See also: Business Intelligence, a timeline

The business intelligence (BI) challenges facing CIOs vary enormously, depending on the sector and their current infrastructure.

“If you’re in the reporting Stone Age, getting from 0 to 60 will be a big challenge,” says BI analyst Andreas Bitterer, VP of research at Gartner.

According to Bitterer, there are many CIOs in global corporations who are creating fantastic systems for their constituents, only for their users to prove to be wasteful.

“Often the CIOs generate masses of fantastic information for their users, who then just throw these impressive reports straight over the wall,” says Bitterer.

A business intelligence report is like a cross between a corporate comfort blanket and a shield you carry into a boardroom battle.

Nobody can accuse you of not being good at your job if you’ve got a massive tablet of data you can throw at them.

Good understanding
The challenge of good BI, says Bitterer, is to create something the users understand.

“You need to find and present information that makes sense to someone, and gives them something that they can act on,” he says.

The problem is, he argues, too often people aren’t able to articulate what they need.

One of the most over-used phrases in any corporation, he says, is “can you give me a report on that”, but the problem is that most organisations have far too much information anyway.

The one issue that all CIOs have in common, no matter where they are and what challenges they have, is that the very people they must work with can make or break any BI project.

Understanding the mentality of the human resources at your disposal is a major boon in ensuring a successful BI strategy. You can tell a business intelligence strategy is successful when everyone wants to take credit for it.

The fact that so much can be achieved by technology is probably a good reason for the eventual failure. At the moment, we are cursed with a bewildering array of new innovations.

The big trends are: big data, consumerisation, mobile access and the emergence of open source alternatives to traditional BI tools.

Tablets are giving users increasing mobility, improving their access to systems when they are out the office and whetting their appetite for more information and reports by offering user friendly interfaces.

Cloud computing is offering more options for business intelligence apps.

Some argue that cloud-enabled BI apps should be avoided at all costs, given that they take sensitive company information and entice it outside the company firewall.

That’s not a distributed architecture, argues Bitterer, that’s a company whose governance is all over the place.

The increase in the number of tablets and smartphones has driven another wearisome (for the CIO) development in business intelligence.

Not only are these devices giving people more chance to access reports, they have created a new type of input.

Social media and social trends are newly created areas that the CIO must cover too.

“People want to experience data the way they do at home,” reports Quentin Gallivan, CIO for open source software vendor Pentaho.  

Big data, big issues
This new, unstructured data, including Twitter feeds, Facebook, Google+ and podcasts, has to be quantified and searched too.

Is this the future? Or are the users are causing more problems than they are solving?

In some ways the consumerisation of IT has already spread to the IT department, says Gallivan.

All the hard work that has gone into business intelligence will have to be redone as BI is drawn from different sources and systems.

Giving users a complete overview of everything, with a point-and-click level of usability and meaningful metrics will be a monumental task, akin to reinventing the computer.

“The CIO will be expected to bring all this together,” says Gallivan.

It’s as if someone is on a work creation scheme for CIOs. Another trend they must keep tabs on is the IBM-type social network analysis attempt to see which individuals have the biggest following and which are the most influential.

As if that were not enough to cope with, there is a groundswell of opinion that traditional business intelligence systems are massively over-priced and over-rated and that open source systems could create a similar level of insight for a fraction of the cost and manpower.

Pentaho, which has created healthcare analytics systems on both sides of the Atlantic, would say that it’s quite doable already. Arguably, the climate for BI projects in the NHS is better now that expectations will be so much lower.

Expectations for business intelligence are one of its biggest problems.

Great potential, little power
The problem business intelligence faced was that the people behind it thought it had so much potential it ought to do something wonderful every time it is deployed.

In these circumstances the upshot is that it can’t be used to do the simple, basic tasks, because it is too complicated to use.

The result is that powerful BI systems are effectively useless in many situations.

This is the age-old problem with technology. When it is over-engineered and all-encompassing, it becomes difficult to galvanise into action because the preparation for any initiative is so daunting.

