DM Direct: Online Newsletter for Business = Intelligence & Data Warehousing Professionals

***** ISSUE : OCTOBER 8, 2002 *****

Back=20 to Issue=20

=20

Business intelligence (BI) solutions have become an important = and=20 integrated function in many companies. However, when measuring the = number=20 of users having the BI software compared to the ones actually = using it,=20 many BI solutions are not as successful as one might believe. Even = if the=20 solutions may be technically efficient =96 delivering the right = information=20 at the right time =96 the number of users actually using these = applications=20 may be low. It can be estimated that only 25 to 30 percent of the = licenses=20 bought by a company for a large deployment to business users are = actually=20 used. This low percentage should ring an alarm to many companies = as the=20 actual business usage of an application may be the most = significant=20 criterion when measuring the success of an implemented solution = (as=20 opposed to project management criteria such as on- time delivery = or on=20 budget).

The reasons for a low usage per purchased licenses are = mainly:

Poor management in implementing the tools.=20
Poor data quality, leading to nonreliable reports or = analyses.=20
Political resistance (never to be underestimated when = implementing=20 BI solutions, as few projects where information and, thereby,=20 empowerment are given to more people can cause such political = upheaval).=20
Poor user education.=20
Poor follow up after implementation.

Also, it ought to be clear that many people are not as computer = literate as they need to be in order to successfully work with = most BI=20 solutions =96 no matter how easy the systems are compared to other = IT=20 solutions. As a benchmark, if the users are unfamiliar with the = usage of=20 spreadsheets, the regular training program proposed by the BI = vendors does=20 not include the time to learn these tools. It follows that if the = users=20 find a BI solution difficult, the resulting reports or analyses = will=20 certainly be less than optimal.

Data visualization (DV) is an area that has progressed rapidly = in the=20 last years. It is a solution that may be easily used by = decision-makers at=20 all levels when analyzing data. Some BI vendors are currently = offering=20 such features with their traditional reporting and analysis = solutions.=20 Apart from being easy to use, DV simplifies the discovery of = exceptional=20 events, potential problems and opportunities in a way that many BI = applications do not.

BI solutions have, by tradition, primarily focused on analyzing = sales=20 performance, followed by solutions for other subject areas in an=20 organization (e.g., production, finance, human resources). DV = solutions,=20 on the other hand, are more often used in research, especially in = life=20 sciences, where a large number of data transactions need to be = understood.=20 DV is also becoming more and more popular for so called cockpits = or=20 dashboards, where management rapidly gets a visual overview of = certain key=20 performance indicators presented in charts. E- business is yet = another=20 area where DV is growing, very often in order to analyze customer = profiles=20 and their navigation between Web pages on a Web site. Presently, = the usage=20 of DV applications is rare for sales applications (while = traditional BI=20 solutions are strong); however, they are very effective for sales=20 analyses.

Tips, Tricks and Advantages with Data Visualization

A number of observations about using DV for analyzing sales = follows.=20 These observations will facilitate the usage of DV applications = for people=20 that may not feel at ease with traditional BI solutions but, = nevertheless,=20 need to analyze information. Even though some DV analyses can get = highly=20 advanced, one does not need to be an expert on Rubik's cube and 3- = dimensional thinking in order to take advantage of DV = solutions.

Ease of Use

If a decision support solution is to be deployed to a large = number of=20 business users, it is important that it is easy to understand and = to work=20 with. It must also improve the results of the work done by the = user. Even=20 though DV usually allows for very complex presentations, showing = many=20 different data dimensions, simple charts can highlight interesting = points.=20 For example, most people understand pie charts and bar charts, and = even=20 these basic visualizations will help increase the user=92s = understanding his=20 or her sales activities. If negative segments exist, bar charts = can handle=20 this nicely (see Figure 1). Pie charts cannot show negative and = positive=20 valued segments together. The next recommended step would be line = charts,=20 which are excellent in analyzing trends over time.

Figure=20 1: A Bar Chart Showing Positive and Negative Values for a Sales=20 Analysis

Standardize Definitions

As always when presenting information, it is important to agree = on the=20 definitions of the data that is used, e.g., the exact definition = of=20 revenue. This matter becomes even more obvious when presenting = figures in=20 visualizations. Just as pictures are worth more than a thousand = words,=20 they also allow for personal interpretation. Seemingly simple = questions=20 can change the interpretation completely. One example is the = distinction=20 between the number of sales per region and the amount of sales per = region.=20 A region responsible for most number of sales can have the = smallest=20 revenues, if the sales are all small. Therefore, the definitions = of the=20 data represented must be clear and agreed upon.

Detail is Not Forgotten

Most DV applications are able to generate the detail behind the = visualizations, as is shown in Figure 2. This corresponds to the = common=20 need to get to the detail level in order to better understand a = chart,=20 especially if it shows something extraordinary. A connection = between a=20 diagram and the underlying detail also means that DV is not = entirely an=20 analytical application. When the detail is accessed, perhaps on a = level=20 where each sales contract or customer is presented, the solution = turns=20 into something that is directly linked to operational activities = or a=20 customer relationship management (CRM) solution.

Figure=20 2: Underlying Detailed Data in a Diagram

Visual Analysis and Less Reporting

Traditionally, the majority of analyses or reports done with BI = tools=20 are presented in classical tables or spreadsheets, showing exact = figures=20 for a certain category of data. The usage of graphs in classical = BI usage=20 makes up for a small part of presented analyses, even though they = are very=20 appreciated at high-level management meetings. (And certainly by = BI=20 vendors when they present their products.) The strength in DV is = its=20 capability to let the users apply graphs to a much larger extent = (than BI)=20 when analyzing data. DV is not well adapted for classical = reporting, using=20 tables or spreadsheets with grouped and summarized information. It = is,=20 however, very strong in performing the analytics, surpassing the=20 analytical capabilities of report-based systems. In practical = terms, this=20 means that DV makes it much easier to graphically discover = interesting=20 phenomenon in the data under analysis, than most traditional BI=20 solutions.

