Last week I launched a challenge to redesign financial info from the Gates Foundation annual report. I was looking for entries that made the data more engaging than a table, matched the level of design of the rest of an annual report, and still reported each element originally presented by Gates.
Every entry that hit my inbox made my heart grow bigger. Here they are, all 10!, along with the commentary submitted by the designer and any notes I had about the redesign. Read through and leave your thoughts in the comments.
The take away message I gleaned from looking through these entries is that every real world design involves some kind of tradeoff. There’s almost always a compromise that must be made somewhere. Real life design is much harder than armchair critique.
I’m starting with Heather – her entry is the longest but she makes a point about not using percent change, which is what most of the other entries did. And I’ll end with Joshua because his made me laugh.
My materials should be considered a draft of a concept – not as a finished viz. I’m not submitting it because I believe it is the single best option – there never really is – but rather because I think it’s an interesting conversation that I’d like to be part of.
A few starting points – Stephanie says in her blog post that she’d like to see some viz that will draw in the more
general public, not just the typical accountant. I think this is a great idea.
The purpose of the financial statement of a large non-profit such as Gates is not to tell a story of change but rather
to tell a story of stability and provide transparency as to sources of assets and liabilities.
The peaks and valleys of one of Stephanie’s proposed redo’s emphasizes change when there really hasn’t been
much change. I also struggle with the inconsistency of scale here.
Additionally, numbers as large as this really don’t mean much to most folks.
The story to be told by this table is that the Gates Foundation financial situation has remained stable over the past year. And the distribution of assets and liabilities has also not changed much, with the majority of assets coming from the Foundation’s interest in the Gates Family assets. (SE Note: Do we really know what constitutes a big change? Unfortunately, I think we lack that contextual knowledge.)
I agree with Stephanie’s point in her original piece that the design of the financial page needs to coordinate with
the design of the rest of the report. I’ve taken a look at some of the previous Gates Foundation reports and I think
the basic design I’ve developed would synthesize well. I kept the font standardized, etc.
The viz is not meant to replace the traditional table. The table needs to stay to satisfy the portion of the audience that prefers a traditional approach to accounting. My viz is meant to accompany the table – ideally in a side by side two-page spread.
Last note – I saw percent change discussed as a possible way to standardize the change over time with these numbers and make them possible to visualize on the same axis. This is a very problematic suggestion. The reason being that the metric of percent change is very vulnerable to the size of the denominator. And the large
differences in scale of the various categories will make percent change really misleading. (SE Note: Agreed! It’s a matter of tradeoffs.)
The top chart displays the actual values of each of the major categories The bottom two chart shows the relative
distribution of the sources of assets and liabilities.
The story that this page of the report (Some of the other pages are actually way more interesting stories) is
1. The total assets of the foundation are increasing
2. The total unrestricted net assets are increasing
3. How the assets are distributed currently and how that distribution of assets is changing
4. How the liabilities are distributed currently and how that distribution of liabilities is changing
And the overall big story is that Gates is getting most of its cash from interest in the Gates net assets. And that
most of its assets are unrestricted, which is very good.
Heather Krause, firstname.lastname@example.org, @datassist
SE Note: Emily blogged about her submission. I told her that I think one of the things that gets me is that financials list the most current year on the left, when we usually read oldest on the left. For example, your first line looks like its going down, but its really an increase of 49%. I took the liberty of swapping the years in my visual but I don’t know that it would be allowed.
- Maintains a consistent color scheme between visualizations depicting data from the same source/categories
- Used green and red lettering to indicate increases/decreases (on bar charts)
- Good for non-color-blind people
- Bad for red-green color-blind people
- Used directional arrows to indicate increases/decreases in data trends
- Utilized bar of pie chart to highlight small data pieces
- Inconsistencies in scales on bar charts
- Length of arrows are not proportional to percentage
- Possibly too many colors
- Fits on three pages instead of less than one page
- Was not a quick process to create
In terms of fixing the shortcomings, it adds additional comparability between the data by adding the percentage changes from 2013-2014 and the total percentages between 2013 and 2014 for assets and liabilities. It slightly fixes the inconsistencies across scale measurements by putting the “in ###s” underneath the labels in the X-axis, but even with those it is still difficult to compare across categories. I separated out the totals for liabilities and assets instead of having them in the same line as the other subcategories, which allows more ease of interpretation.
