Examples of effective dashboards SAP BI

If we look at one of the Web items we worked on in a former we can start to really work through some of the ideas and concepts discussed in this chapter and see if we are following all of the important points while building a dynamic and focused dashboard. So, in the following illustration we can see the column charts that we used in the previous section and let’s discuss what is going on within the chart. Now, the chart itself is in pretty good shape so let’s also assume that we have agreed upon the KPIs so that our focus is on reviewing the finer points of the display and presentation.

As a reminder, this illustration is a comparison of the three different column templates that we worked on. Again, because the illustration is in grayscale, we can’t really discuss the colors. Suffice it to say that one of the items you would evaluate is the color scheme for the columns, to make sure that it’s effective.

Now there’s a distinct difference between Chart 1 (far left) and Chart 2 (middle) in terms of presentation of the information. In your mind which approach will allow you to review this information easier and faster so that you can get back to actually running the business based on this analysis? From the two charts we can see that they differ based on the groups of information that are presented. Chart 1 has the groups based on the Product Groups and Chart 2 has the groups based on the Yearly Sales Volume. Same information in both charts but very different approaches to the analysis processes. I personally tend toward the formatting in Chart 1 since I can immediately see the information, understand the results, and based on the interpretation go out and execute a plan. I know that from year to year my sales have gone up for each of the Product Groups. Very straightforward, very direct and gets the information to the business user directly with very little effort on the business user’s part to understand the formatting used in this dashboard. In terms of Chart 2 the grouping in Sales Volume is nice but what does that tell us and is it intuitive enough for our purposes? We can’t immediately see that the sales from year to year went higher, we also can’t tell or are we expected to realize that in each case the sales for each Product Group increased. We also can’t really identify from the chart that the overall year-to-year sales volume has increased either. We can attempt to interpret the results but that means opening the door to interpretation and that’s not a good thing when it comes to charts that you may find yourself explaining to the corporate world.

Even if we turn the tables a bit and suggest that this dashboard is looking to compare sales in the different Product Groups, will this allow us to suggest that Chart 2 is better at explaining the results than Chart 1? I guess we can give a slight edge to Chart 2 since the Product Groups are together rather than apart but that would still not give a decisive edge to Chart 2 versus Chart 1. Looking at Chart 1 we can still see that comparison very clearly and be able to identify the comparison between Product Groups very quickly and accurately.So if the appropriate formatting and positioning is completed you will see that it can stand up to a number of views of the analysis.

The overall display and sizing is good for both charts. An initial suggestion would be to adjust the Y-axis. The grid goes from $0.00 to$3,000,000.00, but each Product Group has well over $1,000,000 in sales. The Y-axis could be scaled so that it starts at$1,000,000, which would simplify the charts. Another suggestion would be to move the “Product Group” title for Chart 2 so that the business user doesn’t look down from the chart and wonder where the columns are for Product Group.

Now, consider the rightmost chart—Chart 3. It’s format of the data is good and the positioning of the information is very similar to Chart 1 but the titles are a bit either nonexistent or scarce. So we don’t have any titles up the sides or at the top. You can probably see that this is one dimensional versus the others that are two dimensional and that’s OK even though some of the “wow” factor goes away without some tweaking of the colors but I’m not going to say that this is a bad chart template based on the fact that it’s only one dimensional. Remember, the focus of these Web templates is the delivery of the information in a consistent and focused approach, not how many dimensions can we use to get the information displayed. So, all in all, we can see that with the basics that we talked about in the previous chapters we can build a useable and consistent Web item for a dashboard.

Before we get into the configuration details of a good dashboard, take a look at the example dashboard shown next, which clearly incorporates all the concepts discussed thus far. This dashboard was built using the SAP BOBJ component Xcelsius and sources the data directly from the BW system using BW queries and BOBJ Universes. Xcelsius has the capability to set up dashboards with every Widget imaginable in terms of graphics, such as tachometers, aircraft cockpits, and off-the-wall graphics, but here the use of the basic column and line chart types with some actual summary reports thrown in looks great.

Taking a closer look at this dashboard, we see that it has very focused metrics. It presents volume and pricing data for housing sales trends. The space has been used very well. For example, the check boxes for navigation on the line chart type have been incorporated into the display of the chart type. The titles are clear and direct and don’t take up a lot of space; Housing Sales Trends comprises only about 5 percent of the total space available.

