• Votes for this article no votes for this yet
  • Dashboard Insight Newsletter Sign Up

Dashboard Reformation: Government Finance External Department Dashboard

by Alexander 'Sandy' Chiang, Research Director, Dashboard InsightTuesday, November 29, 2011

Introduction

I am revisiting the post on Government Finance Department Dashboards that was showcased on Dashboard Insight in May, 2007. In this post I will highlight the challenges of one of these dashboards and ways to improve it.

The current challenges with the dashboard

The use of interactivity

Dashboards, by their nature, should display only those performance indicators that are most important to monitor. This dashboard provides the option to filter by department and year, which contradicts Dashboard Insight’s fundamental definition of a dashboard.  If the dashboard user will do visual analysis, then this is acceptable. However, assuming that the purpose was to quickly scan performance metrics to see if there are any correlations, trends, outliers (anomalies), patterns, or business conditions, then the department filtering should be removed.

The real estate is not being effectively used

There’s a lot of white space in this dashboard that could be used to represent more data.
For argument’s sake, let’s say that it is important to see every department by these measures. By choosing different visualizations, eliminating unnecessary text, and rearranging the data around, we can maximize the use of the space and display everything.

Poor use of color

I have a fundamental problem with using gradient effects to represent the state of a value as demonstrated by the gauges here. The difficulty with using a color gradation is the uncertainty of the status it represents. For example, what does orangey-red represent? Using discrete colors to represent a business condition that requires action is preferable.
The overuse of different colors in the dashboard also makes it hard to guide the eyes to points of interest and difficult for the user to easily see what needs attention.

The dashboard reformation

In addition to the problems noted above, there is one critical question that must be answered for a proper redesign of the External Departments dashboard: does the department selection provide the same metrics but different values? I.e. each department measures: average internal P.V. Examination Lead Time (Days), Average Entry Variance (Days), Average Internal Audit Lead Time (Days), and Amounts Contribution.

If the answer is yes, then we should provide all the possible values for each department on one screen rather than having the user select a particular department to look at. This will make it easier to compare values across departments if they all appear on one screen.

If the answer is no, then we can simplify the design so that it can complement another dashboard. The less the user has to navigate, the better.

There are other changes I’d like to see happen with the dashboard such as adding a comparative value to the Average Internal Audit Lead Time (Days). I will ignore the value of the performance measures chosen in this example and focus mainly on the design changes.

The following is the end result with the assumption that each department does have the same metrics. In addition, I assumed the dashboard should display the current year by default.


Click on image to enlarge.

Dashboards technically should not require any interactivity to see problem areas. However, dashboards can serve as a starting point for visual analysis. With that said, I kept the year as a drop down option.
Here is the summary of the most important changes:

  • Displaying all the metrics for each department, which negates  the need for a department filter and offers the ability to compare departments
  • Replacing inefficient gauges with space-efficient bullet graphs
  • Replacing  the pie chart with bar charts, which provides the ability to compare across departments as well as within itself
  • Toning down the colors, which make it easier to highlight important values. Those who prefer more color can choose another base color and use different shades of it. However, make sure to choose a color that contrasts with the background for readability.

There’s a lot of real estate still available so it’s easy enough to incorporate the Auditing Sections Dashboard using the same techniques.

One final note, as suggested by the comment in the dashboard, the dashboard can  display exact numbers and details for each visualization when the user places their mouse over it. This keeps the dashboard clean and users can still get exact numbers if needed. However, if the exact numbers are critical to determine a condition, then another layout needs to be considered to make room for the text.

If your dashboard needs bright colors and round gauges, then it’s your data that’s boring.
Anonymous

Tweet article    Stumble article    Digg article    Buzz article    Delicious bookmark      Dashboard Insight RSS Feed
 
Other articles by this author

Discussion:

george sloan said:

Sorry...but I don't agree with this article. I think the dashboard design should reflect what the customer wants, not these types of strict rules. If the customer wants an interactive dashboard with lots of drop downs, or orangey-red colors, then that is what should be provided.

Alexander 'Sandy' Chiang said:

@George - I've backed up all my design choices with clear explanations of the benefits of incorporating them. In response to your comment about designs that reflect what customers want, it's important to figure out what they need rather than what they want.

I prefer to strongly advise my clients to go in a particular direction that's in their best interest rather than just doing what I'm told. Otherwise, I bring no value to the client. They might as well just outsource it to the lowest bidder.

Yes, it's hard sometimes to get people off of crazy colors and pie charts, but if I've at least explained why it's bad, then I can at least feel that they've made an informed decision on the design.

