Friday, June 1, 2012

What is Data Governance?

Data governance is the the confluence of the physical organization, conceptual framework and operational practices that enable the creation, distribution and maintenance of high quality data.

We'll break down each of the major considerations of data governance for further discussion below:

  • Physical organization:
    • This is the topic that comes most often to mind when the term 'data governance' is mentioned.  Executives and consultants alike tend to address this aspect of data governance ahead of the other considerations.  Perhaps it is due to an intuitive grasp of the need to build an organization around the discipline.  On the other hand, perhaps it represents an opportunity to further consolidate power and extend a fiefdom - I'm not judging.  
    • In either case, the canonical organization should break down into a tiered structure that represents the following (top-down organization)
      • an Executive Steering Committee of key decision makers (from Business and IT) who can set strategic vision, challenge key decisions and adjudicate differences
      • a Senior Executive (e.g. Chief Data Officer) with ownership of the program
      • Data Stewards (Business) who will represent each of the key business areas/subject areas and serve as the advocate, subject matter expert and point of contact for governance decisions and ongoing processes
      • Data Admins (IT) who will effect the decisions and initiatives of the governance process;  there are typically multiple Data Admins who will support the needs of each key business/subject area
    • Aside from the the typical challenges related to ever-shifting priorities, executive commitment and political jockeying, establishing the governance organization is a relatively straight-forward process.  The key lies in getting senior buy-in and support of the initiative.  Without this support and mandate, the data governance effort is unlikely to succeed.
  • Conceptual Framework
    • Common gaps that are often found in nascent data governance programs are as follows: 
      • the lack of a clear conceptual framework on which to identify the scope/boundaries of governance
      • the need to formalize the related concepts/domains
      • the establishment of key metrics as well as baseline for measurement (I'm emphasizing this last point as it is often neglected and given short-shrift but is a cornerstone to the conceptual framework as well as to the governance program as a whole - you cannot govern what you can't measure!)
  • Operational Practices
    • TBD...

Thursday, May 31, 2012

JPM Trading Losses Related to Pricing Discrepancy - A Data Governance Issue?

Video Segment:
http://www.bloomberg.com/video/93700605-new-twist-in-jpmorgan-loss-spotlights-pricing.html

Issue:
  • Utilization of a different set of securities prices due to disparate Price Source of Record allowed the Office of the CIO at JPM to mark its books differently than the rest of the firm
  • This difference in standards potentially allowed the trading desks at the Office of the CIO to shield itself from the firm's risk controls thus leading to the outsized losses (currently unrealized)
Governance Perspective:
  • JPM seems to be in need of a single data governance organization with its associated controls and business owner (in the person of a Chief Data Officer) who would report directly to a C-Level executive
  • Having this structure would be the basis of implementing 
    • foundational data management infrastructure and practices
    • unified governance structure with 
      • key metrics & tolerances
      • common language
      • unified standards
      • consistent controls
Hypothesis:
  • Had the firm been standardized around a common Source of Record on Securities Prices, it would have been able to mark the Trading Book consistently with other Trading Books across the firm
  • The Risk Organization could then compare and contrast the Value at Risk (VaR) between the Trading Books within the Office of the CIO and the Risk standards/control tolerances thus potentially highlighting the dangers earlier and minimizing the size of losses

Wednesday, May 30, 2012

Alternatives to the Chief Data Officer - A Sales and Prospecting Approach for Data Management Consultants


  • Hypothesis
    • The emergent industry perspective is that, given the increasing realization by C-level executives that data should be treated as a business asset, Chief Data Officer (CDO) roles are being formalized and created
    • Vendors working in the data management space are eager to engage with CDO's at these organizations as key decision makers for the sales process
    • Is there an alternative path to a key decision maker that may be more relevant and hold greater sway than the CDO?
      • I would submit that Chief Risk Officers and the General Counsel may be equally valid if not more viable key decision makers for vendors undergoing the sales process
      • This hypothesis is based on the combination of 
        • Business need
          • New and existing regulation
            • Dodd-Frank
            • Basel II/III
            • European regulatory reporting (COREP)
        • Funding
          • Firms have allocated capital towards complying with these regulations
        • Well-defined domain/scope
          • Scope of compliance is relatively well-formed vs. a typical reference data/master data management effort
  • Approach
    • Data Management Consulting Vendors should...
      • Align service offerings to key regulatory imperatives
      • Clearly define scope and outcomes of service offerings based on analysis of regulation
      • Take note of timing requirements and factor them into proposed initiatives
  • For further research
    • Apply sales themes based on likely impacted industries and companies
      • Regulatory compliance and risk mitigation
        • Financial Services
      • Cost reduction & efficiency
        • Manufacturing
        • Retail
      • Profit generation
        • Social media?

