent; data content is volatile and frequently inaccurate

Data Strategy Details • Data correction became collaborative, business

owned activity (educated staff on DQ tool usage)

• Began work on business data dictionary

• Established product data profiling and measurement

• Rescoped data governance process


The Core Components

Store Persisting data in a structure and location that supports access and processing across the enterprise

• Everyone has their own copy of data

• There’s no onboarding process or means to identify available data

• Data is structured based on the source, instead of access and usage

• Manipulating data over a time is difficult (due to changes)

• A “one size fits all” approach isn’t appropriate for needs (online, BI, etc.)

Application -based

Raw, Cooked, & Archived

Stored Content

Onboarding Content

Source Data Catalog

Navigation / Access Methods

Unmanaged Versioned

Source Location Req’d

Location Unnecessary

None Documented / Versioned


Change Control

Application Specific Centrally Managed

133C5DD


The Core Components

IntegrateMoving and combining data residing in multiple locations and providing a unified view of the data

Developer Owned

Central Rules

Identification / Matching

Data Correction

System Adoption

Rule Identification

None Governance aligned

Source Specific Centrally Managed

Analytics Systems

Txn & Analytic Systems

.


• Every project team (or user) integrates content based on their own logic

• Even with MDM folks still build their own match logic

• The DW team has ETL standards; other teams have their own standards

• Recording business rules in ETL metadata doesn’t help “self-service” data users

• There no centralized mechanism for tracking where the data has gone

Master/Reference Data

Bulk/Static Multi-Subjects Actively Managed

133C5DD


Company:

Multi-Platform Media Key Facts:

Media Company

• $3 Billion in revenue

• 3500 employees

• 13 media brands/properties

• 30 languages

Data on 360MM consumers

Business Details • Revenue generated from advertising, carriage fees,

and subscriber fees

• Hundreds of content resellers and partners around the world (broadcast tv, social media, online)

• Consumer trend and preference details stored in multiple systems (some unknown)

• Cyber concerns is a corporate “top-of-mind” issue

Data Strategy Details • Establish source data catalog and a centralized data

provisioning team

• Begin data quality monitoring and reporting process

• Expand data governance to address external data access (partner support) issues


Moving Forward…

Identify the scope of your data strategy initiative• Identify the organizations that own/influence the various component

areas

• Start small to limit initial effort; start large enough to ensure visibility

Review your existing strengths and weaknesses• Consider expertise within each individual component area

• Focus on “good enough”, not perfection

Consider future development efforts and ongoing obligations• Look at major application changes and upgrades

• Start with data sharing, access, and movement issues; use the components to categorize needs

1339F3D

sas.com


Evan Levy - VP, Data Management Services, SAS

Thank You!

[email protected]

@evanjlevy

@EvanJayLevy

http://www.sas.com/

http://www.sas.com/

http://www.sas.com/

Le 5 componenti essenziali di una Data Strategy

Data & Analytics

Transcript of Le 5 componenti essenziali di una Data Strategy