In any case, I opine that a good consultant listens to what the customer wants, figures out what they actually need, and proposes the right solution for them.

Nelson Ng said:

@George With respect, I find your comment baffling, especially given that this article is about dashboard design. I suppose what you're saying is valid if:

- your customer is a information interface design expert
- you don't really care (or aren't paid to care) about whether or not your design is actually solving the customer's business problems
- you see dashboards as fancy graphical representations of data rather than real line-of-business decision support tools
- you simply don't think that the design choices you or your customers make have any real impact on the utility of dashboards

Would you care to elaborate?

This in my view is akin to hiring a UI design expert to design the interface for an application, and then simply ordering the expert to execute your own design. Why hire an expert?

The point of my comment is NOT to say that we should ignore our customers. We should listen to them closely, and go further to drill beyond their perceived problems and analyze what their real problems are, as opposed to merely satisfying requests for preferred colors.

Design choices matter, and we are paid to do what we do because design, particularly when it comes to designing displays of quantitative data, is so poorly understood.

Zach Gemignani said:

I think your redesign is headed in the right direction but I'd like to quibble about the changes you made to the "Amount Contribution" section. The layout you created emphasizes comparison across departments rather than the breakout of contribution within a department. Given the pie chart in the original dashboard, it seems likely that the audience wants to know how much "Below 10k" contributes relative to "Above 500k" within one department. By laying those bars out horizontally, you make it difficult on the users to make those intra-department comparisons.

Mark LeGassick said:

@Zach, I agree with you. I believe that dashboards need to be intuitive and quick to understand. I don't think the visualizations for the "Amount Contribution" does either. That being said, I do believe the rest of the dashboard is an improvement.

Alexander 'Sandy' Chiang said:

@Zach - Yikes! You're totally right. I was definitely designing under the impression that the user wanted to compare across departments per segment. However, your assumption makes more sense. A good solution to accomplish that would be to transpose the rows and columns. I.e. metrics as rows and departments as columns. Good catch!

Scott Kent-Collins said:

I think your redesign could also use some more labelling (or a legend)... presuming it isn't elsewhere on your dashboard. There are no figures anywhere on your redesign above:

- no legend to inform what the thresholds are on the bullet charts.
- no information about why the bullet chart lines change colour.
- no specific values associated with the "Amount Contribution" categories.

Also the trend in your cell sparkline for "12 Month Trend" for finance doesn't seem to match the trend in the original "Avergage Internal Audit Lead Time (Days)" chart.

Regards. :)

Alexander 'Sandy' Chiang said:

@Scott - You're right. My legend got cut off when I copied and pasted the image. How embarrassing!

I'd also add for the sparkline that it would help to add the first and last months and show the min and max of the Y axis to give it more context, if needed.

As for the values in the Amount Contribution (and for all bullet graphs, as well), the intent was to have the details show up in a tool tip. However, I should have stated that in the dashboard. I.e. have text saying, put your mouse over the visualization for more details.

Yes, the data isn't the same. :D I made up my own data for the mockup.

Scott Kent-Collins said:

@Alexander... yeah, I thought there would probably be a legend somewhere.

Can't say I agree with the "tooltips" option in this scenario. They have their uses, but only in dynamic environments and don't pull through to printed formats.

I appreciate what you're trying to do here by employing Few's techniques, but ultimately we need to make things easy for the users.

The original (whilst wasting vast amounts of space and misusing colours) also contains specific figures for users to easily find and take away.

Your redesign reduces the space required whilst adding the ability to compare across departments, but makes the user "work" to find exact figures:

- a legend look up to review thresholds and guesstimate actual metric value.
- shows trends but no specific monthly figures.
- tooltips (and wait times) to display actual details.

All making the user work more for the information, but I guess it comes down to the purpose of the tool. Is it to get a quick, general idea on trends/comparisons... or is it for something more specific... or somewhere inbetween.

Cheers mate :)

Alexander 'Sandy' Chiang said:

@Scott - You're right. It depends on the purpose. I went on the assumption that the main goal was information at a glance which is the purpose of a dashboard as defined here on Dashboard Insight. In my write-up, I mentioned that if more detail was required, then the design had to be reconsidered. If printing was required, then the values would need to be shown or, at least, context for the values.

Thanks for pointing these tips out. These are the type of points that need to be considered when designing a dashboard or visual analysis tool.

Site Map | Contribute | Privacy Policy | Contact Us | Dashboard Insight © 2017