News Stay informed about the latest enterprise technology news and product updates.

Key success factors and tips for BI in ERP integration projects

BI and ERP integration is challenging, even when you use integrated technology. Get expert advice, specific tips and key considerations for BI and ERP integration projects.

Business intelligence (BI) and ERP integration has challenges regardless of the vendor choice -- challenges that are both political and technical. BI initiatives often come to an organization from many different angles, supported by company players with different needs and different levels of understanding. There are also the costs, the strategic investments in ERP, the best-of-breed BI silos that are delivering focused value for one department but failing miserably for another. So how does an ERP-focused organization regroup and start making sense of all the nuts and bolts that must come together for a cohesive BI strategy?

"Begin with the end in mind. Start with the KPIs," said R. "Ray" Wang, a partner in enterprise strategy with Altimeter Group. "[Next] follow the information supply chain. Find out which processes touch which people, when and why. Invest in good data governance and processes before buying the technology."

For more on ERP and BI integration
Learn whether to get BI from your ERP vendor or a third party

Learn how to build a BI business case in ERP environments

See why MDM and data quality are key considerations in BI and ERP integration

Building a foundation for BI and ERP integration

"We have to understand the process today and what's the process you want to move to," said Jeff Woods, managing vice president of ERP and SCM for Gartner. For instance, "We can't install a system and magically fix master data management [MDM] problems -- we have to understand if there is dysfunctional use of data. Most of the time, there is no formalized governance, and this can be a problem when you're trying to manage data."

To put that into perspective, it helps to distinguish among the kinds of uses of information you have inside your business because the type of analytics you want will drive your implementation direction.

"If your primary use is reporting capability, you still have a lot of flexibility in terms of what kinds of BI you select," Woods explained. "The ERP vendors are trying to bundle these solutions together to make reporting and analytics information easier to access, and this is what we call integrated analytics."

"The value of integrated analytics is there, but it doesn't necessarily represent transformational business value to an enterprise," he said. "Integrated analytics makes it cheaper to access the analytics you want but doesn't really make the analytics any better or more useful. It's really an IT cost-savings you're getting with integrated analytics."

Another kind of more forward-thinking analytics for ERP-focused organizations is embedded analytics.

"This is the use of analytics information inside business processes," Woods said. "For example, let's say I have one piece of inventory and two customers who want it, and I have to make an allocation decision. Today, the way I make that decision, I might look at, for example, a table that lists whether I'm deciding between an A, B or C kind of customer, to see how the customers rank. I've got some crude rules to help me decide.

"But with embedded analytics within the business process, I might ask more sophisticated questions, say things like, 'Which customer is more profitable? Which customer is least likely to defect if I don't give them the inventory?'" he said.

"Today, it's possible to ask those kinds of questions, but I have to drop out of the transactional application, and I've got to have a really dedicated employee to go over to the OLAP system, run the right report and collect the right information, then go back to the transactional system and make the right decision," Woods said. "What embedded analytics does is engineer that analytical information into the business process. And that is transformative value -- tying together analytic systems with traditional ERP transactional systems."

How data warehouses work with BI in ERP environments

BI systems are getting increasingly nimble in how they tap information, and ERP systems already store data somewhere, raising the question: Do BI implementations in ERP environments still need a separate data warehouse?

"I think there is a need for some form and shape of data warehouse in virtually every environment, primarily because data is coming in from multiple places and that data needs to be commingled," said John Hagerty, vice president and research fellow of BI and EPM for AMR Research. "Unless a business has all its data in a single system, you might not need a data warehouse, but I have yet to see a company that has all of [its] data in a single system."

"So you need some sort of data store, but it could be a small subject-matter data mart," Hagerty said, noting that focused data marts are generally less intimidating to smaller organizations and their IT departments. Either way, the core takeaway is that you have to get to a consistent view of the data, whether it's in a common system or in a data mart that's been cleansed ahead of time.

For BI in ERP integration success, think rapid, agile project planning

Building a skyscraper requires a massive set of blueprints that detail every aspect of construction, which may span years. A good BI plan might seem similar, but reality has the tendency to change the basic rules of the game midway through the project. And that happens with skyscrapers, too -- most people just don't see how floors and plans are altered on the fly. Sure, there are ongoing inspections and budget tweaks, and people getting fired or moving to different jobs. Yet somehow, someway, these amazing things get built.

So for project planning and architecting integration, Boris Evelson, principal analyst with Forrester Research, has some recommendations that might make for queasy stomachs for some IT pros.

"Traditional 'waterfall' project methodology does not work well for BI. Agile does," Evelson said. "So favor interactions over documentation. React versus plan. Think quick, tangible prototypes and deliverables versus long-strung-out milestones."

This means face-to-face business participation rather than working with IT liaisons; personal ad hoc interactions instead of highly defined processes; real-time prototypes versus long specification lists; reacting to change instead of planning in advance.

Obviously, Evelson isn't recommending that organizations throw out their blueprints. Rather, if a company can infuse the BI efforts with a sense of quick deliverables and action, they'll be more likely to create a sense of action, acceptance and excitement.


Dig Deeper on Business intelligence best practices

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.