There is some scrolling in the lower-left report but that is very limited since in this case we are scrolling only 12 months at a time and therefore it will always be the same level of scrolling, which would not be the case if the rows were controlled by something like types of homes where the total number of rows might expand and contract. The yearly average sale price report is well positioned and consistent. We see that the metrics have been structured to between four and six depending on whether you define the use of years as another metric.

This dashboard is direct and focused—Monthly Sales Price and Yearly Average Sales Price with additional historic perspective based on information from 2005 through 2007, then details by Month for 2008. Again, in this case the color is used to identify the primary metrics—top two charts and then it draws your eyes to the lower two reports. I would have used a different background color so that the bottom two reports are not as minimized as they actually are due to the darker black against lighter black background.

The coloring that is going on is very basic, no highlights or fading activities and the scaling is clear and consistent. I believe that there should also be a third button at the top of this dashboard—Price, Volume, and Price/Volume Analysis since we can gather much more information if we integrated these two sets of metrics into an overall analysis but that would have depended on the report requirements and this would be second-guessing the development requirements. In that case, I would have offered an additional prototype dashboard with the additional button to draw the business user’s attention to this possibility. Since we have about 4–5 metrics on each of the screens it would be reasonable to have an additional button for a total of 12–15 metrics for this dashboard. You can also see that there are only four boxes in front of the business analysts. This immediately gets the analyst what they need with no assumptions or guessing required by the analyst. This is a good, solid, well-designed dashboard and takes into account the needs of the business analyst and also of the BI system.

Now, let’s take the lessons learned from the previous examples and apply them to another dashboard and dig into the details of the actual architecture in the WAD in the process. Remember, you first need to have your sources of data correctly formatted and ready to go for the different chart types that you will be presenting. So, you should have already decided on the chart types, reached consensus on the KPIs, and gotten the data sourced appropriately defined to make the configuration process easier and more focused on the design and display than on the data. Nothing is as frustrating as developing the entire dashboard and then executing it only to find that the information that you expected to see is not there. In this case, I’m going to adjust the format to scale into the thousands to help clear some space on the screen next to the Y-axis. The next illustration shows the same chart and chart type with different scaling. You can see that it’s a bit clearer with the scaling turned on, but remember that you have to identify the change in the scaling factor somewhere in your titles. Also, remember that scaling might affect other charts in your dashboard, but as long as all the amounts are going to be scalable, then making the adjustment to the display is safe.

So, for this example I have a series of metrics that we will apply to a dashboard and integrate as many of the concepts we discussed as possible. My metrics are going to be Actual Sales, Cost of Goods, Incoming Orders, and Planned Sales, with a comparison between 2007 and 2008 information with a calculated value showing the Average Price/Unit and the Average Cost/Unit. In addition I’ll be applying several parameters to each of the charts. Initially I will scale everything to the thousands as we did in the previous example and also turn on the Swap Display Axis parameter for each chart type on the Web Item Parameter list. This can be seen in the illustration.

For each metric we will use suitable chart types but again the one thing that we will have a difficult time showing here is the color scheme. Just to say that in each case we have different colors for each of the series in the chart types. I will adjust the font level for the primary metric in the chart but I could have also changed the color to emphasize the primary metric. This dashboard will be straightforward in nature and what I mean by that is we will not focus on additional items within the dashboard such as variables, text, information, hyperlinks, etc., but look to generate a good, solid, consistent dashboard and leave the bells and whistles for further discussions within your company. These are parameters that should be identified in the requirements document and then prototyped for the business users.

This illustration is the finished product for our dashboard discussion. As you well know, there are many components to the development of a dashboard and what I’ve done here is to develop a dashboard that will tell the appropriate story and a bit more just so that the business analyst starts to ask additional questions and possibly decides that there are additional KPIs that they need over time and rather than these specific KPIs shown here, they may require adjusted information over time.

