Kevin Sonsky is senior director of business intelligence at Citrix Systems. During the past 11 years, Sonsky has implemented an enterprise-wide self-service reporting environment that has delivered deeper insights into customer purchasing behavior. At the same time, he has established a grassroots governance program that has successfully standardized on dozens of key enterprise metrics and reports.

In this 30-minute podcast, you will learn the secrets to Sonky's success:

• How a BI team of 3 people can support a governed self-service environment
• How to lay a foundation for governance in a decentralized organizational structure
• The role of executive support in fostering governed self service
• Where to begin a data governance program and with whom
• The value of placing a watermark on certified reports
• The role of business and IT in data governance
•  How develop a standard for self-service reporting tools
• How manage the proliferation of data extracts needed to support a self-service environment
• How to support a community of self-service BI users 

This is an excerpt from the podcast interview between Wayne W. Eckerson and Kevin Sonsky. 

ECKERSON: Our guest today on our inaugural show is Kevin Sonsky, Senior Director of business intelligence at Citrix Systems where he has run the BI program for the past 11 years. During that time, he has implemented an enterprise wide self-service reporting environment using Tableau that has delivered deeper insights into customer purchasing behavior. He and his two-person team have established a grassroots governance program that has successfully standardized on dozen of key metrics and reports used throughout the company. According to Sonsky, it takes years of perseverance, strong executive support, and user-friendly technology to achieve business intelligence success.

ECKERSON: A long time ago you said that self-service BI requires a lot of handholding. Would you explain what you mean by that?

SONSKY: My comments at the time were attributed to both underuse and overuse. The one we feel the most pain from is overuse, which is why we have been running a formal and relatively successful standardized metrics and reporting program for the last year. Underuse certainly requires the introductory training and handholding as well, but a lot of the high handholding is in the overuse area where we are showing very seasoned but somewhat silo’d teams the standardized definitions, data sources, and reports so that we’re all on the same page.

ECKERSON: What were the symptoms of the overuse that caused you and your team to take action?

SONSKY: We made substantial investments over the prior years and built up a repository of data and data stores in our environments, and it became a bit overwhelming to a lot of the users over time. We got to this situation where there was this chaos and sprawl of reports and metrics and analytics all over the place. It caused us to stop and reflect and bring organization to the chaos. We didn’t want to take away the self-service that let a lot of people do their jobs and be able to be close to the data, but we wanted to make sure that as you went up the management chain there was integrity and understanding and confidence in the numbers they’re seeing.

ECKERSON: Did Tableau contribute to the report sprawl and data chaos?

SONSKY: Well, these are the pros and cons of having a tool that you can pickup like Excel. It doesn’t require a lot of training and doesn’t require IT. It allows people to move fast, get their questions answered, and scale. The flipside is they can go off on their own, and there’s nothing to stop them. It’s one of those enabling technologies that you have to recognize may come at a cost. It’s ok as long as you can establish a program with guardrails so people understand that “with great power, comes great responsibility”.

ECKERSON: You were very successful at standardizing metrics and dimensions and even tools with a team of three. How are you able to corral people to start agreeing to standard types of data objects and things like that?

SONSKY: The accelerated progress we’ve seen the past couple of years can be attributed to tone at the top. The executive sponsorship, most recently at the CEO level, and new leadership across other areas has had a profound impact on the pace and direction of our BI governance efforts. So, all the work we were doing in the years leading up to it, we weren’t sure we were making an impact, even though we knew we were doing the right things. Then, when the new leadership teams demanded that we be data driven, make data driven decisions, and work off of a single version of truth, it really paid off. We had already laid the groundwork and had the foundation to accelerate.

ECKERSON: Explain what that foundation is. What actually did you build? What was in place when the new CEO came in?

SONSKY: Even when he came in, we had already started to establish, for example, less formal governance committees. We identified leaders representing each functional group who would be part of a team that could take part in standardization discussions and facilitate prioritization reviews and help work together on the BI strategy and even engagement with IT and technology as well. We had already established ownership for certain metrics and subject areas. So, some of those pieces were all in place, but before the new leadership they were only there if people wanted them. We could only push so hard. Once the new tone at the top was set, teams actively looked for those formal rules of engagement and tools. Folks then wanted to align with the rest of the company. It meant the engagement with us and the rest of the company became a lot more productive.

ECKERSON: So how did you get these guys to agree?

SONSKY: When the executive team sets a tone of aligning to make better decisions, it cascades down to the teams below that. Teams’ objectives now are to ensure they align with their counterparts and functional groups. My team can help serve as an objective facilitation, so we can understand each of the team’s perspectives and requirements, and come to a compromise.

ECKERSON: When people did come to agreement, what was the result? What happened differently?

SONSKY: The end was to develop a standardized report that reflected the certified version of that particular data point or metric and who owned it, and then actually certify it, publish it, and make it available. That didn’t take away all the self-service benefits. People could still do their own analytics, but we told them that their own reports and analytics needed to tie to an anchor report and use the same logic and methodology.

ECKERSON: So, you essentially certified a report. Did you apply a seal or watermark?

SONSKY: We came up with a little logo that goes on the header. This provides the consumer with the confidence that the data is trusted and has gone through the proper vetting. Not only is it meaningful for the consumers, but it becomes a forcing function for those involved in the governance process. When you are ready to stamp something it makes people think a little bit harder about how comfortable they are with something being out there, and that might reveal disconnects that you may not have uncovered if you didn’t formalize the process. It causes people to stop and think.

ECKERSON: Is the IT team or your team responsible for maintaining that report and making sure the data is accurately displayed in it?

SONSKY: No, for every report that we get published on there, we have a business owner. It’s not from IT and it’s not from my team either. My team doesn’t own data or the reports. The owner is responsible that the report is current and refreshed. As a governance body, we’ll go into the catalogue and make sure there is stewarding going on.

ECKERSON: Did you guys bring Tableau in or did it become the default tool?

SONSKY: Both. My team got the very first license, and we looked to it to solve a particular pain point. That problem was that the users were at the point where there was a lot of data and weren’t sure how to access or interact with it. Tableau solved this problem beautifully. It really came about by pure word of mouth and customer success. It was not driven by the top down or IT. We just wanted to make sure that as the growth was accelerating, we had a proper enterprise infrastructure and licensing arrangement to support the growing user base, and that’s when we really built out something that was more enterprise class.

ECKERSON: What are the keys to supporting a community of self-service users at Citrix?

SONSKY: You really have to make people understand the data well because even if you educate them on the front-end tool, if they don’t understand the data model and the data sets behind it, they can get into a bit of trouble. We do spend time working with new users and even some existing users to help them understand both the tools and the interaction with the data. We’re certainly not experts on all the data either, so we’ll point them to folks who know better than us. Then, it’s really about keeping those business relationships active. I think the more we stay connected to those subject matter experts within each of the functional groups, the better we understand the data, serve the next set of users, and share that knowledge across the company.

ECKERSON: If a new BI director came to you seeking advice on how they should do their jobs properly, what lessons or tips would you provide that person?

SONSKY: The first piece of advice I’d give is to try and know the culture of your particular company. Knowing where your company is on the spectrum, whether it be top-down, command and control culture or a loose, self-service one, will determine your approach and how far you can push in certain areas. In either case, the best piece of advice is to not boil the ocean at first. If you’re starting from scratch you want to find a good use case, but you don’t want to pick one that isn’t under dispute. Find one with a little hair on it because perhaps there’s more than one functional team that’s reporting on it, and they’re reporting differently.

I would also advise to engage IT early so they can get the business context for what may likely require their assistance to address. And agree on a very specific tangible deliverable as the output. Measure the success of the process by the output, and, at the end, communicate and document that final resolution, ownership, and success as broadly as you can. Hopefully, they’ll start coming to you, which is what we’ve experienced over time.