Five steps to business intelligence success

Start small but think big

Published: 13 May 2009 15:59 GMT by Danny Bradbury

Tags: business intelligence, enterprise software

Building a business intelligence system is a complex and daunting task, and yet executives need that data, especially in dire economic times when all of the information could help make their businesses more efficient.

Here are silicon.com's top five tips for implementing a successful business intelligence project.

Set your goals
"Critical success factors 1, 2, 3, 4, 5 are simply this: business alignment," says Ronan Miles, chair of the Oracle UK user group, a collection of CIOs who use Oracle database and business intelligence products.

Too often, BI can be a technology-driven project, with insufficient input from business executives. Companies can end up spending large amounts on the infrastructure and reporting tools that make up BI without a clear understanding of what they are trying to achieve.

Simply assuming that it will give you more knowledge about the business is not enough. Think about the employees who will be using the system, and what they need to achieve.

Andreas Bitterer, research VP at Gartner, says the potential applications for BI vary between sectors. "The guys in procurement want to buy low," he says. They may wish to analyse suppliers, volume and prices in an attempt to consolidate and drive down prices through economies of scale.

Conversely, financial institutions might be more interested in driving down risk by avoiding bad credit decisions, he says. "If the financial institutions had more insight into what products [their] customers have across business units that could generate a certain risk score for a customer," he explains.

Nurse your users
"The number one glaring symptom of a failed implementation that we see all the time is a lack of focus on user adoption. You can build the greatest BI reporting solution ever, but if you can't get your users to use it, then it's going to fail," says Jeremiah Johnson, executive director of the US-based user group for Oracle's business intelligence products.

Mike Jelen, vice president of education at business intelligence consultancy the BI Consulting Group, says training plays a big part in getting users comfortable with what can be daunting software. However, it goes beyond that, according to Johnson.

"It's about building out your dashboards to make sure that they're intuitive for the end user," Johnson says. "Often you'll get the prepackaged applications that don't look like anything you have internally, and that can be difficult for non-savvy users."

Oracle user group's Miles recalls one business intelligence application designed to help telecoms employees understand calling rates, volumes and distances. The developers crafted a sophisticated interface that allowed business users to fly around a virtual representation of a country and view the data in 3D. "The problem is that they needed a degree to use it," he says. Sometimes, it is better to do things simply, rather than trying to boil the ocean.

Start small
One common mistake companies make when building business intelligence systems is to try and create a behemoth that covers all aspects of the business."If you release this enormous application it can sometimes be overwhelming for the user," says US Oracle user group's Johnson. "If you phase in parts of your project, that will not only encourage increased user adoption but you'll also be able to gauge how to continue down the road of mapping out future phases of your BI project and learn what works as you go along, rather than building it all out and finding out what your mistakes were."

This can also affect whether you can get sponsorship from senior management - something which BI Consulting Group's Jelen says is a key factor in ensuring the success of the project. Executives, especially during an economic downturn, may not want to commit large amounts of money to a broad-ranging BI project. But a pilot project that can deliver benefits to a small area of the business might make it easier to gain traction in the future.

Think big
Gartner's Bitterer says any business intelligence plans should be a wide-ranging strategic programme, rather than simply a project. "BI is not just a project - it's much bigger than that. You don't just do it on the side. You need a proper organisational structure that runs a proper programme," he says.

Programme management collects groups of projects together to support a broad strategic movement within the organisation. BI's complexity, and the extent to which it touches many different disciplines within an organisation, makes it a prime candidate for programme management.

Bitterer says this doesn't mean you have to get every single stakeholder on board from the first week. But it does mean creating a committee of key individuals in the first instance who can then pull together these resources as a programme takes shape.

The bottom line is that treating BI as a small, discrete project in its own right could leave you disappointed in the long run.

Clean your data
The 'garbage in, garbage out ' principle is as old as modern computing, and it still holds true with BI. Pulling data in from multiple sources isn't simply a case of integrating systems; the data needs to be verified, and its lineage ensured.

"Data validation is one of the first areas that gets cut short, because people say they don't have time," says BI Consulting Group's Jelen. One particular challenge that ties into this is gaining a 'single version of the truth' across your organisation, as many companies' data is stored in disparate systems that have not been pulled together.

Master data management is a critical component of BI, which enables you to reconcile duplicate records so that, for example, you don't end up with multiple different views of the same customer polluting your data when trying to perform trends analysis.

"Master data management is a hot topic. Many companies don't understand across their entire organisation from a horizontal perspective who their customers are, what they are buying, and where they are buying it from," he says.



Reader Comments (0)

Add a comment

Quick Sitemap Links: