Background

Last week in Series Lines: Useful or Chart Junk?, I wrote about Excel’s “Series Lines” feature, and how it seems like a good way to clarify the data in a stacked column chart, until you implement it and realize it just adds chart junk to your chart. I proposed a panel chart to show the stacked chart data more clearly.

Gregor Aisch commented that my panel approach did not allow point-to-point comparisons as effectively as a simple line chart (despite the line chart’s clutter). So I put together the following example of an interactive chart that allows a user to compare a couple series in context of the whole data set.

Here is the original stacked column chart, with series lines cluttering up the space between columns.

Here is the panel-style chart I constructed to show the trends more clearly. As Gregor has pointed out, comparing individual point values (for example, 2008 for alpha and for epsilon) is difficult. Also, if there are many categories (alpha through omega, for example), the data will be compressed horizontally and will be impossible to read.

Our interactive chart will look like this. The user can selectively highlight data using the two listboxes next to the chart.

Interactive Chart: Data

We’ll use this data for our exercise. A1:G7 contains the data used in the previous discussions. A9:G11 contains repeated data for two of the series in the upper table. These are selected using Excel’s INDEX() worksheet function and the index values in H10 and H11.

Cell A10 contains the formula

=IF($H10>1,INDEX(A$1:A$6,$H10),"")

which is copied into A11. Cell B10 contains the formula

=IF($H10>1,INDEX(B$1:B$6,$H10),NA())

which is copied into C10:G10 and B11:G11. The way these formulas work is as follows:

If the index in column H is 1, column A looks blank and columns B:G contain #N/A, which is not plotted in a line or XY chart. If the index is greater than 1, the label from column A and the values from columns B:G are displayed in that row. By changing the values in H10:H11, we can change the data in their rows. If this data is plotted, then changing H10:H11 will change the chart.

Worksheet Controls

It’s nice to be able to modify a chart by changing numbers in a couple cells, but this becomes tedious after a while. Fortunately, Excel has a number of helpful controls that we can drop onto the worksheet to do this in an easier and more natural way. For choosing from among a list of items (alpha through epsilon), we can use a Combo Box or a List Box.

In Excel 2003 and earlier, you can access these controls by displaying the Forms toolbar. In Excel 2007 and later, you need to display the Developer tab*, then pick from among the Forms controls under the Insert button. (There are also ActiveX versions of these controls, and they have more extensive formatting capabilities, but ActiveX controls can misbehave.

* To show the Developer tab, you have to open Excel Options, and find the setting for the Developer tab.
In Excel 2007: Click on the big Office button in the top left of the window and select Excel Options; check the box in front of Show the Developer tab in the Ribbon.
In Excel 2010: Click on the File tab and select Options; click the Customize Ribbon button on the left, and in the right list, check the box in front of Developer.

Here is the Combo Box button on the Developer tab.

And here is the List Box button.

Both the Combo Box and List Box display a list of items for the user to select. I prefer the List Box, because you can several items even without clicking on the control. The Combo Box only shows one item at a time, until you click the control. Of course, if there isn’t room to display a List Box, then the Combo Box works fine.

When you click on the control button in the ribbon (or on the Forms toolbar in ancient Excel), the cursor turns into a crosshair, and you can draw your control just like drawing any other shape. In this example I’ve made two identical controls, one for each series we allow the user to highlight. You could display two list boxes side by side, two combo boxes side by side,  two combo boxes one above the other, or any other arrangement that works in the available space on the worksheet.

Even if I’m using multiple controls, I usually make one and format it, then make as many copies as I need, either by using Copy and Paste or by Ctrl-dragging. After copying, I change specific formats (such as the cell link, below).

When you right click on a control and choose Format Control, you are presented with this dialog. In a combo box or list box, the input range contains the list of items displayed in the control. The cell link is the cell that contains the index number of the selected list item, starting with 1 for the first item in the list. This is how the index values in H10:H11 will be changed.

With the controls formatted and the formulas already in place, play with the controls and make sure the data changes as expected. Note that multiple controls can use the same cell link. Changing one control will change the value in the linked cell, and consequently change any other controls using this cell.

Building the Interactive Chart

The data and the controls are all set up, so now we create our charts. Select A1:G6, and insert a line chart. Resize and format the chart as appropriate.

Now format all the series to de-emphasize them. Thin lines, small markers, light gray color.

It’s easier to keep these gray series, and add duplicate series with bolder formatting for the series we want to highlight. The lighter data is hidden behind the bolder series.

To add the highlighted series, copy A9:G11, select the chart, and use Paste Special to add the data as new series, with the First Row and First Column check boxes both checked. Format these series to stand out compared to the light gray background data, and add a series name data label to the last point of each.

Now arrange the controls next to the chart. The list boxes have ugly black borders, but I hid them behind unfilled rectangles with light gray lines, and I colored the chart area border to match.

In fact, you can relate the list boxes to the appropriate highlighted series by coloring the gray rectangles around the list boxes to match the series.

Using the Interactive Chart

The user can now highlight and compare two series, without interference from the rest of the data which is still shown in the chart. If no other series are selected in the list boxes (or combo boxes), the corresponding data values are all #N/A, so the points don’t even appear in the chart.

Any two series can be selected for comparison. Note how easily the opposite trends of series gamma and epsilon can be seen in this chart. In the original line chart it was very difficult to resolve these series, because the other series were not de-emphasized by shading them gray. Yet here we can still see where these two series lie compared to the overall spread of the data.

Peltier Technical Services, Inc., Copyright © 2011.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

Yes, I did once have a tutorial that showed how to make a speedometer chart. But there were a couple of reasons that prompted me to take it down. First, it was very complicated, requiring a good working knowledge of algebra and trigonometry, and I got lots of emails asking me to clarify the protocol. Second and more important, these charts are particularly ineffective in the display of information. I removed that tutorial from my web site, and it has also disappeared from the computers in my office. No use asking: it’s gone and I will not reconstruct it.

I know you were probably instructed by a supervisor to use pretty dials, and probably 3D charts as well, and pretty colors, and flashing lights, and maybe a buzzer or two. What the boss wants, in this case, is not what the boss should get.

Dials and Gauges

Here are a few articles written about how poor dials are for displaying data, and how the cockpit of a fighter jet or the dashboard of a Formula One racing car makes a poor metaphor for a business dashboard:

Be as provocative as you wish, but back your opinions with substance – Stephen Few

Down With Gauges! – Charley Kyd

Car Dashboards – A Broken Metaphor for Executive Dashboards – Jorge Camoes

The Ultimate Business Driving Machine – Zach Gemagnini

Speedometer Chart – this web site

Bullet Graphs

If you do want to use some kind of single-value indicator, you could use a bullet graph. Though it only shows one value against a background scale, a bullet graph takes up minimal space, so its use leads to greater data density. Bullet graphs are well suited for use in tables, because they can be as small as a cell in a row of data. Here are some sources about bullet graphs:

Bullet Graph Design Specification – Stephen Few

How to Create Bullet Graphs To Replace Gauges in Excel – Charley Kyd

How to Make Vertical Bullet Graphs in Excel – this blog

Better Alternatives

Line charts are a better choice than dials and bullets if you should be showing data over time, instead of a single value.

A column or bar chart with appropriate background shading can take the place of multiple bullet charts (see Horizontally Banded Chart Background and Horizontal Bands in the Background of an Excel 2007 Chart).

Peltier Technical Services, Inc., Copyright © 2011.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

1. financial flash dashboard demo
2. sales flash dashboard demo
3. sales management dashboard
4. airline dashboard demo

As I discussed in my comment to the Spy’s post is that a little animation goes a long way.

If “animation” only means slick effects showing the graphics being constructed, then it is a waste of resources and of the user’s time. A good dashboard does not need animation. The graphics should appear instantly in their completed state.

If the animation shows the user how conditions or performance have changed over time, then it may be a worthwhile animation. It must be done cleanly, with controls that allow the user to stop and start the animation, and stop at any intermediate place during the process. I have described a number of animation examples in this blog, and all of them show the development of one state from another, with playback control buttons that everyone is familiar with. But this type of animation is usually better for more detailed descriptions than are effective in a dashboard.

A good dashboard could also benefit from interaction, rather than animation. Interaction may mean the ability of a user to drill down on a chart to see further information, or the ability to change some inputs to see effects on predicted behavior.

The animations provided in the Fusion Charts examples do nothing to elaborate on the data. They do not clarify information in the data. They serve only to entertain the user instead of instantly showing the information that was called up. Therefore the Fusion Charts animation qualifies as gratuitous eye candy. Such animation does not belong in any effective business dashboard.

In 2008 Excel Dashboard Competition Winners the winners of the 2008 Excel Dashboard Competition are announced:

1. Wade Stokes – International Bank Dashboard

Displaying many disparate Banking Key Performance Indicators, and designed as the basis for the Management review of business performance, it truly achieves More Information per Pixel.

