A business need for better intelligence

Far too many business intelligence systems fail to live up to their promises. Expectations are rarely met, and profitable returns on investment are even rarer

It is almost too difficult to measure the positive aspects of business intelligence (BI). And it is usually impossible to calculate whether the value of implementing it outweighs all the money that has been shelled out on BI in the hope of boosting productivity.

The problem is that companies have spent too much on these systems to abandon them now. Firms will have to keep spending before they get the returns they were gullible enough to expect in the first place.

“The market is increasingly sceptical about the ability of technology to deliver substantial, sustainable returns on BI,” said Ian Charlesworth, a senior analyst at the Butler Group.

The opportunity for the channel is in identifying where the weak links are and eliminating the problems.

The consensus among BI vendors is that, of the six main processes carried out in a BI system, by far the biggest source of error is in the data.

If the data is hopeless, it does not matter how well integrated a system is, or how beautifully the intelligence is presented, it is useless. Actually, it is worse than useless, it is counter productive. This is what the channel should be concentrating on, according to Alan Parker, sales manager at BI vendor Temtec.

“Static and dead reports go back to the beginnings of management theory,” Parker told CRN. “It is time we started doing something about it.”

It is bad enough that only 10 per cent of users – the senior executives – get access to BI. But if the data is no good, then firms are more likely to come closer to a zero per cent return on their investment.

The costs associated with poor data, according to the Butler Group, stem from three major areas.

First, there is the problem of inconsistent management information. This leads to a situation where analysis and results can conflict with one another. This is worse than useless, as it causes firms to waste more time discussing data sources and arguing over calculations and re-running analyses.

Then there is the high cost of maintaining records. Every month, as much as three per cent of a businesses customer records become inaccurate.

Finally, there is the age-old problem of data entry errors. OK, it is impossible to stop human error. But the impact of the daily input of errors can be minimised if firms have a system to nip them in the bud. Otherwise, automated systems will repeat those errors as the information spreads and the damage will be multiplied.

The input of data comes from all the various applications that are used to record transactions. ERP, CRM and event contact management systems are all used to capture information. That information all ends up being duplicated into the information silo (the data warehouse) that is used as the basis for all analysis.

Spencer Marlow, retail solutions manager at Sterling Commerce, said the best way round it is to ensure closer collaboration between trading partners and integrate all applications as tightly as possible. It is going to get harder though. With technology such as Global Positioning Systems and radio frequency identification tags feeding more information into the system, it is vital that data discipline is kept.

“Without global data synchronisation, it could be a catastrophe, exacerbating the profound issues of data quality that already plague our supply chains,” said Spencer.

Haven’t we heard all this before? Data cleansing tools were supposed to eliminate duplicate information and tidy up errors. The extract, transform and load process built into any BI system should put data into the right format for storage.

But has it? No, it has failed spectacularly, said Phil Tee, chief executive of Njini. It is all very well having all this wonderful information on everyone, but it is useless unless we are disciplined about putting it away and finding it again. Phil Tee’s new company (he was the founder of Micromuse and Riversoft before discovering the BI market) may have solved the problem.

Njini has claimed to have created the first software to provide data identity management by tagging data intelligently at the point of origin.

The vendor behind the software claimed it works automatically, in real time and is the only software utility that is capable of performing this operation for in-bound data.

Every time some useful information is stored, it is given a comprehensive set of rules and flags so that it can be easily retrieved. This makes the storage and retrieval of information far more efficient. It also makes intelligence systems a lot more effective.

Categorised as ‘Information Asset Management’, by industry analysts, Njini’s offering claims to bring rapid return on equity and extend information value to business direction.

“Companies need all the information they can get,” Tee said, “But without a system of categorising it as they capture it, it becomes a burden.”

Tee’s experience with Micromuse and Riversoft taught him all he needs to know about managing the channel. Now Tee sees the same sort of growth in the intelligence market.

Structured data grows at 10 per cent a year, unstructured data grows at 60 per cent, explained Tee. The problem is that unstructured data is useless to a BI system, as it can’t handle the data. There is no simple solution to manage and control the proliferation of unstructured data and this is the problem that Njini addresses.

