governed self-service bi

13
Governed Self-service Analytics Presenter: Frank Silva 1

Upload: frank-silva

Post on 13-Apr-2017

262 views

Category:

Documents


8 download

TRANSCRIPT

Page 1: Governed Self-service BI

1

Governed Self-service Analytics

Presenter: Frank Silva

Page 2: Governed Self-service BI

2

Predictive and prescriptive analytics, search, embedded analytics, collaboration, self-service data preparation, big data, data lakes, search-based and visual-based data

discovery, data Visualization, predictive modeling, data mining, statistical modeling, business intelligence, data warehousing, smart data preparation, reporting, dash

boarding, storyboarding, threaded discussions, annotations, automated pattern detection, embedded advanced analytics, search-based natural-language query generation

Page 3: Governed Self-service BI

3

What is Self-service Analytics?

Page 4: Governed Self-service BI

4

Centralized, IT-centricCentralized, economies of scale, governance, standards, best

practices, consistent data, enterprise-wide, certified data, performance

De-centralized, Business-centricShorter time to insight (speed), flexibility, freedom, Local

needs

Page 5: Governed Self-service BI

5

Centralized, IT-centricCentralized, economies of scale, governance, standards, best

practices, consistent data, enterprise-wide, certified data, performance

De-centralized, Business-centricShorter time to insight (speed), flexibility, freedom, Local needs

(A blended approach)Develop organizational, architectural, and

technological framework that combines these two models in to a coherent whole.

Page 6: Governed Self-service BI

6

Self-service Analytics – Does it work well?• Good approach as far as business ownership, high demand, shorter time to insight (speed), flexibility, freedom,

control, and local needs are concerned.

• Although tools have become easier to use, it is still not easy to create a self-service environment.

• Self service can backfire if users fine the tools too complex. Most users settle down with just basic functionality of the tools.

• Conversely, too little functionality creates the opposite backlash – users find tools too limiting and stop using them.

• Self-service analytics requires a lot of hand-holding. Not all power users are skilled enough to perform data blending, modeling and perform data validation.

• Many users don’t have time, patience, or skill to develop reports, dashboards and stories, create metrics, dimensions, hierarchies, engaged in threaded discussions.

• Many require one-on-one training and more importantly time to master BI tools.

• Inability to develop ‘certified data’ – data that has been profiled, cleansed, transformed, and optimized for performance.

• Failure in imposing governance (data, process and tools) and best practices across organization.

• Cross-functional or enterprise reporting is impossible (conformed dimensions and facts, drill-across, organizational KPIs)

• Lack of central Administration (Licensing, scaling, installations, security, support, training)

Page 7: Governed Self-service BI

7

Comprehensive strategyto

develop organizational, architectural, and technological framework that combines these

two models in to a coherent whole.

A blended approach

Page 8: Governed Self-service BI

8

• Make data warehouse solutions fast to deploy and easy to manage through agile methods.

• Use incremental agile approach for building EDW

• Use best practices in data warehousing - star schema, data profiling, cleansing, transformations, blending, loading, optimizing for performance

• Global data models, conformed dimensions and facts certified for reporting.

• Build a data dictionary

• Provide support for users

• Continues improvements

• Extend functionality of BI components (SDK, APIs, mashup) • Two-way communication with users

Centralized/IT Approach

Page 9: Governed Self-service BI

9

• Business Analysts or power users (carefully select the right candidates – capable for data preparation as well as developing analytical components for business).

• Extend global models to support unique and localized requirements.• Edit existing global models and augment or blend with new data from local files or

remote source.

• Use BI tool’s build-in data preparation features to profile, format and model• Develop reports, dashboards, storyboards and engage in threaded discussions

• Work with extract mode for sources that are not in global model.• Minimize working with uncertified sources.• Casual users consume the reports and also share views with others.

• Casual users can become analysts and power users. They can rely on power users to help them make the transition.

De-centralized/Business Approach

Page 10: Governed Self-service BI

10

Centralized/ITEnterprise Data Warehouse (EDW), global models, certified data,

enterprise needs

De-centralized/BusinessShorter time to insight (speed), flexibility, freedom, local

needs

Data warehousing best practices.Global data models, conformed dimensions and facts. Metadata.

Incorporate the new sources and data into the global model making the capabilities universally available.

Page 11: Governed Self-service BI

11

• Strategy formulation, business alignment and adoption strategy. Effective strategy should ensure that enterprise objectives, business strategy, investments, and analytics strategy are aligned.

• Analytics program management – technology, tools, processes and people. Decide on what is best for your organization. Think Big, Act Small!!!

• Governance, data stewardship, standards, best practices, security architecture, project methodology. Stick to basics.

• Best practices and process for incorporate the new sources and data into the global model making the capabilities universally available.

• Be collaborative - User forums, discussions, lunch and learn sessions, Analytics portal. Listen to voice of customer.

• Coaching and training of business users on effective use of self-service analytics tools.

• Benchmark your analytics environment for continues improvement.

Blend the two extremes withAnalytics Center of Excellence

Page 12: Governed Self-service BI

12

Use of Hadoop in a Blended Analytics Model

Staging

Global ModelCertified Data

Incorporate the high value data into the global model making the capabilities universally available.

VolumeVelocityVarietyVeracity

LandingExploration

Data in original form

Page 13: Governed Self-service BI

13