True Stories about the Benefits of Data Visualization

I’m occasionally asked by journalists to describe actual cases when organizations have derived real tangible benefits from data visualization. When asked, I’m usually forced to answer in terms of generalities, for the following reasons:

  • The nature of my work with clients—training and design consulting services—rarely gives me a chance to see the results of my work.
  • On those occasions when I am able to see work that a client produced as a result of my services, I am rarely allowed to share it publicly.
  • People often contact me to say how much they appreciate my work, especially my books and articles, but they rarely share specific, tangible benefits.
  • When people have shared specific accounts of benefits, I’ve remembered only the gist of those accounts, almost never the details.

Even though we have plenty of evidence from years of research to support the tremendous potential of data visualization, we are lacking in specific accounts that confirm beneficial outcomes in the real world, either empirically in the form of measured results or anecdotally.

I was reminded of this frustrating blind spot today when I received yet another interview request from a journalist and imagined myself explaining once again that, although there is a great deal of evidence that data visualization works based on perceptual studies, etc., I have little documented evidence that it works in practice.

I need real stories from you who use data visualization to analyze and present data. Has data visualization led you to important findings? Has data visualization helped your organization increase revenues or decrease costs? Has data visualization increased efficiency or productivity? Have good decisions been made because information was presented visually?

Help me out here. Tell me about your experiences. Be my eyes where I cannot go myself to observe.

Take care,

21 Comments on “True Stories about the Benefits of Data Visualization”


By Andy. August 19th, 2009 at 10:32 am

Stephen, I’d suggest contact some of the presenters from the 2009 Tableau Customer Conference. Most of the case studies provide detailed payback that resulted from the analysis. Whether they can share the data or not is another story. I know I shared the impact in my case study last year, but then I was crucified for sharing it.

By Taggert Brooks. August 19th, 2009 at 12:40 pm

Stephen-

I eagerly anticipate the responses that you compile, and though I’d love to give you an example, I too have none. When teaching business students I’m faced with the same paucity of examples you face with reporters. I suspect another reason is the difficulty everyone has estimating the counter-factual. Tufte uses the example of the shuttle disaster and power point. But can we know that they wouldn’t have made the same mistake in the absence of powerpoint? I suspect your customers like the improvements you help them with, but might have trouble identifying how their decision making changed because of the improved visualization.

By Robert Meekings. August 19th, 2009 at 3:01 pm

Along with the shuttle, I’d add the 1854 cholera outbreak in London and John Snow’s map.

Robert

By Robert Kosara. August 19th, 2009 at 6:51 pm

Good idea, we certainly need a collection of case studies to make the case for visualization’s usefulness. The Discovery Exhibition at InfoVis this year will hopefully provide some good examples. I’ve just had a submission accepted, which talks about three uses of Parallel Sets. None of them are earth-shattering, but they show how different people make use of the program with their data. I’m sure there’s a lot more out there, we just have to find it …

By derek. August 20th, 2009 at 2:23 am

I had my quarterly assessment recently, which was two hours of being quizzed on a myriad of performance indicators, each one of which was accompanied by “can you give a specific example of how you have excelled in this in the last three months?” I quickly ran out of original examples and had to fall back on “You know that case I mentioned a minute ago? I think that also shows me performing well in Value X”

I get the same problem in job interviews. Just to say, it’s a widespread problem :-)

By Simon Russell. August 20th, 2009 at 2:33 am

Hello Stephen, your books and articles are inspiring.
Real results and payback? Absolutely. As an airline improvement consultant, data visualization is critical to see masses of data - millions of rows often. Breakthroughs from data visualization allow aircraft daily or weekly utilization and route patterns to be “seen” - lifting detailed information from time critical schedules. Other visualizations help with airfare pricing decisions - high/low fare spread and seasonality, etc. A dollar value? Yes. Data visualization allows a user to “see” gaps in aircraft use - time holes which can be switched to save aircraft, or add flights. An extra hour a day use per aircraft on an average 10 hour day, can potentially release 1x $80m aircraft(purchase cost new of a 737 or A320),or $3-6m p.a. in lease costs. Data visualization can deliver!

By Tim Dietrich. August 20th, 2009 at 7:32 am

I think the problem lies in the fact that the features are not measured. Good graph designs are both effective (answer the right question, answer the question right) and efficient (least amount of time to interpret). Most (all?) businesses do not measure the effectiveness and efficiency of graph design. A graph that is not effective or efficient has a defect. Good graph design is about the prevention of defects.

In my opinion we only notice the ones with defects. I’ve been in meetings with senior managers and poorly designed graphs have wasted significant time in interpretation or the answer does not get communicated as intended (they have defects).