This dashboard is titled “Sales Analysis by Product Group” and for this example we will accept that the default information around the required KPIs for this dashboard is accurate. As you can see we have six KPIs with two key figures per view and a total of six views. If I do the mathematics correctly we are looking at 12 KPIs for the business analyst to assimilate at one time. This is a reasonable amount of information for the business user to assimilate. In this example we presented the high-level information and summary by year for all of the metrics. We could have added something in terms of a period analysis rather than a yearly analysis or we could also have set up something for a Year to Date analysis for 2009. These are excellent additions that could have been made and we could have added these additional metrics to this dashboard but as we mentioned before, we don’t want to cram tons of metrics onto one dashboard. If there are more we either a) use a tab page to add additional layers to the dashboard and assign an appropriate title such as “Periodic Sales Analysis YTD 2009 by Product Group” and this would have all of the same indicators as the current view of this dashboard but at the period level rather than the yearly level, or b) use some sort of drill-through to get to the additional detailed analysis.

Let’s look at this based on the discussion we had earlier in this chapter. In the case of clear titles and descriptions I think we can see that everything is well spelled out and consistent in nature. There are no assumptions that the user has to make in understanding what these metrics are talking about. If I were to give this to someone who didn’t have any idea of what these metrics were all about they would be able to immediately identify the information and be able to make some specific statements about the Sales and Costs for the years 2007 and 2008. (As a matter of fact I did do this with two other consultants and ask them to explain to me what this dashboard was describing and within 20 seconds they were both able to explain at least three different components of this dashboard.) The color scheme is significantly different for the metrics across the top versus those on the bottom portion of the screen. If this color scheme was available based on this display you would be able to see that there is one metric that is graphically unique and that is 2008 Sales Price and Cost Averages by Product Group. This is the primary KPI for this dashboard. I’ve highlighted it both with colors and also with a 2.5D chart type rather than a one-dimensional display, to identify it to the business analyst as soon as the open the dashboard.

The titles and headers do not take up a disproportionate amount of space, and the information for the bar chart types is set to diagonal so that it doesn’t use up excessive space for the Y-axis headings. Each chart that is identified as Sales or Costs is scaled to the thousandths to conserve the space otherwise taken up by the extra zeros. Also notice that the maximum for each of the chart types will increase as the total values increase, and in all cases the scaling on the Y-axis is the same. This offers the analyst one uniform format for all the charts across the top and another uniform format for all the charts on the bottom. In this case, if we had decided to use another chart type, it would not have improved the statistical analysis available but it would have given it some additional eye candy. We even could have used the column chart type for all the metrics, but I decided to incorporate the use of another chart type, and the line chart type would not have worked correctly since there is really no history or range of information to chart. If we were to use period versus year for the analysis, then the line chart type would have worked.

As for configuration, if you look at one of the views of the data, you can see what was necessary to configure these Chart Web items. Initially, I had to format the view and get it to display consistently—diagram directly over diagram, center the dashboard title, and so forth—and to do this I used my favorite initial approach: insert a table. Once I accessed the WAD, I inserted a table into the design work area and inserted six Chart Web items into each of the cells. As shown in the illustration at right, I set the table to a 3×3 format and, to make the charts consistent and eliminate the gap between the charts, adjusted the width to 75%.

I typed the dashboard title into the middle space of the first row of the table and adjusted the font to bold and adjusted the texture and size of the typed header as shown in the next illustration.

As for each of the chart types, I did several formatting adjustments, but all of the parameters are available directly from the Edit Chart Designer component. Initially I used the wizard to assign the titles to each of the axes and to the top of the chart. This is shown in the following illustration. It’s important that you fill in all three titles—Category, Value Axis, and actual Title.

Once I finished in the wizard, I clicked Refine to enter the Chart Designer, shown next, to adjust the display parameter for one of the titles so that more room is available in the dashboard. I adjusted the format of the Block Style to Cylinder from Rectangle to give the columns more depth.

Next, I moved to the Category Axis element, clicked Title, and turned off the parameter for Visibility for the title along this axis, as shown next. This created more room for the statistics at the expense of the titles, but the additional information offered by the title wouldn’t add any value to the dashboard anyway.

I then turned off all the gridlines for each of the charts. This clears up the background of each of the charts and helps visually identify the information. It also helps with increased performance of the whole dashboard. Another parameter I adjusted is the coloring of each of the columns. I used the same approach shown in the previous examples and used both the primary and secondary colors to enhance the depth and visual appeal and used the gradient to include the two levels of color onto each of the columns. These settings can be seen in the following illustration.