A Case for 'Forensic Data Profiling' in M&A Due Diligence


  • Case
    • M&A Due Diligence
      • Goal:  Company A acquiring Company B to create an organization with substantially larger market share
      • Reality:  The merged Company AB had substantially smaller than forecasted market share due to overlapping customer base and product set - resulting in a substitutionary rather than a complementary outcome
      • Hypothesis:  Company A could have avoided this outcome if they had engaged in 'Forensic Data Profiling' practices to validate their investment thesis
    • Opportunity
      • Information Management firms should leverage opportunistic & context driven initiatives to offer variations of their services rather than engage in programs/initiatives within a data domain context
      • Doing so mitigates the need for creating a business case since a compelling case already exists and is funded
    • What does this 'Forensic Data Profiling' service offering entail?
      • Build of a due diligence data warehouse (DW) to answer the following questions
        • Clients
          • Who are the customers of Company A and Company B?
          • What are overlaps in customers?
        • Products
          • Who are the products that Company A and Company B offer?
          • What are overlaps in products?
        • Predictive models
          • Which customers are likely to defect from the merged entity?
            • Of these potential defectors, which are high quality customers that should be induced to stay via special outreach/incentives?
            • Reverse of above point
          • What product sets should be kept and which should be retired from the lineup of the merged entity?
      • How does this differ from a typical data profiling exercise?
        • Not focused purely on Data Quality (DQ) though DQ issues encountered will be remediated if it impacts the core objectives and highlighted but deferred should it be a secondary issue
        • Not focused on architectural implications as primary objective
          • Data sources and targets will be identified over the course of the DW design/build
          • Architectural rationalization not the main focus

Building a Business Case for Data Governance

Building a business case for data governance

  • What are common drivers?
    • Risk mitigation
      • Minimize processing errors stemming from DQ issues
      • Utilize data that's fit for purpose vs. misusing data based on incomplete/inaccurate understanding
    • Efficiency
      • Increase STP and minimize exception handling resulting from DQ issues
      • Reduction in complex data mapping and architectural rationalization exercises due to the fact that data is already well structured, stored in domain/subject specific sources, high quality and is accompanied by good metadata
    • Operational agility
      • Reduced time-to-market for new initiatives due to the fact that data is high quality, accessible and architected in a way that is 'ready for use'
    • Revenue generation?
      • Data is structured in a manner that allows for ease of analysis
  • What are common hurdles to adopting a formal data governance program?
    • Education
      • Misunderstanding of what a data governance effort entails
        • Commonly thought of as the following issues...
          • an organizational issue - appoint people to governance roles and the problem is solved
          • a technical issue - technology owns this space and should be responsible for its outcome
          • a manual audit control - 'throw' people and resources at the problem and quality can be maintained at the desired level
    • Cost
      • Unknown cost beyond personnel
        • Program initiatives
        • Software
        • Hardware
        • Process changes
      • Common questions
        • How 'big' should a data governance program be?
        • When are you done?
        • Who pays for it?
    • Ownership
      • Ownership models
        • Business
        • IT
        • Joint
        • Operations
      • Accountability
        • Senior sponsorship required
        • Business ownership suggested
        • Owners should be accountable for measurable outcomes
    • Alignment to strategy/initiatives
      • Commonly align to reference data/master data management (MDM) initiatives
      • Questions:
        • Can a data governance program be instituted independently of other data initiatives?
        • When should a data governance program be created?
        • What are the boundaries of a data governance program?
          • Enterprise-wide?
          • Business unit specific focus?
          • Initiative-related (e.g. creation of a new reference data/master data repository; response to regulation - such as Basel II)?
        • Can a data governance program succeed if its scope is
          • too narrow?
          • too wide?

Tuesday, May 29, 2012

Research Material

Books

Magazine articles


Industry publications

Blogs

Idea Generation

Why Data Governance?
  • Inseparable from data initiatives
  • Required for long-term usefulness of data within governance domain
  • Indicator of health of organization (operating risk)?
  • Demonstrates organization's perception of data as an asset
  • Minimizes downstream issues related to data quality
  • Mechanism for correcting data issues
Which Perspective?
  • Practitioner's perspective
    • Implementation viewpoint
    • Policy Maker's viewpoint
  • Focus on Financial Services
    • Capital Markets
      • iBanks
      • Asset Managers
  • Balanced blend of scenarios & lessons learned vs. theory
What are the gaps?
  • Current 'industry'...
    • emphasis on theory in available literature
    • bias towards organizational governance with minimal focus on metrics, implementation and operations/maintenance
    • lack of clarity on value of data governance programs from the following perspectives
      • success of a Reference Data/Master Data Management  (MDM) program
      • minimization of operational risk
      • compliance with existing regulations
      • operational efficiency
      • organizational agility

What are the Objectives?
  • Education (self & audience)
    • Research & synthesis of available content (print, web, industry publications...etc.)
    • Creation of educational and commercial material
  • Capture industry/practitioner's perspective
  • Integrate practice with theory

Data Governance Idea Generation:
  • Governance dimensions
  • Measurement
  • Sponsorship
  • Alignment to strategy
  • Impact on data quality
  • Governance organization - theory vs. practice
  • Governance operations