Adopting BI in an

Organization

Using Proof-of-

Concept

TechniquesGroup 1

AB

OU

T T

HE A

UTH

OR

S

Mark Kromer is a senior product manager at Microsoft specializing in database applications and business intelligence. He manages product development for Microsoft Services BI solutions within the Industry Solutions Group

Daniel Yu joined Microsoft in 2006 as a product manager and currently manages business development for Microsoft Enterprise Cube BI solutions within the Industry Solutions Group

THE PREMISEData collection requires enormous efforts that take considerable time and specialists from many groups, including database administrators, developers, business analysts, and data warehouse expertsHowever, the business units responsible for driving business performance and lines of business, in most instances , fail to realize the value of this dataThough the technology and software that make business intelligence possible have existed for decades, the enterprise wide adoption (and benefit) of business intelligence has yet to materializeThe problem is compounded when different departments make conflicting financial or marketing decisions that may be confusing to customers

Communication

A combination of poor communication between IT and the business, the failure to ask the right questions or to think about the real needs of the business

BI viewed as an IT issue

IT departments make the mistake of looking at BI as

an engineering problem that requires a specific package

solution

ROI It's better to spread that investment over a broader number of users, raising the ROI for each user. Focusing only on analytical users is expensive and wasteful.

Between 70% to 80% of corporate business intelligence projects fail, according to research by analyst firm Gartner

*Source:www.computerweekly.comwww.zdnet.com

THE BASICSO

LTP

The databases store information about business transactions, plus other data such as employee recordsThese databases execute transactions, meaning that they add, update, or delete groups of records at the same timeData extraction difficult because:-OLTP databases contain a large number of tables-OLTP databases constantly update their data-OLTP databases usually store individual records

OLA

P

IT departments usually keep OLAP databases isolated from OLTP databases.OLAP databases use fewer tables and a different type of schema. In addition, they usually keep the number of joins to a minimum by arranging tables a star schemaThe joins between the dimension and fact tables allow you to browse through the facts across any number of dimensions, as well as up and down any number of hierarchiesOLAP databases make heavy use of indexes because they help find records in less time

CUBE

S

A cube aggregates the facts in each level of each dimension in a given OLAP schemaBecause the cube contains all of your data in an aggregated form, it seems to know the answers in advanceSpeed: the largest cube in the world is currently 1.4 terabytes and its average response time to any query is 1.2 secondsUsers can view cube data with any valid tool, including spreadsheets, Web pages, the Cube Browser in Analysis Services 2000, or graphic data browsers such as Microsoft Data Analyzer

ETL: There needs to be a way to move the data to the OLAP database, combine that data into useful aggregations, and then populate the tables. That process is often called Extract, Transform, and Load (ETL)

SQL Server has a built-in utility called Data Transformation Services (DTS) that performs the ETL tasks.

*Source:http://msdn.microsoft.com

*Source:wikipedia

A proof of concept (POC) is a demonstration whose purpose is to verify that certain concepts or theories have the potential for real-world application. POC is therefore a prototype that is designed to determine feasibility, but does not represent deliverablesIn the field of business development and sales, a vendor may allow a prospect customer to trial the product. This use of proof of concept helps establish viability, technical issues, and overall direction, as well as providing feedback for budgeting and other forms of internal decision making processes*

METHODOLOGY

A pre-implementation phase of such projects is scoped for gathering sample data from data sources required for a customer segmentation solution

Proof of concept: A small scale data mart and analysis cube is created based on this data

This phase leads to data gathering, data modeling, and producing the results through an OLAP engine into an Excel spreadsheet

The resulting spreadsheet serves as the catalyst to present the business value to the project sponsors and executives

CASE: CUSTOMER SEGMENTATION PROJECTInitial phase• Identified the needed resources and key stakeholders • Spent four weeks gathering and analyzing their campaign data and purchase logs

Find

ings

of p

roof

of c

once

pt te

chni

que The top 1 percent of base subscribers

accounted for 43 percent of all revenue. By implementing a segmentation business intelligence solution for their marketing organization, the company could achieve a return on its marketing investment of 110 percent

Revenue champions: Distribution curve of customer revenue—when segmented properly—is not a normal one, but rather very skewed toward a few selected segments.

The purchasing patterns of these top-line customers were different from regular users: they bought products and services at different times during the day and were influential to the overall product’s success among regular users

Comparative study: Without the revenue champions, company would have certainly had a low ROI

CASE: CUSTOMER SEGMENTATION PROJECTSegmented Marketing • Sent the same product offering to two sample groups of 20000 customers each• Observation for 2 weeks• After setting specific goals for the campaign, we separated the expected contribution from the overall marketing

campaign• Used standard data mining techniques, and ran hundreds of Monte Carlo simulations and simulation optimizations

and produced a set of actionable marketing activities that would help the service provider target the right customer at the right time through the most effective channel

BI suites used:

CASE: CUSTOMER PROFITABILITY PROJECTCo

ntex

tBecause of the company’s strong base subscriber growth, the enterprise did not put a priority on discovering insights into which customer accounts were most profitable, or which accounts were too expensive to maintain based on service plans and handsets used by subscribers.

Lack of planning resulted in lost revenue because some customers had more operating costs than revenue due to factors including roaming and old handsets.

Met

hodo

logy

After extracting the data from the existing data warehouse, they were able to cleanse and format the data into a small data mart and single cube which was used for processing aggregations, dimensional hierarchies, and data summaries of customer profitability metrics including average revenue per user (ARPU), average revenue by handset, revenue by geography etc

Utilized the capabilities of the BI presentation software to export the results of the trial directly from SharePoint graphs in ProClarity to PowerPoint presentations

Thank You !!