For the bar chart type used for the averages, I turned on the 2.5D view of the bars and adjusted the titles for the Y-axis to diagonal. To increase the display view of the information, I adjusted two parameters in the Web Item Parameters tab in the WAD. First, I increased the width and height of the pixels from the default 300 each to 400 each. Second, I used the parameter Swap Display Axis for each of the charts to make sure that the data structures can be formatted to the chart type.

After making these adjustments, the dashboard was complete. Although this dashboard doesn’t have any “wow” factor, it does have a solid and consistent look and feel that would be very inviting to an analyst who just wants to be able to get the information they need from the KPIs to run the business and then work to improve those metrics.

Let’s take a look at another dashboard, for Sales Analysis, shown next. In this case, the KPIs being used are more practical, meaning they include not only indicators that give us information about the results but also leading indicators that will help us in terms of growing a business. The KPIs include customer satisfaction, actual order value, and market share. With these types of leading KPIs, we can estimate the growth of the company and make critical decisions about its overall health. Again, the dashboard doesn’t include large amounts of nonessential graphics or objects, enabling the business user to quickly understand the information and make the required decisions. The color scheme is basic, not bright and distracting. The KPIs use a very similar color scheme except the primary KPI, so that it attracts the business analyst’s attention first. It is also a different chart type.

This dashboard incorporates several chart types—stacked column, waterfall, and stacked bar. These offer a good vehicle to illustrate the indicators well and drive home the results. The dashboard doesn’t incorporate additional objects, such as List of Values; for example, we could have shown a series of regions with this information divided among them and offered a Radio Button Group to toggle between each of these different regions. We could also have set up variables or check boxes to help with this process. All these additional components are available and are fairly straightforward to incorporate into your dashboard. You will need to adjust the spacing for these different options and therefore possibly move from having 12 KPIs per screen to 10 KPIs, with the additional space allocated to either corporate logos or the components just mentioned. Basically, the enhancements that you can make to the dashboard process are limitless.

This Sales Analysis dashboard shows a series of indicators with formats that will help business users to understand and identify critical KPIs and immediately grasp the point of the information. Notice that we have switched to the stacked approach for this information due to its nature. Basically, there is only one actual value per view on this dashboard, and the other information is for analysis against planned information. So in this case, rather than having a dashboard with two lines—one for planned and one for actual information—we have ranges of information for planned or expected levels of each of the indicators.

This will provide the business analysts additional information and enable them to compare the overall growth and progression of each of the indicators. Which KPI business users are supposed to look at and understand immediately upon opening this dashboard should be clear; the overall structure of the lower-middle view is dramatically different from the others. The waterfall chart type displays this information in a manner that is unique but also very direct and focused. Business users can immediately see that there has been an increase in the overall progress of capturing new customers. The company has increased the number of new customers from 25 in the first quarter to 40 in the fourth quarter of 2008.

In this dashboard view, the information is very clear—quarterly net new customers. Display of the actual number has been turned on so that the business analyst doesn’t have to scroll over the information to see the total number per quarter, which is a time-saver for them. They can also see all four values at the same time, whereas with scrolling they would see only one value at a time.

Let’s back away for a moment and review the entire Sales Analysis dashboard. There are certain items that we need to review and validate to ensure that we have incorporated as many best business practices as possible into this dashboard. So, if we look at each of the areas of formatting, we see several pros:

• All the titles and informational items on this dashboard are consistent. Therefore, once you understand one title and chart range information, you can understand the others. This reduces the amount of time required to get comfortable with each chart type and their individual titles and subtitles.
• All of the color schemes are the same. To achieve this, I created one template object and then just copied it to another object. I had to make additional changes and fixes to the copied object, but having a starting point saved time and ensured consistency.
• The total number of KPIs is 9 to 12—depending on the approach that you use to count them. We have six actual KPIs, but in addition we have the same three indicators for each and this outlines the ranges of Maximum, Average, and Minimum for each indicator. These can be classified as three additional indicators.
• The amount of graphics used is consistent with the required look and feel and will not affect the performance of the dashboard.
• Subtitles and titles are consistent for each of the charts. For the subtitle, if it’s a stacked bar, the subtitle is below, and if it’s a stacked column, the subtitle is to the left side of the screen.
• Business users can easily understand and focus on each of the indicators. I conducted my own test on this dashboard and asked several people who are not involved with BI to look at it for the first time and tell me what they think. Within the allotted 20 seconds per KPI, all were able to understand the information and give one conclusion from the overall dashboard.
• The scaling is consistent and everyone is scaled to the thousands rather than in some with different scaling factors. This dashboard has all at one scaling factor level.
• In each case, the business analysts used in the test case were attracted to the lower level indicator first, and then started reviewing the other charts. This would validate the approach that I took to highlight the critical KPI.