Avoid Information Overload

Be careful not to put too much information in a DV application. = This=20 sounds illogical for BI applications, where the idea is to give = the user=20 the freedom to ask for the information without limiting the data=20 accessible. In the end, most users will only work with a limited = number of=20 indicators. In particular, balanced scorecard applications have = shown that=20 too many key indicators will make the application less useful than = with=20 fewer indicators. This also holds true for DV applications. This = point is=20 to consider how most users work with data. Figure 3, shows a = complex=20 visualization example, adding another two dimensions represented = with=20 different shapes and colors. This analysis may certainly be = powerful and=20 useful to the accustomed analyst, but difficult to interpret for = the=20 average business user. DV can get pretty advanced, and users can = always=20 add more complexity.

Figure=20 3: A 3-D Data Visualization Example

Automate the Discovery of Important Phenomenon

In report-based BI solutions, the user relies on his experience = and=20 intuition when searching for interesting information. With DV=20 applications, sometimes very complex relations in data are = discovered in a=20 more automated fashion as visualization by its very nature allows = for an=20 overview that tables cannot give. The ability to present important = phenomenon in large and complex data sets is certainly one of the = major=20 strengths of DV. Figure 4 shows an example of DV where exceptions = are=20 easily discovered.

Figure=20 4: Exceptional Business Transactions in February

Capitalize on Interactivity

Usually, most users are not able to clearly define what they = want from=20 a new kind of solution. It is often difficult to imagine the = practicality=20 of a product presented and explained in just a few hours. As DV=20 applications are interactive, they allow for adaptations to the = users=92=20 needs. These needs are very likely to change once the application = has been=20 presented and will continue to do so during its life cycle. As a = rule of=20 thumb, if no demands for modifications in a DV solution =96 or a = BI=20 application =96 are made within the first three months after its = roll out,=20 this is usually a sign that no one is using the solution (not that = the=20 solution is perfect and, therefore, the users do not ask for any=20 modifications!). The fact that DV is flexible to users=92 needs, = increases=20 the potential of success when implementing such solutions.

Return on Investment

The return on investments for DV solutions is, as with other BI = solutions, difficult to estimate. It is easy to calculate the = costs, but=20 the potential increase in revenue related to improved decision = making is=20 not an exact science. However, one aspect that can be relatively = easy to=20 calculate is the time saved by the user. Graphs produced with = classical=20 spreadsheet programs that may take 10 to 20 minutes to do can = literally be=20 done in seconds with DV tools. Often, this time saving will assure = a=20 return on the investment.

Standardize Graphical Representations

DV means the visualization of data. This certainly includes the = usage=20 of colors, something that experience has shown can become part of=20 standardization within a company. The color red, for example, may = only be=20 used when highlighting potential problems, and green is used to = highlight=20 favorable things. In such a situation, these two colors are not to = be used=20 when giving colors to products, sales regions, etc. This way of = working=20 with certain reserved colors facilitates further the = interpretation of the=20 DV analyses.

Conclusion

These examples allow the business user to benefit from = analytical=20 applications in the form of DV without needing to know how to use=20 classical spreadsheet solutions. After all, what counts is not how = advanced a solution is, but whether it is used or not. It will be = used if=20 it is easy and adds value to the results of the work done.

Many analysts predicted that managers on every level would be = using BI=20 solutions. Even though the number is constantly on the rise, it is = not=20 near some of the forecasts. Today's BI applications may be = powerful and=20 relatively simple to use but, in most cases, only used by the = power users.=20 Other managers do not have the time to do the analyses themselves = because=20 they don=92t know what to look for. DV solutions effectively = highlight=20 exceptions, and the manager can spend more time treating a problem = or an=20 opportunity, and less time trying to discover or define it.

DV applications exist today as outlined in this article, but = they are=20 still rare in the sales operations area. Implementation of DV = solutions=20 offers companies the opportunity to take the lead in improving = their=20 analyses, moving them ahead of the competition. The bottom line is = that it=20 is necessary to understand your business, find and define = potential=20 problems or opportunities and improve the decision-making process = as a=20 result. DV applications are easy to use and understand, can be = deployed by=20 a large numbers of users and are very effective in depicting = business=20 activity at a glance.

Gabriel Fuchs is the senior manager at La Suisse Insurance = Company in=20 Lausanne, Switzerland and founder of the business intelligence = department.=20 He is presently involved in analyzing sales activities using data=20 visualization and visual geomarketing solutions. You can reach him = at sgfuchs@bluewin.ch. =

^=20 top

About DM Direct
DM Direct is a weekly, HTML-based = e-mail=20 newsletter distributed to more than 26,000 subscribers in = the IT=20 industry. The editorial focus of DM Direct parallels that of DM = Review magazine, and the featured articles are only available = in the=20 newsletter. Past Issues:	To subscribe: Send an email to DMDirect@tfn.com and specify = 'DM=20 Direct' in the subject line. To unsubscribe: Send an e-mail = with=20 "SIGNOFF DMDIRECT" in the first line of the message body to mailto:%20RemovesDMDirect@tfn.= com

	DM DIRECT=20 BI SPECIAL REPORT *Strategic Solutions for Business=20 Intelligence*