Business Intelligence

Informasjonsledelse - "Closed loop"

Informasjonsledelse inkluderer hele verdikjeden fra kildesystemer via datafangst datavarehuslast, OLAP-lag, sluttbruker-/presentasjonsverktøy, håndtering av styringsinformasjon i ledergrupper, initiering av endringsprosjekter og kvalitetssikring av datavarehusprrosessen som "closed loop". Dette omfatter både strukturerte (numeriske) & ustrukturete (tekst, multimedia etc) data. Informasjonsledelse dekker også styringsmodeller for økonomi- og virksomhetsstyring.

Business Intelligence Architecture

Business Intelligence can be seen as a separate set of solutions apart from the transactions systems in a corporation. As opposed to the transaction systems Business Intelligence is focused on extracting information for 4 purposes:

1. Scorecard

2. Reporting

3. Analytics

4. Triggers

 

The 3 top layers (of the information pyramid) covers information for use by humans. The 4th level delivers information from the Business Intelligence solution directly to a target system (may be a CRM system og a transaction system) feeding it with "actions".

The architectural aspects of a Business Intelligence setting have several levels:

1. Business Architect

2. Enterprise Architect

3. IT Architect

Depending on you belonging to the business side or the IT side of the business the archtects role is to look at needs and processes and how these may interplay through enabling technologies. Aspects important to Business Intelligence Archtecture are:

SOA (Service Based Archtecture)

MDM (Master Data Management)

Key Performance Indicator Database

Many businesses fail to have 1 overview of all Performance Indicators. In doing so they loose the total picture and development is by conicidence more than a structured and targeted process.

One way to ensure a total overview is to store all indicators in one single database complete with source information, definitions, responsibility, users, and our experiences using each indicator. Also, it is useful to have a tag to indicate if a KPI is active or if it is just in our reserve pool.

Business Intelligence Process

In implementing a solution for Business Intelligence it is important to use a formal approach or methodolody to describe the neccesary process steps. There are many ways to do this. However, some common denominators should be present:

1. Measuring the executive information satisfaction though a survey

2. Analysing the infra structure and existing systems

3. Mapping KPI's, reports, analysis solutions and 1:1 triggers

4. Making a prototype for pilot testing

5. Lauching reports as they become available

6. Describing how the reports are to be used (e.g. which meetings and by whom)

7. ++

In this example we do not cover a management model (e.g. scorecard etc). This will complicate some, however, it is important to integrate the management model 100% in this work.

Mapping database

It is common to start Bi and data ware housing project without mapping thoroughly the total of reports in an organisation. This leads to a random solution covering only parts of the total need for reporting.

The key is to map all reports in terms of:

1. Source

2. Name and ID

3. Descrition

4. Owner of the report

5. Lead times and frequency of use

6. etc

If you do this, then you will get a total picture of the needs and how these are solved. You will also find many reports that are produced that users no longer need. This "clean up" gives you a solid fundament for an effective development of reports and a means to prioritise the most important reports first.

On several occasions I have made such databases for clients. The database gives a good overview and is vital to have a rational approach towards reporting.

The same database princliple can be used for Key Performance Indicators, reports, analysis solutions, and even 1:1 triggers.

Information Management

Information can be divided in 2 halves and several levels. Imagine a triangle where the left half concerns numerics. This half has 4 levels:

1. Scorecard

2. Reporting

3. Analytics

4. Right time 1:1

The 4 levels increases detail from "the vital few" down through more deailed reporting, analytics of large amounts of data down to real time usage of 1:1 data.

Likewise the right half of the triangle covers unstructured data in 4 levels:

1. Scorecard (How KPI's are improved)

2. Reporting (Meta data)

3. Analytics (Patterns)

4. Right time 1:1 (Indexing)

The same increase in details are iminent here as well. But we focus on two types of data. Firstly we cover meta data i.e. data about unstructured information and then the unstructured data itself. Sometimes the data is divided into internal (ECM) and external (e.g. web) data as well.

Performance Measurement

Most corporations are good at measuring the financial aspects of the business. And most corporations use traditional budgets and maybe some even use financial ratios. Do you need more than that?

Well, if you want to be pro-active and hollistic this just is not enough. Some measurement models argue you have to measure the market in terms of sales, customer satisfaction and customer retention.

Analytics & Business Intelligence

Analytics is used as a term in many areas. However, in terms of Business Intelligence we can subdivide the term to classify different needs on the business side and solutions on the IT-side. One way to do this is to split it into 3 areas:

1. Historically

2. Current

3. Future

Historical information may be used to analyse patterns and gather experience on past operations. Current information may be used to analyse our operations at this moment and do swift changes. Future information adds a bit of guesswork and may be based on buth historical and current information. Typical future analytics include forecasting, prediction and so on.

Corporate & Enterprise Reporting

Within a corporation there are vaste and differing reporting needs. Much of such reports take time and effort to be produced, takes time to distribute and often follow no common priciples and standards. The result is that reporting in many instances varies in quality, the correctness vary, there is a need for manual adjustments, and as people move on within or outside the corporation the reports may still be produced even if the users are long gone.

Now, to make sure reporting is current, up to date, has good quality and relevance it is important that users reporting needs are mapped. Then as the reports are being set up there is an updated database showing current reports, users, source systems and so on. This way all reports may be evaluated in 1 place. The total reporting need should then be seen as a part of Business Intelligence and enabled through a data warehouse if the corporations sizewise and numerically allows for it.

Scorecards & Management models

The first scorecards we saw in business use contained no business model. They were merely a collection of indicators and in some instances just financial ratios. "Tableu de bord" was a french invention (Invention is perhaps widening the definition a bit - several other sources of business scorecarding exist). Then in the late 1980's some work was done to see if it was possible to identify indicators that said something about future performance.

2 types of indicators were used and linked in what whas later to be coined Balanced Scorecard. The Balanced Scorecard was a challenge to large for most companies to combine with operations. So operational balanced scorecards being actively used was to become the exception from the rule. The rule became using a lot of resources to create unclear and foggy cost/benefit related to the Balanced Scorecard. I personally used more that 15000 hours research, training & consulting so find pitfalls, the rights steps and in what order they should be taken.  I found several reccomendations I have chosen not to disclose openly. Some consulting customers have experienced the results in Performance Measurement projects with substantial cost/benefit.  

Most of this work is resting in peace as I continue to work and remains a reference source for my further work. I will take up some of these factors in this blog as time passes by.