Questions and Answers: The Trickle-Down Theory in Business Intelligence
With pervasive BI growing, and barriers to implementation and use dropping, where is BI headed?
- By Linda L. Briggs
- October 15, 2008
As business intelligence moves deeper into the enterprise, it is also becoming more pervasive at small, midsize, and large companies alike. As barriers to BI drop, companies of all sizes are looking at business intelligence in new ways.
We spoke with Chris Rafter, vice president of consulting services at Logicalis Integration Solutions, about trends in BI, and asked what BI questions clients are asking him these days -- as well as the questions he wishes they were asking. Logicalis provides technology solutions to over 6,500 clients worldwide. Rafter is a long-time expert on information and application strategies in enterprise computing environments, and has applied his deep expertise in both business and financial issues to Fortune 500 clients for many years.
BI This Week: You've worked with all sizes of companies for years on enterprise BI systems. What are a couple of the strongest trends you're seeing in the market right now?
Chris Rafter: The ones we see most often are the growth of "operational BI" and the movement of BI into smaller and mid-market companies. I believe these two trends are linked somewhat. BI is no longer just for Fortune 500 companies. Many managers with large-company experience are joining these smaller firms and applying what they have learned about business performance management (BPM), reporting, and analytics.
It sounds like smaller and mid-market companies are increasingly using BI more effectively. In what ways in particular?
We find there are far fewer barriers to information sharing and delivery in smaller companies, which means there is much more tactical use of BI to help make day-to-day decisions. This makes small to mid-size companies more agile, more responsive to their customers, and allows them to stay competitive with much larger companies that cannot move as quickly. Operational BI, as it is sometimes called, acknowledges that there are important, business-impacting decisions being made every day at all levels in the organization -- not just in the corporate boardroom.
How is real-time BI being used to drive lower level and tactical decisions?
It starts with something called "Decision Impact Analysis," which is an introspective look at what decisions get made across the organization every day and what impact those decisions can have on customers and on that company's business.
We regularly find decisions being made daily within companies that have very strong and far-reaching impacts in areas such as profitability, risk, and customer satisfaction. These decisions are being made by salespeople, customer service managers, and finance managers. These decisions have consequences ranging from whether a customer is retained or lost to major issues around procurement costs, product profitability, and other areas that might expose the company to risk. Imagine these decisions being made 100 or 1,000 times per day in the average company and then you can see why people are starting to look at this.
We identify these categories of decisions and rank them by impact. For those with the greatest impact, we ask questions such as "How could this decision-making be improved?" Very often, the problem is not with the judgment or skill of the decision-maker, but rather with the information he or she has access to - and just as important, with the timeliness and "freshness" of that information. That's where BI comes in.
You mentioned predictive analytics. How important is that becoming to the clients you work with?
Predictive analytics applications and models are definitely out there. The math behind them has been around for a very long time. We've used them ourselves for some of our techniques for master data management (MDM), matching and merging of customer records to eliminate duplicates, migrating from deterministic to probabilistic matching, and other similar tasks.
There are some great tools that perform forecasting and inventory management. The models take a bit of time to "tune" up front, but once they're working, they can be almost frighteningly accurate. The companies getting the most out of predictive analytics are those who aren't afraid to take a little risk and try something new. They also need to select areas of their business where there's a high degree of payoff to counteract the risk.
What kinds of questions are customers asking you these days about their BI and data warehousing issues?
The most common questions we get are around products:
- What's the best BI platform?
- What's the best database for a data mart?
What I wish people would ask us is:
- How much should I spend on BI to make sure I get value from my investment?
- What is my business case for BI? Who is it going to help most in my organization?
- What do I need to do to get my company ready for data warehousing and BI so that its introduction is successful?
I've never seen a BI rollout where the users hated the BI platform. There are many great platforms and tools out there, at every price point. I like to see organizations spending time looking inwardly at how they're going to use the BI product, rather than getting caught up in product features and marketing hype. Think of BI applications as creative canvases.
Do concerns differ based on whether the company is Fortune-500 size or midsize and smaller?
There is definitely a difference in the tone of discussion between the Fortune-500 size companies and smaller-to-midsize companies. The Fortune 500 have a much larger population of data to manage, so there's more interest in data governance, MDM, and data retention and lifecycle. This complexity also prompts them to ask about things such as metadata stores and data dictionaries.
The other area large enterprises ask us about is managing increasing volume demands [as it affects] both system performance and data storage. Storage is not getting any cheaper, and I've seen far too many large-enterprise data warehouses where performance was seriously lacking. It's too late when it starts affecting users.
By contrast, smaller firms are more about just "getting it done." They're dealing with much smaller warehouses of information, so there is less of a concern about data quality and governance. We also find that the business users in the smaller organizations get more involved in BI and take more of a hands-on approach. We've had line-of-business managers come to requirements meetings with hand-drawn mock-ups of dashboard screen layouts they'd like to see built. They are often very specific and vocal about exactly what it is they want to see.
Do you think companies are seeing a true return on investment on their BI investments yet?
Yes, they definitely are, but not in all cases. The problems we see arise when companies overspend on BI. When it comes to an ROI for business intelligence, companies need to realize that the "return" part of the equation is not infinite, but the "investment" part can certainly seem like it sometimes. You have to strike a balance. You can't spend $3 million on a BI platform and expect a three-month payback.
The smartest companies are calculating their expected benefits and savings before they go shopping for tools. It's a lot harder to overspend when you set a budget up front and are realistic about how much benefit you're reasonably going to derive. All too often, the process is reversed and people price out an expensive platform, then go off searching the enterprise looking for a way to justify paying for it.
There's no shame in starting small. Some companies are better off with a smaller data mart with 60 reports that generates positive returns for their business than a massive enterprise data warehouse that's bleeding it dry.