Lately, I’ve been on the lookout for modern BI. It’s not as common as BI vendors would have us believe. One key is intuitive visualizations. When Matt Chambers, data architect at Clemson University, framed his data problem in terms of “data storytelling,” he had my attention. Visual data is powerful – but visualizations that actually help people to do their jobs better are a rare find.
Chambers learned that the hard way:
One of the things I’ve really been working on in the last two years is the data storytelling component, like being able to walk someone through a visual narrative and help them understand what they can actually take action on.
But even a skilled data analyst can get into trouble with visuals:
I come from a really strong technical background, so when I first started working in data visualization and analytics, I would build these really advanced dashboards or tools and then just expect people to use them. Upper-level management doesn’t necessarily want to drill through a dashboard. They want [visuals] that are easy to understand.
How to start small – but get buy-in
Chamber’s journey towards data storytelling at Clemson is about starting small with Tableau software, and growing adoption to the point where an enterprise license was needed:
I had to prove that Tableau was a viable solution. Over time, we were able to build it up to the point that the EVP funded the enterprise license. Right now, I’m up to about 200 consumers of data on the system.
So how did we get here? Rewind two years, when Chambers initiated a BI tools evaluation. Why? The classic problem:
We have a business intelligence team that’s in central IT. Most of what they produce are transaction-based reports, like line item-based, non-visual reports. Most people would go and download for raw data, and then build their own visuals in Excel, or whatever tool their specific department used.
Chambers, who works in finance, supports all seven colleges within Clemson University. What became data storytelling began simply: with the desire for a more appealing format. Chambers:
We wanted to be able to provide a standard platform for all of the colleges. It’s the same dashboard, but it’s pulling your specific college’s data. That way, we don’t have the issues where individual units are pulling data from data warehouses, and trying to build their own reports. Who knows that they’re pulling the right data elements? There’s just so many issues with everyone trying to do their own thing.
The search for a data visualization solution
Chambers led a tool evaluation that included Qlik, Pentaho, and several others. Why did Tableau stand out? First, the amount of connectors, with plenty out of the box: “I don’t remember anyone else having a connector for Splunk at the time.” Second, Chambers liked the ease of use and extensibility. Chambers did not find it difficult to create custom visualizations. Tableau’s Story Points functionality looked like a good fit for data storytelling.
BI instigators could learn a lot from how Chambers built up. He started small, but with a pretty influential group: the first Tableau project was for Clemson University’s Board of Trustees. Chambers knew the Chairman of the Board wanted better reporting. Game on:
The Chairman requested a little more transparency into the financials and also just to make it easier to understand. Before, we would send out a crosstab view of the budget versus the actuals report. It wasn’t necessarily easy to understand. We actually took that, and rebuilt into a dashboard. Now it’s obvious where the biggest variances are. You can look at it in five seconds, and figure out if there is an issue with federal funding or something like that.
Chambers found a viral way to build more Tableau support: Share the visual. Chambers:
We debuted it at the Board of Trustees meeting. Some of the deans were there from the colleges. They all started expressing interest in having something similar. They wanted something that was standardized that all of them could be able to use. After we launched the Board dashboards, we moved to the dean and division dashboards on the financial side.
Data visualization in action – sample screens from Clemson
Though Chambers couldn’t share mission-critical images with us for obvious reasons, he did share a couple of sample screens with cleaned data. This overview of Clemson financial performance by year and category gives an idea of a top-level dashboard prior to any drill downs:
The caption on the bottom right notes some attributes of Clemson’s year-to-year growth. The next graphic is another top-level view, this one showing the flow of majors after a school re-org:
“They can access the data story whenever they want”
New BI tools are nothing without adoption. I asked Chambers if he faced any resistance.
There really wasn’t a lot of pushback because it made things so much easier than before. They used to would have to go into multiple different systems, run queries that take 5-10 minutes to run, pull all the data into Excel, and try to aggregate it into something that made sense. It was just a huge undertaking for them.
They literally just go to the dashboard, and all the new data pushes in at the close of business every month, so they can see everything there. The way we built it, we give them a lot of information. We also give them a lot of different ways to drill down or filter things, so they can answer the questions that are important to them.
User feedback has been good, from the top down:
We’ve had great feedback. The Board loves it because they can get access to it every month now. They used to only get these cross tab reports every quarter. Now, they get it every month – we push in the projections, the budget, the actuals. They can access the financial health – the data story – whenever they want… They’re really excited about it. So far, since we’ve been debuting the financial dashboards for the deans and division heads, they love it too.
BI comes down to adoption. No need for guesswork:
We can see if they’re actually using the dashboards when we log in the server. It’s being used very frequently. We’re excited about the adoption and how much people are using it.
Given Chambers’ success to date, I wanted to get his advice for other organizations struggling with “data storytelling.” One key: design for your audience. Different users want different levels of drill down:
I think the biggest thing is knowing the audience. If it’s executive level or something like that, odds are they don’t have time to sit down and click through a dashboard. But a line manager or someone who’s more in tune to the business will use that kind of tool.
Chambers went into detail on designing for levels. For example at the dean or business office level, they probably want the ability to filter for different departments, clicking/drilling into data discovery mode. A business analyst will want to take the dashboard even further, downloading it and making their own version of it.
They’re exploring the data at a different level of granularity. Based on the hierarchy of the organization, you can design the tools to support that role, basically.
One more tip from Chambers: strike a balance between complexity of information and ease of use. Clemson’s team designs for the 80 percent of queries to help keep complexity in check.
The wrap – “exploding adoption” is fueling analytics growth
Chambers’ work is just beginning. There are plans to expand his team given the “exploding” adoption. Human resources dashboards are next, followed by space and facilities management. Chambers will also be working on custom consulting projects for departments. These wouldn’t be part of the regular dashboards. They would be designed for a specific data problem or goal, such as improving retention.
Perhaps the only team at Clemson not yet in line is the football team. Fresh off their national championship thriller over Alabama, their dashboards can surely wait.
Image credit - Tableau screen shots provided by Clemson University, with sample data. Feature image - Whats Your Story Sticky Note © Ivelin Radkov - Fotolia.com
Disclosure - Diginomica has no financial ties to Tableau. Their internal PR has been getting us introduced to use cases, which we are always glad to report on.