This example goes to show that there are a variety of ways to visually represent data, and that when you report data you should know your audience and their style/level of competence regarding various data visualization techniques. The type of story they want to see will influence the types of visualizations you use.
SE Note: You might be cheating by expanding this on to multiple pages! But maybe you’re just a dreamer who thinks data viz deserves more than one page in an annual report. In which case, I’m totally on board.
The way my solution addresses those shortcomings is by simply not distorting the data.
1. Using the same scale for the bar charts representing the amounts in USD for the years 2013 and 2014–because dollars are dollars, whether they are in 2014 or 2013.
It’s clear by looking at the horizontal bars that the majority of the assets (98%) are in Beneficial interests, followed by Property and equipment (1.6%) and everything else is negligible (0.3% or below). Just imagine that $ 43.4 billion is 2,767x bigger than $15.7 million. That’s a huge difference and it should hit the viewer in the eye.
2. Using a simple variance bar chart for the percent change. You can quickly see that “Cash” had the highest percent increase–and not “Beneficial interest” or “Total Assets” as perceived from the slope of the lines in your chart.
SE Note: I like the addition of extra data, like reporting on the past 7 years, though it’s kind of cheating in this challenge. However, the bar graphs are an issue for me because all but the first ones are undetectable. What’s the point of having a visualization there if you can’t see it? The only story it tells is that the first line item is a lot bigger than the others. But again, every solution here is going to have a design tradeoff somewhere. Thanks for your entry. You’re the only one who took issue with my design on Twitter and actually tried to work out a full solution.
We don’t know the exact context in which this graphic was originally prepared, so I made some assumptions about the purpose and audience of the graphic in order to guide my redesign. The graphic comes from the Consolidated Financial Statements of the Bill and Melinda Gates Foundation. It’s a somewhat technical document that I assume is intended for people closely involved with the foundation, like partners, investors, etc. It is not a document intended for the general public or for internal accounting people at the Gates Foundation. The audience expects some detail and is familiar with financial tables, but their main interest is in overall trends and magnitudes.
Given the audience, my strategy was to maintain something of the financial table look while adding the minimum number of visual elements to communicate trends and magnitudes. I didn’t want to lose the sense of accounting integrity that I assume the original table suggested to the audience. What the original table is lacking is any visual indication of the changes in value from 2013 to 2014, which I assume is one of the two main points of interest. I think the other main point concerns the magnitudes of the values and the fact that there is one dominating value in each category, assets and liabilities.
So here’s what I did. The only visual elements I added were sloped lines connecting the values of different years that indicate the percentage change in value. In addition to the lines, I rearranged the text so that values were organized by magnitude and by amount of change. I also got rid of the last item “Total liabilities and net assets” because it’s exactly the same as total assets (I’m sure the accountants won’t like that!).
– minimize added visual elements
– highlight changes in value
– organize/group by magnitude and percent change
– reduce the significant digits
I did not attempt to find a story in the data, I am merely presenting the financials in the best way I could think of, limiting myself to about 1-2 hour to design and produce the visual. I decided to show absolute amounts in a tabular format and visualize % change from 2013 to 2014. This gives a quick indication of where the largest proportional shifts occurred. So as not to lose the picture of actual $ amount changes, I coded them as line/bar thickness. Note that I also reversed the scale on Liabilities so the decrease in Liabilities is on the same side of the graph as increase in Assets. I wanted to give a consistent indication to what I labelled beneath the graph ‘Money OUT’ and ‘Money IN’. A boxed summary above the visual gives an additional emphasis to the big picture of total changes and I leave it for the viewer to conclude that total changes in assets and liabilities matched.
SE Note: Width and height can be tricky! What do others think?
I think the multiples in this case just won’t work. In both assets and liabilities the top category dwarfs all the others, both in absolute values as well as in changes from the previous year. That’s why I focused on exposing the contribution of the changes to total assets, total liabilities and the resulting net assets.