It was probably inevitable that the advent of Web 2.0 would lead to a flurry of ‘2.0' coinages. One such is ‘business intelligence 2.0' as a term denoting a set of new business intelligence (BI) approaches and technologies. But is it just another piece of PR branding, or does it bear a real relationship to developments in the way BI systems are designed and used?
There has certainly been a change in thinking about what business intelligence can, or should, offer the enterprise. In the modern organisation, the development of service-oriented architecture (SOA) has made it easier to link lots of data sources together so that information flows freely between them in real time. Organisations increasingly share data with customers and suppliers and the organisational structure is less hierarchical with -decision-making more widely dispersed.
CIOs recognise the potential BI offers for making the business more profitable. A Gartner survey has found that it is top of CIOs' list of technology priorities for 2009 but, against that, the recent past has been littered with failed BI implementations.
Some would argue that the traditional model of BI, in which data is extracted from databases, loaded into a data warehouse and then, days later, turned into reports for executives to study, is not up to the job of providing the kind of information needed to make decisions in the fast-moving, inter-connected modern world. The central question that the term ‘BI 2.0' is trying to address is: how can we use data in a way that leads to better decision-making?
A large part of the answer is that we need to push information out to a much wider group of people, and not just senior managers. "If you take a step back from the technology and look at how people and organisations work, there are overlapping areas of expertise," says Richard Neale, product marketing director of SAP Business-Objects.
"It's a very complex web. BI 2.0 should be about developing BI to support this decision-making process, break down barriers between departments and people and get them to work together."
Frank Buytendijk, vice president for enterprise performance management at Oracle, agrees. He argues that the new model of the enterprise - "Enterprise 2.0", naturally - is more networked and less hierarchical. "Traditional BI tends to be aimed at organisational hierarchy," he says. "You have to have a 2.0 approach, otherwise the way you manage that business doesn't match the way the business model works. Decision-making processes can be more participative."
Of course, vendors have a vested interest in ‘democratising' BI as a way of selling more software licences. But, says Royce Bell, CEO of Accenture Information Management Services, there is a strong case for letting more people see key information. Bell argues that younger people coming into organisations today feel more comfortable than their predecessors with studying and analysing data. Letting more people see the information enables the organisation to become more agile in its response, he says: "For people to have a richer experience and get a reaction quicker, we need to devolve decision-making down, which means we need to provide those people with the information."
That democracy extends to sharing data outside the organisation. "Data should not necessarily just sit within your firewall; data could flow to your key customers, could be sent to your suppliers or sourced from your suppliers," says Andreas Bitterer, research vice president at Gartner.
By working with suppliers and partners, suggests Accenture's Bell, firms can track the way their goods are transported globally, and save money by finding more intelligent, efficient ways of moving them.
But sourcing and sharing data like this isn't easy. Bell sums up the problem. "I think I know what's happening in my own company now, and I need to reach out, not just to people in the extended supply chain, but out in the real world. I've got all this data, I've worked out what I'm doing. Now how can I connect that with the outside world?
Given that the outside world is so volatile, that's an interesting struggle."
The other problem is that allowing more people to see and analyse data may improve the quality of decision making, but only if they are seeing the right data. There is a concern in organisations that traditional sources of data are too narrow in scope. BI 2.0 is partly about taking data from a wider range of sources, some of them outside the organisation, and seeing how they tie together.
Traditional BI could tell you whether an advertising campaign, for example, has had a direct impact on sales, but BI 2.0 might use the data on rainfall to examine the relationship between sales and weather patterns, or how your firm's success on key performance indicators (KPIs) compares with industry-standard benchmarking data from the web.
Crucially, it is the users themselves who choose which data they look at. Vendors are offering much greater flexibility to users to create personal dashboards displaying their own bespoke reports instead of looking at a pre-defined set of reports.
This ability to combine different sources of data has been accompanied by different ways of presenting data, following the web model of mashups, or composite applications that mix and match tools, data sets, ways of viewing them, and so on. For some time, BI vendors have been making products that create clear visual representations of data, through dashboards and scorecards and traffic light colour-coding to demonstrate performance against KPIs. BI 2.0 takes this further by drawing in web-based applications: the insurance market Lloyd's of London, for example, combined complex insurance- risk data with Google Earth to create simple visual representations of insurance risk.
The sources of data used in BI 2.0 could even be extended to use text analytics to analyse information from unstructured sources on the web, such as customer blogs or forums. This kind of information can be immensely valuable as a guide to what customers are thinking, says SAP's Neale: "You can look at trends. You can say, ‘The number of calls coming into the call centre is increasing and they're increasingly negative: maybe we've got a product quality problem.' This unstructured information might be a leading indicator for what's going on in your business. You see customer complaints before you get hit in the sales pipeline. If CIOs were thinking about the potential of using unstructured information in their business, in a BI context they could really have a profound change on the way an organisation works."