But that is not the problem that most IT directors identify, according to Andy Hayler, founder and chief strategist at Kalido. Hayler probably has a better idea than most, since he was an IT director until he came up with a solution that he recognised most companies would want.

It is not about the quality of the data, or the efficiency of its storage, he said, it is about getting to the sources of data. Hayler said it is so important to ask where the information is coming from. In many companies, there is no single view of the world, he said.

Isn’t that what SAP is for?

“Well, that is my point,” answered Hayler. “At Shell, where I used to work, there are multiples of different systems. It wouldn’t be so bad if they were uniform, but they are all subtly different from each other.”

Worse than that, a SAP system only really accounts for between 30 and 70 per cent of all the company information, he explained. The rest will be in a host of specialised systems, such as CRM, or the supply chain system.

The traditional way around that was to have a data warehouse, which draws on multiple sources of information, copies it and puts it in one consistent form. Then you can put some tools on top of that and run your queries on that information database.

Sadly it doesn’t work.

“The structure of the data is too brittle. Databases like this aren’t good at coping when the source of information changes,” said Hayler.

For example, when two companies merge, the way they store information is different. It takes months for the two sets of information to be ratified so they are consistent with each other.

“The market has been crying out for a data warehouse that copes with change. I know that from my experience of having been on the wrong end of the problem as an IT director. This is why we built one,” claimed Hayler.

All the big vendors say the solution is standardisation. Surely then, that the answer?

Not so, said Hayler. The problems are in the data, not the systems that manipulate the data. Shell spent $1.5bn implementing SAP, but they still ended up with 100 different SAP systems, as well as several legacy JD Edwards systems, he added.

“It is unavoidable really. You don’t get uniform markets, so you can’t have a one-size-fits-all system. Different factors are at play in different countries. In Germany, for example, the sales of lubricants might be driven by the car market. But in Vietnam, the sales of the same products could be driven by the motorbike market. You can not reflect all these conditions when there are so many variables,” he said.

Does this mean BI resellers don’t listen to their customers?

“No, but I typically think a new information system is great on the day it is put in, but six months later it falls over,” Hayler told CRN.

That is when they actually become counter productive, because after that, everyone has a different view of the information and interprets it differently. “You get to the situation where everyone spends all day arguing over the figures. The people from the German office might say one thing, but in the UK office they might see things completely differently,” Hayler said.

Resellers are wrong if they think BI is about reporting tools, like those from Cognos and Hyperion.

“Well, they are very interesting and very pretty but they don’t always work,” said Hayler. Besides, there needs to be a new approach, because the traditional approach to selling BI is becoming very hackneyed. “If I had a penny for every time I have heard the phrase ‘one view of the truth’ I’d be a rich man.”

There is a school of thought that BI tools are no longer a rich corporation’s plaything. Now the market has matured, SMEs might be prepared to buy them.

Really? Can BI solutions be sold to SMEs? Definitely, said Temtec’s sales director Alan Parker. “The same principles apply to any organisation, no matter the size. As soon as managers recognise the need to measure the output of their work and understand the causes of deviation, a dynamic interactive tool will accelerate their effectiveness.”

Support, training and consultancy will all be demonstrably worth paying for. Successful solutions call for a holistic approach. There also needs to be given time to evolve. That won’t happen without quality support, training and services. Businesses need to own their solutions. So the smart reseller should concentrate its efforts on providing a skills transfer process and great support and back-up to customers.

BI is for everyone, agreed Richard Neale, product marketing manager at Business Objects. “No matter how small your company is, you can use intelligence. You can always get information that will point out something you didn’t know about your spending patterns, or where your customers are. It might be something you want to do something about.”

Take for example the SME whose salesforce has collectively realised that travel expenses are not being scrutinised. As word gets around, slowly but surely the expenses payouts will creep up. This is something that a trends report would tell you straight away. You could nip all the expense fiddling in the bud and save yourself a lot of money.