Antivia has been talking and writing a lot recently about our vision of Universal BI, where interactive dashboards are the perfect tool to allow everyone in an organization to get exactly the information they need to do their jobs. So, last week, at SAPPHIRE NOW in Orlando, it was great to see a real-life example of Universal BI in a presentation from FedEx Office.
Andy Mills and Charlotte Huff discussed their SAP BusinessObjects Dashboards (aka Xcelsius) implementation and the transformational effect it had on the running of their 1,800 stores where previously “too much data” had been the biggest issue. The presentation finished on a quote from their COO Kim Dixon:
“I consider ICE a true game-changer for FedEx Office field operations. It provides great visual dashboards of key metrics. With ICE, it’s easy to spot areas of strength and weakness so you can take quick action to improve performance. This really is the next generation of reporting and managing our business”
A perfect way to sum up the promise of Universal BI.
In a future post I will explain why FedEx Office’s implementation fits the Universal BI vision so well, but here I want to discuss several comments made on Twitter after Anita Gibbings posted the following picture of one of the dashboards, during the session:
The comments were:
1) “Sorry, shocking data viz”
2) “Oh dear”
And in response to the idea that this was an effective dashboard:
3) “no way”
Given that this picture is entirely indicative of the successful dashboards, which have transformed the way FedEx Office’s store managers run their business, what could have provoked such a strong negative reaction from a single screenshot?
I suspect the answer lies in the use of gauges and possibly (but less likely) the fact that the charts do not start at zero, both of these are widely regarded as enemies of good, effective data visualization.
Now, I understand the academic reasoning behind the issues people have with gauges and non-zero based charts (perhaps a subject for another blog), but that does not mean they are always the wrong thing to use.
Indeed, as I believe this example shows, they are sometimes exactly the right thing to use.
Remember, this system was deployed to 1,800 retail stores, to users who are responsible for running an efficient store, users who had previously struggled with having to battle through “too much data”.
In the spirit of Mico Yuk’s advice to “Make ‘User Adoption’ your ONLY KPI to Measure BI Success”, I suspect the biggest barrier to effectiveness facing this system was people not using it at all. I also suspect that using more sophisticated visualizations (e.g. bullet charts, which some regard as “always better than gauges” but are often unfamiliar to end-users and so are off putting to them) would have had a significant impact on adoption.
It is often a case of “familiar is better than more efficient”, particularly when the speed at which people absorb information from a dashboard is almost never a critical factor in management-style dashboards.
So, Andrew and Charlotte at FedEx have a successful dashboard rollout, a strong relationship with their users and a BI base on which to build. Their future may include dashboards with “more efficient” visualizations, and if so, they are much better placed to introduce them with this success already under their belt. If they had focused too much on visual efficiency, too early, then things would probably have not have worked out as well.
To put it more bluntly, employing the mantra “adoption trumps visual efficiency” will see you gain much, much more than you lose.
I would be interested to hear from those who disagree, specifically:
- What would you do to improve the dashboard above ?
- How would your proposed changes make the dashboard materially more effective?
- Why are you confident your change would not have impacted user adoption ?