A more controversial element of BI 2.0 is the idea of ‘real-time reporting' so that the decision-makers can see relevant business data as soon as it is created. This can be done either by getting data into and out of the data warehouse more quickly, or by bypassing the data warehouse altogether. In the latter model, the data warehouse is simply used as a repository for historical data, while middleware is used to enable the relevant business process data to flow straight from the application to the desktop. It's an attractive idea: in a fast-changing business environment, surely it must be advantageous to know about events as soon as they happen?
Up to a point, this is true, according to Neale. "There are some areas where real-time information is of massive value to the organisation - if it's very volatile and the value of the information is changing over time. If you're selling a finite resource, such as an airline seat, its value goes up as the amount of the resource goes down. You've got to have the most up-to-date information because the value of the product you're selling is based on up-to-date information," he says.
But real-time data is more expensive to deliver, he adds, and is often not necessary: a financial services company needs to know how many mortgages it sold yesterday, not five minutes ago. Indeed, Accenture's Bell believes that in some cases, real-time -reporting could do more harm than good. "Real-time financials are pretty meaningless.
Real-time alerts on things like sales may be useful but you've got to think quite carefully before you react. Knee-jerk reactions are often a poor reaction until you've done an analysis," he says.
The challenge that all BI implementations face is that giving data to people is only useful if they know what to do with it. Even with high-quality data from key sources, it is easy to make bad decisions. The next phase of BI will both be about -using more sophisticated analytic software to understand trends, and about hiring people with strong analytical skills. The use of statistical techniques to understand customer behaviour, for example, and to predict future trends, will become increasingly important, says Bell.
"Businesses will start to look at things that were only done every couple of years on a scenario planning basis every month or every week. You'll get a lot of that continued testing of a particular strategy or set of tactics against the market, with continual- readjustment," he argues.
What does all this mean for the CIO? Many have already had to adapt to the challenges brought by increased consolidation in the BI marketplace, with Oracle acquiring Hyperion, IBM acquiring Cognos and SAP acquiring Business Objects. If this has reduced choice, it has also in some cases made it easier to rationalise the number of BI platforms and to improve integration.
But letting users choose what data they want to source and what reports they want to receive means losing a certain amount of top-down control. Greater democracy leads to increased requirements for good governance and cast-iron security, says Gartner's Bitterer.
"You certainly want to have decentralised users, but also more centralised governance of the whole infrastructure.," he says. Bitterer is also sceptical about whether- the change in approach BI 2.0 brings will necessarily lead to more successful BI implementations. "The technology is rarely the problem. Today we can turn out thousands of reports a minute, we can fulfil reporting and analysis needs for tens of thousands of users," he says.
"The issue is more home-grown: -internal politics, lack of definitions, lack of metrics. People don't even know what they want, so they typically come back with the lowest common denominator, which is ‘Give me some reports', and they think this will turn into some big benefit, which it typically doesn't."
Poor data quality also remains a serious problem in many organisations; without accurate data conforming to standard definitions, the quality of decision-making is inevitably compromised. His recommendation is that CIOs, in tandem with the business, take more central control over how BI is implemented and used. BI, he points out, is a "hugely complex thing with many moving parts: tables and metrics and metadata and master data and dashboards."
Bitterer recommends creating a BI competence centre: a strategic group that -defines what the business wants to do with BI and identifies the priorities, the gaps and how to align BI investments with what the business requires. Without this strategic- approach, BI implementations are likely to go on failing, however good the software.
Case Study: Channel 4
In 2006, Channel 4 launched 4 on Demand (4oD), its first video-on-demand service, offering viewers the chance to watch the broadcaster's programmes on their PCs or on televisions through a third-party service provider such as Virgin.
Using SAP BusinessObjects Data Integrator, it created a data warehouse to capture two sets of information. One set showed which programmes were being watched, the number of viewers watching them, and the platforms they were watching them on; the other showed whether Channel 4 and its resellers were meeting their SLAs to deliver and broadcast the programmes on time.
Channel 4 was the first broadcaster to offer a video-on-demand service, says John Telford, project manager. "When we went live there were different opinions about what business model might be successful, but without doing the BI work, we had no way of finding out what would be successful and popular with the customers," he says.
Reports that are created using SAP BusinessObjects XI are delivered daily to executives' desktops or mobile devices. Data is represented graphically or in table form, says Telford, and SAP BusinessObjects alerts are used to highlight missed SLAs.
"You see a list of programmes that have gone into VoD in the last month and any programmes that miss their SLAs are highlighted in red," he explains.
The solution has brought very direct benefits. "It's allowed the C4 business to grow video-on-demand in terms of the volume of programmes on 4oD and the number of platforms it's on. It's also allowed us to refine the offering so we can be sure that it's in a direction that will lead to most profits," says Telford.