Prototype Driven Requirement Elicitation for Business Intelligence September 11, 2015 Mike Krajnak Steve Strohl 1 mkrajnak@icct.com sstrohl@icct.com Agenda Introduction Why Projects Fail Why Data Governance Requirements Prototyping Success Stories 2 2 © 2015, Information Control Company About Mike Professional • • • • Business Intelligence Requirements Practice Lead CBIP, CDMP 35+ years of IT experience 20+ years building analytical solutions Little Known Facts • • • • 3 Started career as system’s programmer Scoutmaster for 7 years in the Boy Scout troop that Steve’s father started Met wife on trip to Israel Traveled to China to adopt daughter About Steve Professional • • • • Master Data Management and Data Governance Practice Lead Sr. Business Intelligence Architect 35+ years of IT experience 15+ years building analytical solutions Little Known Facts • • Started career Battelle on defense systems Spent 7 years in Alaska – Lead Architect on Exxon Valdez Oil Spill Project – Reported BI data to national news © 2015, Information Control Company Business Intelligence (BI) Overview Raw Data Fuzzy business rules Unclear data Relationships 4 Meaningful Information Reports Getting the right information . . . Decisions . . . to the right people . . . Opportunities . . . at the right time. © 2015, Information Control Company Business Intelligence Requirement Gathering Landmines o It is estimated that 85% of defects in developed software originate in the requirements* o Fixing defects is costly $ o Unrealistic scope and expectations hurt timelines Time o Multiple data sources and fuzzy business rules cause complexity o Business requirements are hard to articulate (users do not know what they want) o Lack of data governance (data quality/ Integrity) cause confusion 5 © 2015, Information Control Company (* Young, Ralph R. Effective Requirements Practices. Boston: Addison- Wesley, 2001.) Multiple Approaches to Solve this Problem Monolithic Waterfall Cons Pro 6 Pure Agile Pro Cons Long time between Complete and robust requirements and deployment requirement planning Business and IT working together Loss of connection to the big picture causes requirement drift Requirement changes have serious project impact Tangible defined handoffs between stages Flexible to handle changes Lots of rework due to requirement changes Linear dependencies cause project delays Stable Processes Clarify requirements as you go Sprints too small to create deployable artifacts © 2015, Information Control Company Combine the Best of Both into a Hybrid Approach Water Scrum Fall Monolithic Waterfall Cons Pro 7 Pure Agile Pro Cons Long time between Complete and robust requirements and deployment requirement planning Business and IT working together Loss of connection to the big picture causes requirement drift Requirement changes have serious project impact Tangible defined handoffs between stages Flexible to handle changes Lots of rework due to requirement changes Linear dependencies cause project delays Stable Processes Clarify requirements as you go Sprints too small to create deployable artifacts © 2015, Information Control Company Our Approach “Water Scrum Fall” Process Design & Build 8-12 weeks Data Governance (common thread) Planning 6 weeks Requirements & Prototype 6-8 weeks Test 1-3 weeks Release 2 weeks Total: 23-31 weeks 8 © 2015, Information Control Company Why Data Governance? According to the Data Governance Institute, Data Governance is … “The organizational bodies, rules, decision rights, and accountabilities of people and information systems as they perform information-related processes.” It refers to the operating discipline for managing data and information, including the: • People • Processes • Technology And categorizes data as a key enterprise assets 9 © 2015, Information Control Company Data as an Asset: The Problem No governing body No data policies, procedures or processes No business glossary No data quality checks No remediation process The Result 10 Badly formed or incorrect social security numbers Incorrect or obsolete addresses Incorrect dates (birth, admittance, discharge, policy etc…) No standard descriptive or type values Duplication of data across source systems Different data for the same person across source systems © 2015, Information Control Company Bad Data Good Data 11 © 2015, Information Control Company Data Governance Program View data as an asset! Data Remediation Data Quality Rules Roles and Responsibilities Governing Body 12 Policies, Procedure And Processes Our Approach– Planning • Identify the business needs, suggest projects to meet those needs • Validate that the pre-requisites can be met on each project • Estimate the costs, and rank them by return on investment • Identify stakeholders to champion the projects. • Find people to fill the roles for the BI project Planning 6 weeks o o o o o Project manager BIBA – Business Intelligence Business Analyst Data Analyst User Interface Specialist Report Specialist Deliverables: • Prioritized list of projects with estimated cost and ROI • Roadmap showing project timelines 13 © 2015, Information Control Company Our Approach– Requirements and Prototype Requirements & Prototype 6-8 weeks 14 © 2015, Information Control Company Our Approach– Requirements and Prototype SME Interview Business Questions Glossary Data Governance Business Model Working Prototype Artifacts Created Interview Guide List of Business Questions Business Terms Business Glossary Facts Qualifier Matrix Source Qualifier Matrix Data Profiling Results Logical Business Model Physical DDL Semantic layer Prototype Reports Functional Specification Business Requirement Document 15 © 2015, Information Control Company Data Stewardship Roles Contributors Facilitator 16 • Identify the people responsible for identifying the business element and data quality rules • Identify the person responsible for managing the workflow process Approvers • Identify the people responsible for approving, modifying or rejecting the data element or any part of the data element. Reviewers • Identify the people who will view the data elements but only have read authority © 2015, Information Control Company Our Approach– Term & Rules Work Flow Interviews SME’s Approvers Approvers Eval Consumer (read/write) Contributor Business Q’s Eval Reviewer (read only) Business Terms Submit Business Abstain Terms Business Business Terms Approved Business Terms Rejected Terms Business Approved Business Terms Rejected Terms Approved Publish Consumer (read only) Published Area We have now reached a Consensus! Review, Submit All business terms have been approved. Reviewer (read only) Approved terms can now be used to build the modeling objects and prototype package Facilitator 17 © 2015, Information Control Corporation © 2012, Information Control Corporation Prototype - Which would you rather have? 18 • Manual or online requirements • Paper based or accessible artifacts • Wire frames or working prototype • Which is easier to develop from? © 2015, Information Control Company Consensus Tool - saves 2 weeks of time gathering requirements – links requirements to a logical model 19 Consensus Accelerates Output DDL Table with sample data Semantic Model Documentation 20 © 2015, Information Control Company Demo the Prototype and Obtain Business Signoff 21 © 2015, Information Control Company Why Prototype Requirements? 1. More Accurate Requirements 2. Less Risk 3. More Satisfied users 4. Less Cost 22 © 2015, Information Control Company Success Stories Ten Success Factors 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 23 Cooperation between Business and IT Championing by the Business Stakeholder Narrow Defined Scope (No Boiling the Ocean) Prioritized list of Business Questions Having a PM that understands BI Mitigation Strategy for Data Integrity Issues Following Data Governance Principles Available Personnel with Correct Skills Using accelerator tools (Balsamiq, Consensus) Stand up and Status meetings Retirement 6 month project (Balanced Score Card) Production results after 2 months Continued rollout at month 4 and 6 Restaurant 4 month project Produced dashboards across 4 subject areas Colocation, management support © 2015, Information Control Company 24 © 2015, Information Control Company