Thursday, November 14, 2024

Business Intelligence Software: 10 Common Mistakes

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

After more than a decade of researching and writing about business intelligence software, data warehousing and performance management implementations, I’m astonished that the same mistakes made in 1999 still occur in 2010. When I polled some distinguished experts to list the biggest mistakes they see when implementing these systems, they too say the same things keep occurring.

Here’s our list of the top 10 BI software mistakes you should avoid, why and how:


1) Lack of executive sponsorship and active business involvement.

Everyone knows that any major IT effort needs executive sponsorship, but in the case of a BI software implementation the big mistake by the CFO, the chief marketing officer or other sponsor is to not be actively involved. It takes frequent injections of business process and strategy savvy to guide the IT team and prevent scope or data creep.

BI projects typically go way over budget when a directionless IT staff isn’t given enough parameters about how much data is enough. They get into the bragging rights trap – my warehouse is bigger than yours. Costs skyrocket, response times lag, results are muddied and the entire project capsizes from its own weight.

Without continuous guidance from the business side, “IT tries to stuff everything into a warehouse to address absolutely any question a user could conceivably ask,” notes Howard Dresner, a former Gartner analyst who coined the phrase business intelligence and author of a new book on performance management on performance management.

“As a result, IT burns tons of cash, takes far too long and creates inordinate complexity. Warehouses needn’t be big, they just need to be useful.”


2) Inadequate scrutiny over the data.

Just having the right extraction-transform-load tool doesn’t make the data correct and current. Poor quality data can destroy the credibility and utilization of data warehouses and business intelligence systems.

This is not an IT challenge but a business challenge. If the key business staff aren’t involved in identifying the right data stores and solving the inconsistencies – how many definitions of customer are in your systems? – the project will fail.

3) Not easy to use.

Too many IT implementers forget that the biggest benefits of a BI software solutions come from widespread deployment. This means the user profile will range from a doctorate in mathematics to an associate degree from the local community college.

The software user interface, graphics and what-if query capabilities have to be intuitive. If the fancy chi-squared distributions are the most prominent tool, you’ll freak out many users. Keep the heavy-duty tools easily available, though – the power users want everything.

4) Poor performance.

User expectations about query response times will be much higher than you realize. If the data warehouse has more than one terabyte or more than 100 heavy users, consider more processor horsepower via a data warehouse appliance. Develop more cubes or other ways to optimize performance now, not afterwards.

A related mistake is not recognizing the enthusiasm factor. A successful system begets huge interest.

“Some organizations are cursed with success and can’t seem to keep up with user demand,” warns Wayne Eckerson, director of research at The Data Warehousing Institute (TDWI), the pre-eminent organization in the BI field for IT pros.

5) Too many or too few BI software tools.

Both Dresner and Erickson warn that IT has to be careful about how many tools are available. Too many tools lead to a lot of confusion and soaring training costs. Too few tools frustrate the users.

Just relying on the tools provided by an ERP vendor may not be the way to go, warns Dresner. Think strategically about the toolset, he recommends.

6) Going it alone.

Ten years ago it was hard to find a lot of BI expertise in specific markets. Now the maturing of the field and plentiful resources make it a crime for any organization to launch a program without thoroughly vetting the process, project, products and people. TDWI membership should be a pre-requisite before moving ahead.

If you’re in a big company, urge the CIO to develop a BI Competency Center. A core group of experts within your organization can become internal consultants to business units. The competency center approach will help avoid a huge number of mistakes and wasted money.

7) Allowing the spreadmart plague to spread.

Eckerson invented the term spreadmart in 2002 as a label for the proliferation of mini-data warehouses and business intelligence systems based on spreadsheets. Typically a department would try to solve a business problem by creating a spreadsheet with lots of macros linked to transaction systems. These spreadmarts were typically undocumented, impossible to audit and extremely fragile. But they’re easy to set up and use, thanks to the ubiquity of Excel.
“Today, spreadmarts are the bane of IT departments who can’t control their proliferation, and the nemesis of CEOs who can’t gain an accurate view of enterprise activity because of them,” notes Eckerson.” In many respects, spreadmarts are the corporate equivalent of terrorists—just as soon as you eliminate one, ten more spreadmarts pop up to take its place.”

Eckerson offers a number of tips on how to combat the problem on the TDWI Web site.

Essentially, IT has to develop and support a superior solution. And the CFO has to use Sarbanes Oxley as a bulldozer to crush as many spreadmarts as possible. Especially maverick systems that directly feed into the P&L.

8) Inflexible design.

Thanks to globalization, an extremely volatile economy and other factors, building a rigid data warehouse and business intelligence system is a sure fire route to misery.

Your business advisors should be probed for insights about what strategies and tactics could change. They should offer odds or likelihood that key parameters will shift, and then IT should consider which parts of the system are most likely to need updating or revision.

9) Ignoring external data.

The best business intelligence and performance management systems incorporate data from external sources. Weather forecasts are obviously an important factor in determining optimum shipping routes, for example.

Mark Graham Brown, a performance management expert and author, says that external factors such as economic, political, regulatory and consumer trends may need to be considered and incorporated into a BI or performance management system to make it truly effective and useful.

10) Wrong customer data.

If customer satisfaction is a key metric for your organization and the IT department is asked to implement a performance management system to create and track this, ignore urges to just use survey data.

As Brown notes, an annual survey won’t be of much help when you need weekly or monthly updates. And few organizations are surveying customers frequently enough to make a meaningful online metric. You need more granularity. He recommends a customer aggravation metric – collect service call or help desk data, and then score the inquiries based on severity.

Sad to say, there are other commonly made mistakes. What do you think are the biggest mistakes you’ve seen? Too many Key Performance Indicators (KPIs) in the performance management system? Maybe we can prevent repeating the same mistakes in the next decade of the technology.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles