7 Recommendations for BI Success
September 9, 2008 | Leave a Comment
One of the biggest reasons a BI initiative fails is lack of ownership by the ‘business’ units of an organization. A recent article in DM Review provides seven recommendations in an attempt to remedy the situation. The article provides a few new and interesting ways to think about this problem, that I like, however, the authors lay it out a little more verbosely than I would like. I’ll summarize the main points below:
The authors do a good job articulating the causes of user satisfaction or dissatisfaction with BI:
- The pace at which users receive answers to questions.
- The degree of relevance between the answer and the question asked. (An answer with little or no relevance to the user’s question is as good as no information at all.)
- The (perception of) reliability of the answer. (Do I trust the answer enough to use it?)
- The cost of information compared to its value. (Users are not interested in answers where the cost exceeds the value of the information.)
The article continues with seven recommendations to resolve these problems, however, I’m not so sure that - in the end - the recommendations squarely address the causes articulated above. Still, the recommendations in their own right are worthy of consideration. For purposes of clarity I will paraphrase the recommendations in my own words:
- Create an organizational unit that is responsible for the design, development, and support of your business intelligence initiative.
- Funding for the BI foundation should come from IT or another central budget source, while funding for individual subject area reporting and analysis should come from the business unit.
- Develop a federated enterprise data warehouse - make sure it is flexible and can grow as the business unit’s needs grow.
- Use technology that helps you manage the BI foundation when data structure changes are needed.
- Use an iterative approach to BI application development - seek frequent user feedback and deploy in days and weeks, not months and years.
- BI tools provided by your enterprise software provider (ERP) are useful for creating an Operational Data Store (ODS), not an Enterprise Data Warehouse (EDW).
- Create a data governance group that is led by your BI Organizational Unit.
I’d be interested to hear what others think about some of these recommendations and the counter arguments they may have.
Forrester Releases Comprehensive, Yet Overwhelming Overview of the World of BI
August 28, 2008 | Leave a Comment
Business Intelligence Architectural Stack from Forrester Research
Business Intelligence (BI) is a no brainer, right? Every organization needs it. Yet there are so few organizations that get it right.
This diagram from Forrester Research (via The Dashboard Spy) serves up a clue as to why.
While, it’s purpose is to provide an overview of the BI landscape, it certainly highlights the complexities in the field. Not only that, but it doesn’t even get into the non-technical side of BI - changes in business process, organizational politics, and transparency - which, are the real challenges of BI.
Still, one should not be daunted by the BI challenges ahead. Sure, that’s easy for us to say, we’ve been doing this for years, but I’ll let you in on a few of the secrets to our success:
- Start small, do not try to do it all - It is better to find one or two key subject areas from which to build, then to develop an enterprise data warehouse from the ground up right from the start.
- Use the technology that you already own first - So many organizations already own incredibly useful BI tools. There is no need to make a major technology investment in tools and technologies as a way to kick start your BI initiative. Start with the tools that came with your enterprise software. Your ERP or CRM system may have provided a means for you to extract data. Your database probably has reporting tools built into it. Use these tools first, and grow into others as your use and knowledge grows.
- Focus on outcomes - If you have not been able to gain traction on your BI initiative, it is almost never the fault of your tools. Usually the fault lies in the lack of alignment between the organization’s strategies and the people and processes in place to execute on those strategies. The tone of your BI endeavor should be outcome based, stress transparency, and serve to align the organization tightly around its mission and values.
In fairness, the Forrester document does not purport to highlight all that is involved in BI. It is clearly intended to be more of a technical overview. Nevertheless, I can’t help but feel that there are a whole segment of leaders that might view a document of this nature and do one of two things:
- Go out and spend several hundred thousands (or even millions) of dollars on a new BI toolset.
- Nothing.
Let neither of these options be acceptable and if you follow these three simple principles, you will be better off in the long run.
Data Visualizations: Will the big BI vendors catch up to the niche players?
June 5, 2008 | Leave a Comment
The New York Times has really been stepping up its infographic department in recent years. Take a look at this data visualization showing how different demographics voted in the recent democratic primary contest.
All politics aside, I’m most interested in the fact that although this was more than likely developed by an information or graphic designer, it looks like it could have been produced by a BI tool. Of course a tool is just a tool without the data organized in such a way that makes the tool useful (but that is a subject for another post).
What fascinates me is how far the major BI tool providers are from producing these types of visualizations. It seems to me that they are so focused on the integration issues with all of the consolidation in the industry that there isn’t enough time to add the features that decision makers want - new and exciting ways to communicate information that helps them take action.
If you’re interested in data visualization, I highlighted a few interesting developments in the field in my recent column in Campus Technology Magazine.
It seems to me that that the quality gap between the data visualizations produced by tools and those produced by designers is getting smaller, but it is the niche players in the business that are making this happen. I’d love to be proven wrong though.
Thoughts?
Write a book supported by econometric models in minutes
April 15, 2008 | Leave a Comment
This one goes in the category of ‘it has to be seen to be believed.’ Phillip M. Parker of Websters-Online-Dictionary.org has been experimenting with the use of computers to automate content development. This video shows how he has been able to produce books and reports fully backed by data and econometric models.
Imagine, being able to write your next report, business case, or economic study in less time than it takes for one to read it. Watch the video to see how it is done.
Moving Beyond BI Shadow Systems
February 16, 2008 | Leave a Comment
A recent post on The Data Doghouse, a Business Intelligence blog, explains that data shadow systems are still extremely prevalent across industries and will continue to be so for the foreseeable future. Data shadow systems are groups of spreadsheets and local, customized databases - often built in Microsoft Access or Excel - created by business groups to provide data for their users. An excerpt from the article shows just how prevalent these shadow systems are:
In a survey we conducted in the fall of 2007 we found that the medium number of data shadow systems at an enterprise was 30. According to the survey, data shadow systems are prevalent in all industries, companies of all sizes and throughout various business functions in the enterprise.
The presence of shadow systems is a trend we’ve seen with our own clients in education, government, banking, and insurance. For many of our clients, different shadow systems are providing business analysts with the data they need to answer important business questions, recognize trends, and make informed decisions. Obviously, these shadow systems are providing a valuable service. However, despite these benefits there is a downside to operating shadow systems. Let’s take a look at some of the negative aspects of shadow systems.
First, it is often time consuming to implement and maintain data shadow systems. Data shadow systems are typically cobbled together by business users. Let’s face it, designing these systems is not the strength of business users (this is something that is bettered handled by the IT department). This leads to a design framework that requires lots of labor-intensive maintenance activities and a lack of felxibility and extensibility.
Second and most importantly, these shadow systems do not enforce rigorous data management techniques and principles. For example, consistent business rules may not be used across the organization to clean and codify the same data. The result, data quality and consistency suffer and it is difficult to know how confident decision-makers can be in the numbers being reported.
So the question is, what can be done to move beyond the use of shadow systems in your organization. The best approach is to engage the IT department and use the organization’s existing Business Intelligence (BI) tools to tackle this problem. BI tools should be used to centralize data, implement consistent and documented business rules to clean data, and develop reports that meet the needs of business users and stakeholders.
6 Best Practices for Successful Business Intelligence
February 11, 2008 | Leave a Comment
Business intelligence (BI) is not about technology. No doubt there is much technology involved, but a sound business intelligence strategy concentrates more on methods. The outcome of intelligence gleaned from a strategic reporting or decision support system should be an action or decision. The decisions made from business intelligence will likely lead to changes made in strategy and/or individual business process.
These changes in strategy and business processes will necessitate changes to one’s enterprise resource planning (ERP) or other transactional systems. For example, new business rules or codes may need to be added to the ERP in order to operationalize a decision that was made. This will require an understanding of the business rules engine of the ERP as well as the implications from a historical measurement perspective of changing or adding codes to the system. People will need to be trained and constituents may need to be informed of new rules.
BI is much more about organizational alignment or people and processes around a common set of strategies and goals then it is about technology. To that end, follow these 6 best practices to move your organization from silo-based planning to one that is aligned around a culture of evidence:
- Define areas for exploration - What subject areas need to be studied? Not all can be effectively studied at one time - not at the start - therefore, you will need to prioritize. The organizations leaders will need to set the priorities based on the key strategies that need to be affected.
- Articulate problem statements - Now you have identified the subject areas to be studied. What are the problems in that area? Simple year-over-year trend analysis will often highlight where the problems are lurking. The problem should be stated as follows: [Subject Area] is down by 15% compared to last year.
- Identify causal factors - Perhaps one of the most overlooked steps in the process. Your problem statements only tell you what is happening. It is critical that you find out why it is happening. Statistical models need to be employed to determine the key drivers influencing the problem area. Identification of the key drivers in the area will help you isolate the problem and determine the factors causing the problem.
- Determine corrective action - Once the causal factors have been identified decisions can be made and corrective action taken. True evidence-based decision making.
- Align people, process, and technology - Decisions inevitably lead to change. Most often the change comes in the form of a new business process. The new business process will need to be codified in the ERP system and people may need to be reorganized and/or retrained. This will be the hardest step in the process toward a culture of evidence, yet it is also the most critical.
- Measure outcomes - Now that people, process, and technology has been aligned to solve the problem, you must measure the effectiveness of this action. Be careful that you allow a sufficient amount of time for the impact of the change. In fact, decision makers should agree about the length of time they will permit for the action to take hold and pre-determine a point in time for re-evaluation.
As you can see from the diagram above, the process is cyclical. As decisions are made and corrective action is taken, key drivers will change. This will constantly cause the organization to reevaluate and revisit its strategies and tactics over the course of time.
Colleges Take the ‘Business Course’
February 2, 2008 | Leave a Comment
In a recent article in the Chicago Tribune, Ann Meyer explains a growing trend in higher education. More and more institutions are facing an increasingly competitive landscape and are using business principles in their marketing and operations to attract students, improve their financial position, and maintain a competitive position in the marketplace. The importance of embracing business principles is summarized nicely in a quote from the article:
“The business side of all of higher education is increasingly important,” said Ronald Ehrenberg, director of Cornell University’s Higher Education Research Institute. Heavily dependent on tuition, many independent colleges are, as Ehrenberg puts it, “on the financial bubble.” They need to pay attention to larger trends and react by carving out new niches, he said.
The article highlights the accomplishments of two small independent colleges in Illinois, Elmhurst College and North Point University. Led by entrepreneurial college presidents, these institutions used marketing, communications, and sound business decisions to create new institutional identities and reposition themselves in the competitive marketplace. The results are impressive - increased applications, higher enrollment, revenue growth, and surging endowments.
The institutional changes made by these two institutions were no doubt made using one of the most powerful tools available in higher ed, data. The question is, how does an institution leverage its data to make the right decisions. The answer, by employing the right analytic techniques and business intelligence tools to identify improvement opportunities and measure success against strategic goals.
Let’s take an example from the article. Elmhurst College has been able to attract students with higher test scores and grades, and in the process boost its selectivity. How is this done? One way is to utilize business intelligence tools to closely monitor the institutional funnel (i.e., number or prospects, applicants, offers, and enrolled). At the same time, selectivity and recruiting goals must be set and progress should be tracked and reported annually against these goals. Higher powered analytic techniques can also support these initiatives. For instance, predictive models can be used to identify quality prospects that are likely to accept offers of admission.
This is just one example of how business intelligence and data are used to support strategic decisions. Businesses have been leveraging these tools for some time and now colleges are relying on them to compete in this increasingly competitive market.
Defining business rules in the business’ language
January 11, 2008 | Leave a Comment
One of the biggest problems with any business intelligence project is for all involved to fully understand the data model and it’s relationship to the business of the organization. Without an excellent business/data analyst, much will be lost in the translation.An article in DM Review addresses this very issue, however, largely from the technical person’s point of view. Since this blog is primarily aimed at business leaders and decision makers (information consumers) that require business intelligence to make informed decisions, I won’t go into gory detail summarizing the article.
The most useful tool mentioned in the article was this spreadsheet (pictured) to help put business language around a logical data model.
What’s a logical data model?
[Silence fills the air.]
…Exactly.
As a business decision maker you don’t really need to worry about it so much, however, your analysts and technical staff developing the business intelligence architecture sure do.
Essentially it is the relationship between and amongst the tables that represent the business data.
Without a rock solid understanding of the ‘business rules’ that describe the data, the business intelligence architecture will fail.
This tool is as good as any I’ve seen to help business decisions use their own language to precisely describe the relationships in the data model.
Think of this tool as the Rosetta Stone of BI. Okay. Never mind. I’m sure I’m overstating it. But it is useful nonetheless.
Barriers to Institutional Effectiveness Research and the Need for Self-Service Analytics
January 7, 2008 | Leave a Comment
In a recent article published on the Achieving the Dream website by Vanessa Smith Morest and Davis Jenkins titled Institutional Research and the Culture of Evidence at Community Colleges, the authors point out several important barriers to high quality institutional effectiveness research in community colleges. These barriers include:
- Lack of dedicated IR personnel. IR personnel may not have the training for quantitative analysis.
- “Data analysis” is typically “Data collection” for compliance and accreditation. These data are typically not very useful in college management.
- Data contained in the student information system may be “dirty”, and are designed to support administrative functions, not research.
- Skepticism with college presidents about using quantitative methods to manage the institution.
I’ll add that responsibility for data entry, warehousing, and reporting are often scattered among several offices making coordination of a research agenda difficult at best. In addition, faculty are often not engaged in the management aspects of the institution. This may, in part, be due to a lack of timely and accurate information. In recent years, however, the proliferation of business intelligence (BI) and analysis tools (e.g., Business Objects, SAS, SPSS, Tableau, etc.) allow information consumers to quickly access analysis in a “self-service” environment.
Most BI software products have matured to the point where creating reports and even running sophisticated statistical analysis is point and click. This BI software ease of use helps to make the most of the limited staff dedicated to IR. In addition, the self-service nature of BI software helps to extend human resources by making data and analysis more broadly available to administrators, faculty, and staff. There certainly exists the potential to increase the number of research topics as well, since faculty may utilize these tools for investigation of success rates or retention of their own students. Instead of having one or two IR staff stretched to analyze a single topic over several months, there may be dozens of concurrent analyses by curious faculty.
This helps to address items #1 and #2 on the barriers list.
A complication for most institutions right now is the lack of standard sources for data to answer questions. The lack of data access usually encourages the proliferation of “shadow systems”. These shadow systems in turn lead to multiple and inconsistent reports for topics like headcounts, retention, or success rates. It’s not always the case that the student information system (SIS) is at fault. Most of the data needed to answer questions or to conduct analysis is contained in the SIS. Information consumers usually don’t know how to get to data they need. This is where self-service plays an important role.
By implementing BI tools, information consumers can typically make sense of what should be the “authoritative data source”, the SIS. Surfacing data contained in the SIS will often illustrate where data quality problems are, and point to deficiencies in business processes. Most data quality issues in higher education are the result of breakdowns in processes, in other words: it’s usually not a technology deficiency. This helps to address item #3 on the barriers list.
As for item #4, and the coordination across multiple offices, governance structure and commitment to a “culture of evidence” are usually necessary at the highest levels. Providing self-service to analytics across an institution using BI tools is one way to foster this culture of evidence.
Higher Education institutions should incorporate learning systems into the business intelligence framework
December 28, 2007 | Leave a Comment
The Sloan Consortium, an association committed to the advancement of online learning, released the results of a a survey it commissioned on the growth of online learning in higher education. It should be of no surprise to learn that online learning continues to grow as a mode for learning, although the growth rate from 2006 to 2007 did decline slightly when compared with 2005 to 2006. Still, nearly 20% of the nations 17.6 million students are enrolled in at least one fully online course.
Of particular note is the fact that nearly 54 % of institutions are either somewhat or fully engaged in offering online learning as a key part of their strategy to deliver education.
As online learning becomes a more strategic component of the institution’s success, so to does the data within the learning management system (LMS) compared with other institutional systems. Unfortunately, too often the LMS is left to be an island of data unto itself and is not usually integrated with the administrative or ERP systems.
This can be detrimental to decision making as it is not always possible to form a 360 degree view of student activity and learning assessment without the data from both the LMS and the ERP.
While institutions have increasingly showed a willingness to interface these two systems at a transactional level, it is time that institutions consider integrating the data from these two systems to create a fully informed set of decision support tools. Increasingly accrediting agencies are asking for information on student engagement and learning outcomes. These are two areas of measurement that must come from a combined view of the data.
In five to ten years this problem may be easier to solve as the systems are likely to become one. Until that time, one should consider a business intelligence strategy that combines data from these disparate systems to create fully informed profiles of students, their engagement levels, and learning outcomes.