Perhaps no phrase strikes more fear in the hearts of IT managers than "over budget."
Unfortunately, with business intelligence (BI) projects, going over budget is more than just an occasional occurrence, according to David Hatch, an analyst with Boston-based Aberdeen Group.
"The fear of hidden costs has kept many companies from making and investment [in BI]," Hatch wrote in a recent report.
Hatch said that companies often struggle to complete BI projects within a defined budget and time frame, and they routinely underestimate the costs associated with maintaining and enhancing BI systems once they're up and running.
Another 67% reported running over their BI project budgets, with nearly a quarter reporting budget overruns of 10% or more. With the worldwide recession and tightening IT budgets, any overrun could mean doom for a BI project.
That leaves around 40% of companies that completed BI projects on time and a third that kept costs in check. Hatch identified five key areas that he said helped the companies that excel at deploying BI stay on time and on budget.
- Realistically estimate project duration. BI vendors "often make claims that their solution can be deployed in a matter of days or weeks, not quarters," Hatch said. While often technically true, in order to achieve such quick deployment times, companies must have defined processes for identifying, cleansing and integrating data already in place. "Best-in-class companies are more likely to have these tasks automated," Hatch said. Otherwise, BI deployments can take weeks, months or longer, which should be understood at the start of the project.
- Reach out to end users. In order to keep BI projects on time and on budget, it is critical for companies to understand the data needs of end users. To do this, Hatch says, companies must bring together all the stakeholders involved in a BI deployment – IT, management, and end users – and he recommends creating BI competency centers to do so. "When you get to a larger enterprise where the BI tool is going to support multiple departments, you need to identify special needs and use cases to avoid alienating subsets of end users upon deployment," Hatch said. "There is a lot of project angst that goes on if this step is not performed well."
- Establish formal training programs. To make the most of a BI deployment and ensure end-user adoption, companies should establish formal training programs. "Too many times, companies will roll something out only to find the technical savvy and skill set to use it just aren't present," Hatch said. Creating a formal training program and factoring its cost at the start of the project will probably save both time and money down the road. "Training is a new cost entering the equation," he said, "and if it wasn't budgeted for, you could have a problem."
- Understand the impact on IT infrastructure. To prevent cost overruns, companies should determine up front how the BI applications will be delivered, based on end-user requirements, Hatch said. In some cases, it will be standalone deployments; in others, BI apps will be integrated with other enterprise applications, such as CRM and ERP. Either way, companies must consider the implications for their IT infrastructure, and they must budget money and resources accordingly to avoid surprises during deployment.
- Don't forget ongoing costs. Finally, too many companies fail to take into account costs related to modifying BI systems as analytic requirements change over time. "When business needs change, you have to change your reports," Hatch said. There are also maintenance costs that need to be accounted for. Companies should invest in BI systems that are easy to adapt as requirements change, he said, and factor in the related financial and personnel costs at the start of the project.
Based on survey results, companies that followed these strategies consistently completed BI project deployments on time and on budget, Hatch said. They realized "5.8% improvement in on-budget completion of BI projects, more than five times the rate of industry average companies," according to the report.
They also completed BI project deployments in around 14 days on average, three times less than the average BI deployment time overall, and they took just over half a day to make a change to existing BI reports or analytic views, compared with the industry average of just over three days, according to Hatch.
Of course, automating data cleansing and integration processes itself takes time. It can take months or longer for a company to get its data management house in order, and upkeep is a never-ending job as new data sources emerge and old ones get left behind.
Creating a BICC and keeping user skills up to date are no easy – or quick – tasks either. But, Hatch said, companies that put in the needed time on the back end will reap quicker BI deployments – and less pain -- on the front end.