As mentioned previously, the format of the query that supports this dashboard is as critical as what chart type is chosen to display this information. In this case we are using queries with four columns—Lowest Level, Average Level, Highest Level, and then the Actual itself. This will allow the series within the chart type to pick up the appropriate information by column. This drives home the need to have the correct query to support the correct chart type. Now the trick to developing this chart type is that the initial three column values of the queries need to be the differences rather than the total. So, your Lowest Level is the total amount used—in the case of Sales Revenue, it would be $500,000,000.00 (scaled in the graph). Now, the Average is actually$600,000,000.00 but in your InfoCube you are storing the difference or in the query you are calculating the difference—so the Average is actually $100,000,000.00. Using the average number of$100,000,000.00 and using the Stack option for the bar chart type you get the \$600,000,000.00. You do the same for the Highest Level, and when it is presented on the chart, you get your correct sections for the ranges.

To configure this dashboard, I again started with a table in the WAD. I then added the required Web items—Column and Bar Chart types. Just remember that if you need to adjust the size of the table, you should use the option to reduce the size. This will help eliminate any spaces between the chart types. I’ve used 80% in this case to help structure the Web items as well.

Now the rest of the configuration for the stacked columns comes into play with the series formatting. Within the wizard, I assigned the chart type stacked columns and then switched to the Refine approach to get to each of the series to configure at that level, as shown next. Using the different shades of color helps emphasize the sections of the stack.

For Series4, I changed the look and feel via the Refine view to the line chart type and also adjusted the format of the Area Property. In this configuration, I adjusted several items, as shown in the next illustration:

• I changed the Chart Type to Lines.
• In the Area Properties, I adjusted the two colors—Color and Secondary Color.
• In the Area Properties, I adjusted the Marker Shape to Triangle.

In the case of the waterfall chart type, I had to do a couple of additional configuration changes, but I used the same stacked column to start, and then just worked with the colors within the series to make them invisible, creating a “floating” center section. I created that type of chart in about five minutes. With a little touch-up, I achieved the finished product. The next illustration shows the configuration for the top section of the waterfall chart— I didn’t really make the section invisible but rather made it blend with the background so that the color is the same as the background color.

To complete this, we can now look at Series4, which is the line chart type. In this case, I adjusted the look and feel of the line and marker itself. As shown in the following illustration, I enlarged the size of the marker so that the complete value can be seen within the object. I adjusted this to a size of 30, and chose Circle for the Marker Shape. Finally, I turned on the (visible parameter) Show Labels option. The important item here is to leave the Format field blank. This will allow the values to show up as absolutes rather than fighting with currency or unit of measure (UOM) signs.

The additional configuration is very similar to that of the previous dashboard. All in all, this is a consistent dashboard and will accommodate many different requirements by your customers. Yes, you could get really fancy and add all the bells and whistles, even going as far as creating your own HTML that you type directly into the XHTML tab within the WAD. If you want items that consist of links to other formatting options that are not available within the WAD or links to other systems, you will have to use this functionality. All in all the WAD component of BI to help with the display of our KPIs within a dashboard format.

If we take this configuration and go just one step further using the Visual Composer (VC) toolset, we can see the finished results of a dashboard. Now this example has more dynamic visual appeal, but the basic information that you see is very similar to what we just developed from the WAD. The next illustration shows the finished product of a WAD-developed report in VC. You can see that this type of data is very similar to what we’ve reviewed, but now the “wow” factor comes into play.

The shading of the chart background, different approaches to use of variables in the report, and just overall eye appeal is much better in this example versus what we just created in the WAD. We could easily have included the variables, radio buttons, navigation pane, and so forth into this dashboard, but you can see that this will be fairly straightforward. Be selective with what is incorporated into your dashboards. If your business user is interested in having what-if scenarios available on a dashboard, think about putting this type of chart type on a separate tab within your dashboard. This will make very clear what is required reading versus what is available to use for estimates.