Donald Farmer, VP of product management for QlikTech, likens the modern BI market to the networking market of the 1990s.

In those days, when he was installing networks for small health centres, there were two types of technology available. Cheap and nasty Ethernet and IBM’s reassuringly expensive Token Ring, which could cope with all eventualities.

The eventuality that Ethernet became the technology of choice.

There’s a lesson here for the business intelligence sector, says Farmer. Sometimes, by narrowing your focus and limiting your expectations you go much further.

“I didn’t have time to learn Token Ring,” says Farmer. “All my clients were small practises who just wanted something that did the job.”

These days Ethernet is a carrier-class technology. Like Ethernet, BI systems can start small and evolve as the requirements become more complex and demanding.

Anything over-complicated runs the risk of joining Token Ring in IBM’s museum of forgotten technologies.

QlikTech’s Farmer argues that BI is far too over-complicated. Whereas some BI systems with complicated requirements take a year to create, QlikTech can create a proof of concept in three days and a system in three weeks.

“You have to move fast otherwise by the time a BI system is built, the original conditions might have changed,” says Farmer.

The advocates of big data might argue that BI is more sophisticated than ever, but don’t get carried away, adds Farmer.

This is a time when the discipline of keeping things simple is more important than ever, he argues. “You have to keep everything cheap, simple to display and elegant,” he says.

The problem with big reports on big data is that you need to know which questions to ask in advance, and that never happens in real life.

Then you have to know the relationships between data sets, and those are rarely obvious.

So you need something that can be rapidly re-configured, argues Farmer. You need something that mirrors the way humans think, which isn’t regimented and cast in stone.

“The unexpected questions that arise half way through a study are the ones that help you find the best insight,” he says.

Don’t assume you can dismiss big data altogether, warns Eddie Short, a partner for business intelligence at KPMG Europe.

Even though ‘big data’ has all the hallmarks of being a big fuss over nothing, and there’s no signs that anyone is actually taking it seriously enough to develop products for it, that doesn’t mean it should be dismissed.

Short agrees that it might be all hype, “but who would have thought a few years ago that you would be able to build a business based on Google?” he asks.

Business intelligence is supposed to be leading the changes, not developing so slow that it can’t keep up.

“A lot of IT is there to support an organisation when it’s business as usual. Not business intelligence. It should be flexible enough to support the organisation as it changes,” says Short.

If all the information about business intelligence was put into one datawarehouse in a datacentre (or alternatively in the cloud) and a series of queries were run, what patterns would emerge? What lessons would we learn?

Dan Burrows, senior consultant in business technology at IT consultancy Waterston’s, identifies two lessons of past BI failures that should be heeded by CIOs.

One is that business processes, which is always overlooked should be the things you concentrate on now before you’re forced into any panic buys.

The other lesson is that IT suppliers will always insist you chuck out the baby with the bathwater and buy the latest and greatest (in terms of cost) technology in their portfolio.

To give a specific example, many hospitals and PCTs already had decent workable patient record systems that could have been adapted at a fraction of the cost of the NPfIT disaster.

“There’s a history of good IT systems being torn out by the NHS,” says Burrows.

Human touch
The point is that systems don’t have to be perfect and you can’t pre-plan processes, according to Mike Lynch, former CEO of Autonomy.

It’s far better to get people to adopt a system first,and let it evolve as the human networks grow, much as Ethernet evolved as physical networks grew.

“Eighty-five per cent of information in a business is human information,” says Lynch.

“Video, voice and text – that’s where all the interesting stuff happens. It’s growing faster than the structured stuff, and it’s where CIOs are looking to find value.”

Which seems to imply that CIOs should be followers, rather than leaders. The trick is to keep running in front of the parade and pretend to be leading it.

“We live in a real-time economy and need to act as things happen to ensure success,” Lynch adds. “Gaining intelligence from data requires understanding it and acting on those insights automatically and in real time.”