You may be asking yourself: Should we implement a BI system?
Although an increasingly popular and accepted technology for improving business performance, BI – like any other system – must be carefully considered before deciding to implement it. Benefits must be clear and understood and they must justify the investments.
So let’s elaborate on the exact benefits of implementing a BI system in your organization.
First, a bit of introduction to my perspective on this subject. If you ask companies which benefits they want from their BI systems you will get one answer. If you ask a BI expert you will get another. In my experience, most companies only grasp the benefits in tiny pieces and they either can’t express the benefits they are getting or they are not aware that they can get them. Oftentimes both.
For the purpose of giving the most complete picture of Business Intelligence benefits I have chosen to describe them from an expert point of view. I.e. from the perspective of a professional who has been working with multiple BI systems at multiple organizations for the past 15 years.
Reduced labor costs
The most tangible benefit of BI is the time and effort saved with manually producing the standard reports for your organization. It is rarely the largest benefit though. However, because it is so tangible it is often part of the equation when a decision must be made about implementing BI, and if it turns out that these savings alone can justify the BI system, then it is the easiest way to justify it.
BI systems reduce labor costs for generating reports by:
• automating data collection and aggregation
• automating report generation
• providing report design tools that make programming of new reports much simpler
• reducing training needed for developing and maintaining reports
Reduce information bottlenecks
The BI system allows end-users to extract reports when they need them rather than depending on people in the IT or financial department to prepare them. The BI system will even allow authorized users to design new reports to match their requirements.
BI systems reduce information bottlenecks by:
• providing individualized, role-based dashboards that collect the most important data for daily operations
• letting the user open and run reports autonomously
• providing documentation of KPIs and other information
• allowing users to analyze and validate the data without involving IT specialists
• allowing users to create new views of data as needed
Make data actionable
What happens when employees in an organization get too much data, too little data, too old data, too detailed data or just irrelevant data?
Nothing happens. Everybody is just wasting time and resources.
Most organizations use extensive amounts of resources putting together piles of standard reports that are delegated throughout the company. To make sure everyone has every information they need, all kinds of reports are sent to employees – usually on a very detailed level. As a result employees feel overwhelmed by the amounts of information that don’t give a clear picture of the overall situation. And moreover, since so much effort is required to assemble the reports they usually arrive at the employees’ desktop days or weeks after they were most relevant.
All put together this means that the potential corrective and opportunistic actions that these data could have led to, are missed due to either being too late or because the employees overlooked or were out of time to find the relevant trends in the myriads of information.
When employees try to find head and tail of the data they even often find that the numbers are not comparable between different reports and end up analyzing the differences instead of interpreting the actual numbers. And since trust in data is lost, nobody dares to make a decision based on the numbers.
But worse yet: Many employees don’t have the training and knowledge to interpret the numbers in order to identify threats and opportunities.
BI systems make information actionable by:
• providing information through unified views of data where KPIs are assembled and calculated using a central repository of definitions – a data model – to prevent conflicting definitions and incomparable report data
• providing to-the-minute information in real-time reports that show the state of the business in this very moment – not a historical view of how it looked days or weeks ago
• allowing users to search and design reports autonomously instead of being dependent on specialists in the IT department
• showing data in a context, e.g. by benchmarking KPI values against comparable values (e.g. averages, budgets/target and last period) to let the user interpret whether the KPI value is acceptable or needs corrective action
• using rules to highlight KPI thresholds as “good” or “bad”
• providing integrated documentation to help the user understand the meaning and definition of the KPIs
• providing links back to the operational systems that make it easy for the user to carry out corrective actions (closed loop)
• making data collaborative, e.g. let the user forward and share selected data with other users and assign targets and responsible persons to KPIs
• only showing data relevant to the specific user in a role-based environment to avoid “Information overload”
• showing data on a high, aggregated level where overall trends can be easily spotted and then let the user drill-down to detail data in a top-down manner
• using intuitive visualizations that enhance on the nature of the data such as graphs/charts and gauges
• forwarding relevant information based on the occurrence of predefined events, i.e. only sending certain reports when specific business events occur, such as too high stock levels, customer churn etc.
• shortening the analyze-decision loop to avoid losing the train of thought
Decisions need to be made every day and, as we all know, decisions have varying quality. Good decisions can provide tremendous benefits. Bad decisions provide no benefits – they may even cause you losses.
BI systems help make better decisions by:
• providing decision makers with rich, exact and up-to-date information
• letting users dive into data for further investigation
In this context the term decision maker needs to be seen in a broad perspective; it is not only management that makes decisions. In fact, the decisions that affect an organization the most are those made by people all over the organization, from the sales person who decides to give a customer a discount to the procurement assistant who decides to buy certain products for inventory.
A decision can be made the moment you have all the relevant information at your hands. In other words, the faster the relevant information gets into your hands the faster you can make a decision.
Fast decisions are important for two reasons:
1. It makes the organization more responsive to threats and opportunities
2. It shortens the time between thought and action. Most people will lose their train of thought if they need to wait a long time for further information about the problem they are dealing with.
BI systems enable fast decisions by:
• combining multiple data sources in common reports, thus saving the user from manually combining data in spreadsheets etc.
• providing analytical and ad-hoc reporting capabilities that allow users to quickly retrieve new or different combinations of data as needed instead of having to request new reports in the IT or financial departments.
• providing reduced system response times by using pre-aggregated data or other techniques for fast data aggregation.
Align the organization towards its business objectives
The most successful organizations are those that succeed to make every person in the organization work towards a common goal.
BI systems help organizations align all parts of the organization towards common business objectives by:
• centralizing KPI definitions. BI reports don’t calculate KPIs using autonomous queries and scripts. They retrieve KPI values and definitions through a central repository and thus prevent conflicting KPI definitions and values
• guiding information presentation using advanced visualisations, benchmarks and KPI thresholds thus ensuring a common interpretation of the KPIs
• providing a single source of information. All reports collect their data from one source – the BI system.
• “pushing” selected information throughout the organization. By enabling organizations to push KPIs and other information to the end-users the BI system helps focus employees’ attention on the most critical success factors.
• assigning targets to KPI values for each organizational unit to be used for measuring the ability to achieve the goals set forth and thus pushing the organization towards the defined goals. These uses of BI systems are sometimes referred to as Performance Management.
Traditional reporting systems aim to give users data according to a fixed and predefined structure. This rigid approach gives the organization answers to exactly the questions it is able to specify in advance. And no more. Modern business intelligence systems on the other hand provide ad hoc query capabilities that allow users to poke randomly around in data to get answers to any question that comes to their mind. This allows users to strengthen there understanding of the underlying patterns of the business and thus to gain new insights into the dynamics that lead to success or failure.
Such analysis is often referred to as OLAP: Online Analytical Processing.
In another application, some BI systems provide special mathematical algorithms for finding hitherto unknown patterns in data – so-called data mining. Such algorithtms comprise Cluster Analysis, Decision Trees, Neural Networks and Rule Induction.
Data mining algorithms are advanced statistical methods that attempt to uncover patterns automatically and thus help the organization answer questions such as “which variable is most important in determining customer churn?” and help discover rules such as “92% of all Bluband butter is sold with Agege Bread”. Most Data mining techniques require a deep understanding of mathematics to make their results actionable and they require large amounts of data in order to be statistically significant. Thus, data mining is not for everyone.
Habanero Data Blogging Team.