What is Business Intelligence?

Business Intelligence (BI) is an umbrella term which pulls together a whole raft of processes and technologies, as can be seen from the various elements on this web site.  BI really needs to be separated into two key process types.  Firstly you need to decide what to measure; and secondly put in place the technologies to collect and use the measurements.  Inevitably this is an iterative process and is described by our BI Cycle. 

Particularly when looking at corporate strategy and business alignment some elements of BI can be packaged into approaches such as Corporate Performance Management or Business Performance Management. 

This takes a set of metrics and simplifies and accelerates the creation of budgets and plans and the comparison of these with actual; often known as the Plan - Monitor - Analyse cycle.  By cascading this approach throughout an organisation it becomes clearer as to which parts of an organisation are performing and which aren’t.  This is typically reflected in Balanced Scorecard, although there are other approaches.

The design phase, where decisions are made about what to measure, will include thinking about Key Performance Indicators, Balanced Scorecards and other Metrics.  Other decisions will be what areas of the business to include and in what order. 

BI is applicable to Operations, Quality, Supply Chain and Logistics, Purchasing, HR as much as to more obvious areas such as Finance and Sales.  However no project can do everything at once so prioritisation is important.

The design phase will also need to include some thoughts around presentation of information – do people prefer graphics and pictures or numbers.  Would a specialist dashboard be appropriate or will lists of numbers in Excel suffice?  Should we use a balanced scorecard or a different methodology?

The process to collect and use the information at the highest level consists of three basic elements:

• Collecting and cleaning relevant data
• Storing the data
• Methods for looking at and exploring the data

These various steps can be performed with some specialist software or none, and different software tools may have different terminologies or combine the steps in different ways.

Collecting and Cleaning Data

The manipulation step is a key element and will deal with different types of data issues:

• Different systems naming fields differently (eg Customer vs Client)
• Different systems naming data differently (eg Altimus vs Altimus Ltd)
• Standardisation of descriptive fields (eg UK vs United Kingdom for Country)
• Standardisation of Groupings (some systems may have two product group levels, others may have 3)
• Introduction of new analysis items (eg system holds Country, but analysis required by Region)
• Validating data (eg comparing sales from the invoicing system with sales from the financial system) and reconciling and eliminating errors. And so on

Storing Data

Data is generally divided into dimensions and measures.  Dimensions are what the data is going to be analysed by, and facts which are the actual numbers.  Dimensions would typically include such things as customers, products, GL Accounts, Country, Site, Time; whereas facts would be items such as Sales Value, Quantity, Budget Value, throughput or any number of others from any area of the business.

Once stored data can be further processed into ‘Cubes’, a cube performs some specialist pre-calculations so that when, for example, a user requests sales for a customer for the whole year the cube has already added up the total, so it doesn’t have to be calculated when the request is sent.  This further speeds up the return of information to the user.

This traditional approach is the one used by Microsoft and most of the other traditional BI software companies like Cognos.  However there is a different approach which combines the data warehouse and cube into an in-memory representation of the data, often referred to as a data cloud.  This approach is utilised by tools such as QlikView.

Neither approach is intrinsically better than the other, although they do have strengths and weaknesses.  The choice of model will depend on the specific circumstances of the Company and what the BI program objectives are.

Using The Data

Dashboards are designed, like an instrument dashboard, to give a quick view of the health of an organisation by graphically displaying a small subset of the organisation’s metrics i.e. the Key Performance Indicators (KPIs).  A scorecard is likely to contain more metrics, but be updated less frequently.

Analytics refers to a whole class of systems for looking at the data in different ways.  This can be graphical, numerical or use a whole variety of specialist tools.  A very good  example of these is ProClarity’s decomposition tree which quickly and simply allows a user to see how various data items are made up. 

For example which customers contribute the most to sales, or which products are most profitable.  The key difference between dashboards/scorecards and analytics is that the former pretty much define the data that is going to be viewed, and in what format, whereas analytics allows the user to explore the data in any format and in any combination they feel appropriate.

Data mining is another different approach.  Here specialist software is used to automatically search for patterns in the data.  Analytics support the question from the user of ‘I wonder if X is linked to Y’, because they suspect something from their knowledge of the business.  On the other hand data mining will say that ‘A is related to B’ and leave it to the user to make the relevant business connections (if any) and to act upon the connections.

Getting the BI tools and techniques, metrics and processes right is a challenge, but even more challenging is acting upon the results that BI gives you to make better decisions, faster and drive your business forward.

To find out more about Business Intelligence and how it can benefit you, call Altimus and speak to one of our experts on 0800 804 6442, or contact us through Information Request