2. Jim Uden – Outpatient Surgery Center Dashboard

Developed for Meridian Surgical Partners, as a one page snapshot for the review and presentation of partnership level business operations and trends. Jim also includes probably the best associated description of dashboard content and the thought processes involved which we’ve seen.

3. Hitesh Patel – Pharmaceutical Sales Dashboard

Developed by Hitesh Patel and Mike Askew of Data Intelligence, for Bristol Myers Squibb. A key report for the Regional Sales Managers, containing the information required to run the business in terms of cash, growth, share, and competitive performance.

In More Information per Pixel!, they display a very nice report by Professor Rolf Hichert:

Rolf very effectively uses the technique of small multiples with sparing colors to highlight high and low performers among a portfolio of organizations.

These MicroCharts guys are really onto something.

A dashboard report is not a set of dial gauges that mimics the cockpit of a 747. A dashboard report is the combination of a large number of small, well-designed charts and tables, thoughtfully integrated to pack a great deal of information into a small area on screen or on a printed page. In this post I describe some free and commercial resources for creating dashboard reports, particularly within Microsoft Excel.

Dashboard Resources

Stephen Few of Perceptual Edge has spent a career learning and teaching what techniques effectively display data so that human eyes can readily perceive it as information and human minds readily analyze it as knowledge. Stephen has written two books entitled Show Me the Numbers and Information Dashboard Design, which explain how to present information in ways which are known to be effective, based on the psychology of cognition and the physiology of vision. These books are very good references for understanding perception and cognition in terms of informational display.

Businessman and Excel MVP Charley Kyd of ExcelUser explains the benefits of dashboard reporting, and shows how to construct robust dashboard reports with clearly written procedures. Charley has incorporated Few’s approaches with his own background in Excel financial analysis to produce a comprehensive guide to using Excel to produce efficient dashboard reports. Charley’s commercial materials include an e-book, Dashboard Reporting with Excel, and a number of example workbooks that show the concepts and techniques behind effective Excel dashboard reports. In addition to construction of dashboards, Charley shows how to design a system of workbooks and directories so that regular updates of a dashboard is painless, taking minutes instead of days.

In his intelligent and informative blog Charts, Jorge Camoes discussed dashboards. Jorge takes a pragmatic approach, and offers instructions on dashboarding in Excel as well as reviews of other charting products. His Excel posts include How to create a dashboard in Excel, Excel Dashboards: do you need VBA?, and 10 tips to improve your Excel dashboard. Jorge has reviewed the use of Crystal Xcelsius as a dashboarding environment, and found it lacking (which matches my own experience), and he has provided a link to a dashboard best practices paper by Dundas, maker of fancy charting software. Visit Jorge’s archive of dashboard-related posts.

Excel MVP and author Mike Alexander of DataPig Technologies has written about Excel pivot tables, integrating Excel and Access, and about dashboards. Mike has written about Crystal Xcelsius, and his new book, Excel 2007 Dashboards and Reports for Dummies is being released this month.

Excel MVP Chandoo of Pointy Haired Dilbert has written a series of posts about dashboarding techniques for Excel. He has a comprehensive list of these posts at Excel Dashboards.

The Dundas white paper Dashboard Best Practices is [no longer] available both online in html format and as a pdf document. In Jorge Camoes’ post Dashboard design: we need best practices for best practices, Jorge finds that the content of the white paper accurately reflects the philosophies of the experts cited above; unfortunately he also finds that the Dundas charting package and documentation seems to promote conflicting practices, with special graphical effects obscuring the accurate dispay of information.

Sandy Cavalaris, who writes the Excel with Monarch blog, describes how to use Excel in conjunction with the Monarch data management product to create a dashboard reporting system (part one and part two). I know very little about Monarch, but it seems like a worthwhile approach to try before dumping hundreds of thousands of dollars in a monolithic ERP system.

Dashboard Advice

I have the following advice to anyone who is developing dashboard reports. You already have Microsoft Excel, so you should try to use it as a dashboard application before investing in expensive and obscure software. Excel can link to local data and to corporate databases, so it can wield the power of large packages with less overhead. Start with Stephen Few’s book on Information Dashboard Design to gain insights into what makes effective data displays. Read Charley Kyd’s web site and e-book to get practical approaches to dashboarding in Excel. Read Jorge Camoes’ archive of dashboard-related posts for more practical insights into Excel dashboarding. Then build and tear down and build again, until you have a display system that is informative and leads to framing good business decisions.