I agree with Taggert. We only get to observe one outcome of a graph. What’s needed are experiments in which we get to see both outcomes– good and bad designs assigned to two populations and compare measures of effectiveness and efficiency. Arguably, this work has already been done by Cleveland and others. That’s how we ended up with the knowledge to design better graphs!

By David Jaques. August 20th, 2009 at 8:37 pm

Stephen,
We met at the Tableau Conference where our team presented very tangible results of data visualization at our Medical Center. This is used realtime to assure proper nursing levels for all our patients, to identify safety and quality of care trends that deserve more attention, to address capacity and efficiency in operating rooms and to reduce premium pay through more controlled pay practices. Discovering the type of patient that may have frequent readmissions to the hospital is a current active exploration which will direct specific efforts to focus on these ‘at risk’ patients to provide more frequent follow-up and detailed discharge information. The use of visual analysis has stimulated our organization to pursue a data transparency model that will encourage others to explore and identify solutions to their questions by creating interactive dashboards that push the data to our frontline managers. We are adding people to this effort because we are convinced that this will be a foundation element for quality healthcare delivery.

By Rob. August 20th, 2009 at 10:32 pm

I stumbled across a major error in what I though was a clean dataset using visualisation earlier this month. I posted the details here http://blog.une.edu.au/robbi/2009/08/06/data-testing-using-visualisation/
I am not sure if testing using visualisation is a discipline in itself but suspect it could be. Tools like Tableau are able to provide instant visual confirmation of the validity of large volumes of data in ways that I believe would be hard to perform using more conventional means.

By Wendy Cook. August 24th, 2009 at 4:10 pm

I posted Stephen’s query on the IABC (International Assn. of Business Communicators) discussion group on LinkedIn, with a link to Stephen’s blog. Maybe some of them will have some input since those folks are on the front lines. If there are other such LinkedIn or social media groups, the word could be further spread.

By jerome cukier. August 26th, 2009 at 3:56 am

Stephen,
One way to tackle that problem would be to set up a gallery of before/after examples.
There is a vast number of data visualizations out there which are hardly helpful, because their authors were not aware that by simple design choices, they could make them much easier to understand, interpret and operate.
Even when the benefits of a better display are difficult to assess, the difference between a normal chart and one with a reasoned approach on design is striking, which is what I take from your books.
usually we don’t like to show our old displays but if they are successfully redone according to the rules that you preach, I think they could inspire more people.

I work for a government agency so I won’t have a story where a visualization generated billions in sales. however as for a chart having impact, how about that famous one in an inconvenient truth (al gore on a crane lift)? CO2 emissions in the US haven’t changed since, but it became one of the top issues in the public debate of all western countries almost overnight.

By Withheld. August 27th, 2009 at 1:40 pm

I employ many of the data visualization techniques outlined in “Now You See It” in all of the current metrics we use in our business. To directly address your question and quantify for you an improvement situation.

We run a mail-order distribution warehouse where units of inventory are picked into shipping boxes. We recently used “new” methods of data visualization, specifically heat-mapping, box-plots, and Pareto analysis to visualize the frequency of visiting each “pick bin” and the aggregate walking distance to each location. Identifying the deficiencies by this visualization then allowed us to re-layout our warehouse and realize a 15% increase in productivity. The best thing was, the visualization techniques allowed me to explain very clearly to all shareholders in the process the Present State and Future State and how we were going to get there.

So there you go, actual results!

By Dawn of KeckCAVES, UCD. September 1st, 2009 at 8:33 pm

Also http://keckcaves.org

We have found that 3D visual analysis of complicated geoscience data provides dramatic new insights into morphology, dynamics, and processes. Good visualization tools allow us to quantitatively measure features that can’t be done numerically at this point. The main web site is one paper that documents this.

By Steven Co.. September 7th, 2009 at 6:56 am

Hi Stephen

I am working in Finance/Accounting and I can tell that using visual to present Data has a dramatic impact on the communication. I am working toward Germany and with the Language gap sometimes you really have to be careful on how you communicate your data. I am using your work, Jakob Nielsen, Steve Krug, Garr Reynolds to help me getting the most efficient message.
I think that in every technical field people feel they don’t need to communicate properly, cause they like analytical stuff they don’t have the skills to use visual etc….

It’s a real pain and I am sure that in Finance like other technical field we would really benefit if people could learn how to better present data. With information overload I can only see good prospect in the field of data visualisation.

By Mark Schifferli. September 7th, 2009 at 3:57 pm

I wish I could publish how my company employs information visualization techniques, but we use these custom tools internally, and consider them such a competitive advantage that we decided not to share them. Suffice it to say that our visualization tools help with both analytic exploration and prioritizing near real time operational decisions, and have greatly improved the effectiveness and efficiency of the teams for which they were built.

