What is it?

This BOBI award is given to a piece of work that provided new insight, direction, visualisation or revitalisation of current business practice.

Winning Entry

A LOAD OF NEW BALLS – creating an interactive online data-visualization for respondent profiling  (John Aitchison & Martin Conroy, First Line Research)

Judges’ Comments: Easy to read and interesting. Addresses a need that’s definitely increasing with clients, especially Marketing. An innovative approach with a clear application & all done with no budget – wow!

Data-visualizations are all the rage. Visually playing with data allows people to quickly spot patterns and get a feel for meaning. We noticed that the better examples were online and interactive, and that PowerPoint and the like couldn’t deliver the same impact or flexibility, which we found frustrating. With this in mind, we set about building our own online interactive data-visualization. We identified an idea that was a good fit with our type of work and set about learning new skills and hacking about with code. The resultant “balls” technique is now a popular part of our deliverables.

I remember the bubbles very well. They gave a very clear image of the different segments we were analyzing. Personally I’m a fan because they tell much more about the data than one simple view, adding another dimension which is dynamic. I liked being able to see which patient populations belonged to which segment. So much easier to understand than just the numbers on a chart. (Market Analyst, Roche).

You probably know marketing colleagues who suddenly get busy on their smartphones whilst researchers “talk numbers”, but happily re-engage when a colourful chart hits the screen. Behavioural studies have shown that rich visual content drives engagement much more effectively than text.

Neuroscience tells us that over a third of our brain mass is given over to processing the data we take in through our eyes. About 90% of the data our brain receives via our senses is visual, and we process it 60,000 times faster than text.  In summary, the potential for data visualization is to effectively communicate large amounts of information, quickly, to lots of different people!  The success of books like “Information is Beautiful” by David McCandless, the proliferation of infographics in mainstream print media, and the sheer range of online “data-viz” executions made us want to break out of the confines of chart libraries.

To be honest, as a pharma client I rarely see any new analysis techniques, the final outputs that agencies deliver are pretty much exclusively in PowerPoint  (Market Research Manager, Sanofi)

We are not ‘big-data’ practitioners but we do deliver data-rich market research projects to clients and had been encouraged by how well received one particular ‘compound doughnut’ slide had been. It was highly visual, animated, and conveyed a core message. At the time we noted how it held audience attention. We later discovered that the slide had been reused often within the company, and even made it to Board level. Thus motivated, we started thinking about how to create different, hopefully even better, interactive data-visualizations.

Knowing that we’d have to invest precious development time (we are a four-person company) we wanted to identify a technique that was reproducible. We also had a hunch for something that could visually represent alternative views of respondent profiles, which would suit our type of projects. For example, in a segmentation analysis that describes and quantifies different types of patients, we might typically present the outcomes as a glorified table – adding size, shape, colour and language to highlight significant differences between segments. That’s fine but it is static, and makes for a busy slide.

We wondered if a data-visualization could do a better job.  We recalled seeing a ‘bubbles’ execution online where the objects were different sizes and colour-coded, but static. And we’d seen another where dots moved about the screen in response to the user’s choice of axes. So we investigated the possibility of adapting from them both and building what ultimately became this ‘balls’ execution! We soon realised that such a technique would only work with discrete data. In effect, each respondent in a data set would need to be allocated to one ball only. Which essentially meant that multi-code questions were out. This seemed initially like a big blow but actually pointed us toward a respondent profiling application even more clearly because, by design, segmentation and typing techniques allocate respondents to one category only.

Now we needed to build a proof of concept, so we dusted off our JavaScript, enrolled on some online courses, and set about playing about with various bits of open-source code from around the internet. We had frustrations, and fun. For example, the code that governs the movement of the balls was adapted from an open-source D3.js model, and we tinkered with speed and “gravitational weight”. The right settings create a pleasing effect when changing axes, but most settings produce either speeding small balls, or immovable large balls, or even balls that gravitate to the wrong category! It took us about three months (February to April 2015) to create the basic data-visualization template that underpins the examples shown. We started incorporating it into client results presentations a few months later, following a few technical tweaks.

We’ve prepared a three-minute online video introduction to the balls: http://flres.uk/balls-dataviz (opens on YouTube, 3 mins).

Here is the text-only instruction on interacting with the balls:

  • Each ball represents a unique combination of respondent attributes
  • The larger the ball, the more people with that unique combination
  • Hover your mouse over a ball to see the values of each attribute for that group of respondents
  • Clicking a menu button at the foot of the screen (or the top left of screen in the second example) changes the dominant variable, and the balls (i.e. unique groups of respondents) move about accordingly.
  • Chop and change menu selections to discover new relationships.

We have used the balls data visualization in three projects to date. The first was developed from our original proof of concept and was used as a demonstrator. The other two were incorporated into commercial projects, as additional deliverables, to seek client reaction and guidance as to usefulness. One of those projects we are able to anonymise and use as an example, the other we are not for reasons of commercial sensitivity.

NB – we have had to remove some examples, in this blog version, due to client confidentiality.

Example: Sanofi, diabetes
: In late 2015 we worked with Sanofi and their communications agency to survey n=500 diagnosed sufferers of a particular condition. We needed to summarise the breakdown of patients according to disease type, gender, age, treatment approach, and specific treatment type. This objective presented an ideal opportunity for the balls data-visualization.

Example: for Conference:  In early Spring 2015 we developed a proof of concept which was subsequently developed into a demonstrator. It dovetails with work we do on making better use of survey meta-data, and attracted attention at Conference last year when we showed it on our stand.

I’m personally very proud of what we have achieved with this data-visualization and consider it a big success story for our small research consultancy. It has been delivered with limited human resource, on a zero budget, and has challenged our creative and technical capabilities.

I’ve just taken a look at the visualisation tool and it’s really neat.   I am certainly open to using this tool to look into the data further – the bubbles easily differentiate category sizes and relationships (Market Research Manager, Sanofi)

We have made a flexible tool that is not to be found in ready-made commercial packages or charting libraries, and that we hope will serve us well in years to come. And it has pleased clients. There have been “Ooohs” as the audience watch the balls skit about the screen for the first time, followed by “Aaahs” when they appreciate the added-value that the visualization brings to their speed and breadth of understanding.We believe we have created a useful and innovative deliverable as well as, we hope, another means by which researchers can keep marketers away from their smartphones during results presentations…