Information kills

In the business intelligence sphere, disruption is the name of the game, from business models to competition.
1 June 2008

The disruptive potential of business intelligence (BI) has been well documented. From Amazon to Google to Tata, effective use of BI can disrupt business offerings and the entire sector in which a company operates.

The key word in the above sentence, however, is intelligence. Merely churning out reports that provide reams of information is not providing intelligence, and not useful if your corporate strategy includes imminent world domination.

The trick is in reaching a certain level of maturity in the implementation and use of BI applications. Many companies haven't reached this level yet.

Says SAS Institute BI portfolio head Carel Badenhorst: "A lot of [local companies] are still at the stage where they are trying to integrate disparate data sources. We're still at the data warehousing or management reporting stage with a lot of companies. I think South Africa is ahead in a few industries, for example, banking where [organisations] are way ahead in the way in which they interact with customers."

Says EOH Consulting divisional director Hubert Wentzel: "BI should be disruptive if it is going to do its job properly. Acting as a change agent, BI should continually challenge business requirements and business thinking. If it succeeds, then BI becomes more strategic and is not seen as a simple reporting tool as it was perceived in the '90s.

"Very frequently, it is too much a case of driving the car by looking in the rear-view mirror. BI is seen as a reporting tool, but the technology has evolved to be more responsive," he says. "Companies can now almost report daily, instead of running reports monthly. This enables them to perform trends analysis and impacts decision-making capability."

Says Sybase SA BI practice head Estelle de Beer: "People who say BI is not disruptive are looking at it as a reporting and analytical application; that's the front-end. The way we see BI is the whole thing - analysis, data modelling, data quality, data storage and presentation to the end-user; the whole environment. From that perspective," she says, "BI has started becoming a disruptive installation or process.

"If you look at the demand - what people want - they want access to larger volumes of data than in the past, more detailed data and they want access to it in a different way. They want it to be web-based, they want to be able to drill down into the information, to explore the data. The challenge is that companies have to change their infrastructure to satisfy these demands because you cannot anticipate what users want to ask, so you can't predefine or pre-calculate answers. You can't give a summarised version."

Let them eat cake

As 3fifteen BI business unit manager Rian Durandt puts it: "Ten years ago, the tool sets cost an arm and a leg, and only executives had access to the data sets created. The information was limited to a few people. Once it became more available and distributed to every man and his dog, information became a commodity."

BI should be disruptive if it is going to do its job.
Hubert Wentzel, EOH
Users today, he says, want information that impacts their bottom line, and they want it in Excel.

"The whole paradigm shifts when the design of the solution and the development paradigm comes from the mass user. This is incredibly disruptive to how we build solutions. You can't use the old methods to build; solutions are needed in weeks, not months, and must cost less. And in terms of the information sets used, mass users want to use totally different data sets to what is expected," he says.

Adds SAS's Badenhorst: "It is not a case of doing data warehousing or management reporting; that is kind of commoditised and down pat for most companies. This is where things move

Dirty data

Cognos MD David McWilliam notes that a BI system is only as good as the information it presents. "To fully leverage their BI systems, organisations need clean data, a well designed and architected repository (i.e. a data warehouse), and reporting tools that are aligned to the needs of the organisation and can be easily used by staff at all levels of the organisation."

Says Sybase's De Beer: "The continuous integration taking place in an organisation, between ERP and CRM systems or accounting and CRM systems, has highlighted the need for clean data. And that moves onto master data and meta data.

"When we talk about master data, it means that there's a desire from the end user to have one consistent set of information about an entity, whether that is a person, customer, partner, supplier, product or service. This is important when trying to do simple things like gauge 'what is the risk we carry with X company?' You need a single view of those entities. It sounds simple, but is technically quite complicated. For example, I would be Estelle de Beer in one system, Mrs E de Beer in another and EDB in a third. This is how entities are represented in a system, and when trying to bring [this information] together, is not represented in same way."

She says one cannot simply change operational systems like SAP or Oracle. "The trend has been to buy them off the shelf and, while consolidating, de-duplicate data, try to match entities to each, and the end result is a single view. This has been very successful in certain cases, if you apply very stringent data quality and matching routines when populating the data warehouse. The trend has been to house master data in the BI environment. This works well, but also has its disadvantages. For example, only people who have access to that environment benefit, as it's not accessible to all the different applications and end users. But," she says, "now that people understand the need for and benefit of master data, it's easier to understand the procedure and build a business case to create a single source of master data outside the BI environment."

Open source issue

Making things just that little bit more interesting for BI vendors has been the disruptive moves within their own sphere, particularly from open source BI vendors. Says Paul Morgan, MD of Asyst Intelligence: "The open source software [OSS] guys have moved very quickly and gone from simple reporting tools into full suites in a couple of years. As far as disruption goes, all they've done is lower the cost, but they don't seem to be hitting the new ideas yet. They're following the lead of the proprietary vendors, but doing it cheaper. Let's see over the next few years; maybe the really exciting stuff will come from the OSS guys."

Notes PBT Group COO, Martin Rennhackkamp: "The OSS tools in the BI space are widely accepted overseas, but here there is a reluctance to go there. Local companies are slowly starting to use OSS databases, but they stay away from the reporting tools. They're too scared to stick their necks out for the new kid on a new block."

That said, he adds: "We're finding OSS technologies are moving quicker, embracing real-time feeding of warehouses and cubes more quickly, quicker than the corporates. On a functionality level, they're almost surpassing the big vendors. This will make the equation very interesting around the end of next year when they get far ahead. Take support for real-time data feeds and online event processing where you basically do analysis of events as they happen and apply BI in real time. It's good for the retail guys and cellphone guys who want to see usage patterns as they happen. The new OSS vendors are adopting that a lot quicker. Can you imagine what the big guys have to do - they've got to R&D it, which takes forever, or go Pacman another company that has done it and integrate that."

Never a dull moment

There are clearly some rather interesting times ahead in the BI space, between the upstart OSS vendors and clients that want to do whatever they want with their data, whenever and wherever, and usually via an Excel spreadsheet. This is a space that certainly bears watching.