By Ajay. October 2nd, 2009 at 3:30 am

Stephen,

Have good decisions been made because information was presented
visually?

Yes. I had to dig a little for this one (and I realize that response is a bit late) but here’s what Nicholas Kristof, New York Times columnist had to say (see quoted) -

“In september I traveled with bill gates to africa to look at his work fighting aids there. while setting the trip up, it emerged that his initial interest in giving pots of money to fight disease had arisen after he and melinda read a two-part series of articles i did on third world disease in January 1997. until then, their plan had been to give money mainly to get countries wired and full of computers.

Bill and Melinda recently reread those pieces, and said that it was the second piece in the series, about bad water and diarrhea killing millions of kids a year, that really got them thinking of public health. Great! I was really proud of this impact that my worldwide reporting and 3,500-word article had had. But then Bill confessed that actually it wasn’t the article itself that had grabbed him so much — it was the graphic. It was just a two column, inside graphic, very simple, listing third world health problems and how many people they kill. But he remembered it after all those years and said that it was the single thing that got him redirected toward public health.

No graphic in human history has saved so many lives in Africa and Asia.”

Regards,
Ajay

By Ladi. October 13th, 2009 at 11:47 am

Steve,

I think the answer lies in the semantics. Data visualization is meaningless. What is being visualized is information.

I’ve developed a travel dashboard for a pharmaceutical company that tracks travel activities. The travel dashboard shows late booking, advance booking, one day travel etc. The fact that employees are aware of this dashboard, unnecessary traveling were greatly reduced. This resulted as a cost saving for the organization.

Ladi

By Stephen Few. October 14th, 2009 at 9:17 am

Ladi,

Thanks for sharing your example. Assuming that I correctly understand what you’ve said, I don’t agree that the problem is one of semantics. I believe that framing it as a problem of semantics diverts our focus from what really matters. I use the terms “data” and “information” interchangeably. Distinguishing these terms in a manner that makes “information” different from and superior to “data,” is little more than a semantic exercise–it solves nothing in and of itself. Regardless of what we call this stuff that we analyze and present–data or information–what matters is that we understand it and communicate it clearly and accurately. Many business intelligence software vendors claim the ability to “turn data into information” without actually doing anything to bring the stories that live in the numbers to light. They offer nothing but hollow manipulations of semantics–a common practice in marketing.

By david marcus. June 6th, 2010 at 8:00 pm

bugger! just lost my post so here is a terse version:

semantics/meaning is important. the big issue is what is ‘presentation’ and what is ‘analysis’ - they are often impossible to separate.
One example: a few years ago i worked on a review of australia’s quarantine system. there was ‘data’ around on how many items were detected coming into the country but no information on whether this level of detection was good or bad. we did some analsyis and found a lot of variation. we chould have used tables but instead used very simple charts, eg http://dl.dropbox.com/u/126571/quarantine%20chart.jpg

See also other charts in chapter 5 at http://www.anao.gov.au/uploads/documents/2000-01_Audit_Report_47.pdf

By the time the report was public the government had announced an extra $600 million for quarantine. In my view the charts were just too powerful inshowing gaps in performance. But, it was the analysis underpinning them that stopped them being dismissed.

on a similar point, presentation can usually just make it easier to see what is going on. See my humble example of a heat matrix at Y:\Tessera\Tessera Website\Demo Models\naplan schools testing visualisation.html . Mind you, anything is better than indented tables.

CHeers

by the time the

By Mel Stephenson. August 12th, 2010 at 1:31 pm

Stephen

I too was quite astonished by how little was out there on this topic. The effect of data visualisation and data analysis has certainly been unbelievably transformative on my very conventional, small company (no false modesty here - there’s only 12 of us). Without raising prices and with only a modest increase in turnover we’ve managed to double net profits every year for 5 years. Everyone in the team is better paid and having that live visual performance feedback has helped bring real harmony to the team.

As a consequence I asked Tableau if I could present at this year’s 2010 Tableau Customer Conference specifically on the topic of the cold, hard cash value of investing time and money in data visualisation. To my pleasant surprise they agreed and I’m presenting Tableau Server - The Ultimate Profit Driver on the Tuesday. I’d be naturally flattered if you could come along to hear the presentation and perhaps offer your views on what we’ve done to date. Perhaps some aspects of our experience could be useful to cite for those ever curious journalists.

Keep up the great work - your well-thumbed books cover our offices!

By Jan Willem Tulp. June 17th, 2011 at 1:21 am

check out Discovery Exhibition: collecting visualization success stories (visweek 2011) http://www.discoveryexhibition.org/pmwiki.php