A different approach to Business Intelligence

Everyone knows the problem

  • Information hidden within a mass of transactional data

  • Inconsistent information causes unhappiness

  • Labour-intensive environment creates a large backlog of requirements

  • Users look elsewhere for answers

  • Credibility hole


Everyone knows how to fix it

  • Provide a single, consistent source of the truth

  • Do it quickly

  • Hit the mark


Few do – why?

  • Too much noise

  • Hard to hear the real question

  • Hard to formulate a punchy answer – it all sounds trivial

  • Planning gets disrupted constantly

  • System extracts are complex and data is poor

  • Bottom up approach


Overall Approach – Saving it

  • Find a friendly senior manager to work with

  • Understand his real problems and what information goes with them

  • Find out what you have and agree on what is possible

  • Do it with a dedicated resource

  • Model his piece of the data

  • Deliver in small chunks

  • Find the next one


In parallel

  • Assess BI maturity

  • Map out Key Performance Areas

  • Draft Conceptual Model

  • Identify burning needs/quick wins

  • Group and prioritise 

  • Estimate effort and time

  • Estimate benefit

  • Buy-in and budget


Maturity Assessment

BI Maturity Model:

Maturity Model.jpg


Simple Information Model:

Simple Information Model.jpg

Objective:
Deliver one version of the truth to business

  • Actions.

  • Data quality must be addressed at source.

  • ETL level data cleansing creates more problems than it solves.

  • Business rules for derivation must be formally defined.

  • Data management must be a formal discipline.

  • Conflicting or multiple sourcing must be resolved and agreed to by business prior to commencement of ETL development.

  • A single ETL will exist for any atomic-level data item.

  • Any interpretation of the business rule will be contained in the ETL.

  • No changes to atomic-level data in the repository will be allowed by any other agency.

  • Embed a simple version of this “one version” as a key component of upper-level KPI reporting.

  • The cascade effect is indispensable for adoption by the organisation.


Objective:
Make information accessible, thereby liberating business from limited availability of technologists

  • Actions

  • Provide a well-structured information repository

  • Publish all business rule definitions

  • Publish all ETL timings (eg pre- and post- monthend, weekly, etc) with appropriate warnings.

  • Provide a front-end tool with a friendly user interface, and ability to navigate cube structures.

  • Provide metadata information in an unambiguous and “instinctive” way

  • Provide context-based, example-based training on demand to users

  • Manage change to source systems formally so that BI definitions are not invalidated

  • Initiate early adoption in a trial business area for the “information steward” role.


Objective:
Support a broad spectrum of business decisions, with increasing impact on good decision-making and process efficiency

  • Actions

  • Collect all data items from particular tables when these are first visited to develop new ETLs.

  • Build an understanding of the evolution of business KPI’s by close alignment with business decision-makers

  • Provide for inclusion of externally-sourced data in the repository

  • Risks must be explicitly recognised and managed

  • Business Activity Monitoring to provide process execution statistics to BI.


If you can do all this own, you’re superman!