Real-time BI not viable - try right-time BI!

Real-time business intelligence (BI) is the latest buzzword in BI. It is expected to contribute heavily in enabling faster decision-making and speedier business responses.

However, without a clearly defined information strategy, integrated technology architecture and suitably adjusted business processes, procedures and responsibilities, real-time BI offers few benefits. Right-time BI, on the other hand, does.

"While BI has evolved from being an optional investment, to becoming a key investment as a decision-support mechanism for organisations seeking to retain their competitive leadership, organisations in South Africa are only just coming to grips with BI.

"Few have the physical, technological and operational infrastructure in place to gain tangible benefits from real-time BI. Practical benefits can, however, be obtained by speeding the collection of strategic information and aligning reporting with business rhythm," says Carlo Gunter, COO, e.com institute.

A critical step would be to define what information is needed, when and by which business process in the organisation. "Business activity monitoring is high on companies' agendas at present, for example," Gunter explains.

"An accelerated flow of information is needed to enable decision making that is more responsive to the environment. The key is to move data extraction, transformation and loading (ETL) closer to real-time, but this is easier said than done as batch-driven operational systems fail to handle real-time BI. The answer is to move to right-time BI," he emphasises.

There are a number of new and hybrid methodologies that are being used at present to enable right-time BI. They include enterprise application integration (EAI) coupled with ETL and ELT; and a shift to the use of a more agile, distributed intelligence and operational BI model.

Explains Gunter: "There are two types of data management mechanisms - ETL/ELT and EAI. ETL is the standard batch-loading of data as applied in traditional BI models, with a static data warehouse, while ELT is used for transformation of data on the fly, and real-time reporting from operational systems.

Where both are utilised within an organisation, they become more supportive of each other and begin to converge to reach a single goal - appropriate BI at the right time.

The key to success is building smart real-time BI frameworks that focus on more responsive processes that offer a bigger payback.

There are currently heated debates around distributed BI systems. According to Gunter, they defy the traditional centralised and static data warehouse BI model, to create a distributed BI network (BIN) by embedding knowledge-gathering agents into distributed islands of data. And they are here to stay, he says, as they are more critical to real-time, or closer to real-time, transformation of data for reporting.

These knowledge agents are essentially an analytics layer that sits above the application layer and acts as the interface between business decision-makers and the information systems. The network, not the data warehouse, thus becomes the primary source of business event-based intelligence, feeding rich client and browser-based reports (such as interactive operations dashboards) via Web services.

"We are already receiving requests of this nature from clients," says Gunter. "Companies want real-time applications to enable performance management (PM) and dashboarding. However, few have suitable PM or even BI infrastructure strategies in place to enable them to leverage real-time BI applications. Despite what BI vendors are saying, PM and dashboarding applications are not silver bullet solutions - there are physical, real-world operational and process constraints."

The BI tool is only as good as methodology. "PM and dashboarding applications are built around triggers and the identification of bottlenecks, but they are useless unless they are linked to solid strategies," says Gunter.

"Understanding the performance and dynamic behaviour of workflow is crucial in being able to modify, maintain, and improve it. At present, users don't know what to do when the dashboard goes 'red' and alarm bells start ringing. It's a challenge to interpret reports into actions."

In SA, we are still very new to real-time BI and are just finding our feet with regard to traditional BI. "Take for example the difference in the requirements and expectations of customers in SA and the US," notes Gunter.

"Amazon needs real-time analysis of its clients and supply chain, and customers' expectations regarding service are very much higher. Locally, while there may be a real-time information requirement, there is a lack of will to pursue this avenue. We need to catch up both in terms of business culture and business methodology."

Locally, the banking industry comes closest in terms of demand for real-time BI, with an operational requirement for credit card fraud protection or for strategically ensuring competitiveness. In the pharmaceutical industry, real time BI is used in processing plants to analyse batching and to enable an immediate halt to production if anomalies are spotted. The travel and hospitality industry also has some requirement with inventory monitoring, allowing them to spot shortfalls or overcapacity and change prices accordingly.

"However," says Gunter, "If the data warehouse and BIN are designed and integrated properly, businesses will attain new levels of visibility. The aim is to bring business challenges to light daily, not just weekly, monthly or quarterly, and communicate these challenges directly to the business manager for an appropriate response.

"With the advantages of Web services and the ability to create loosely-coupled technology environments, organisations can leverage historic methods and bring speed and intelligence that business managers need, to meet new regulatory guidelines and exceed business expectations of employees, customers, partners and shareholders," he concludes.