Rubicon Project is a leading technology company automating the buying and selling of advertising globally. Its Advertising Automation Cloud is one of the industry’s largest independent technology platforms used by more than 500 of the world’s premium publishers to transact with more than 100,000 advertising brands across a massive marketplace that powers more than 180 billion ad trades per month. We had the opportunity to interview Jan Gelin, Rubicon Project’s Vice President of Engineering for Automation Cloud who took us through a whole new level of sophistication when it comes to data driven operations.
So, why did you start using Geckoboard?
We're a metrics driven organization. We have a Greenplum data warehouse. We have a few petabytes of data in our back end data center, but we also are heavily a user of graphite. I think we have 250,000 or more graphs in graphite. I know, actually, down to each individual server, what revenue that server is generating. We of course have Microstrategy and we have big data warehouse, so when it comes to business intelligence and analytics, they do deep dives into data in terms of understanding what's going on. That's very sophisticated, algorithms and so forth that they use for that data. But then, for every week I was tasked with getting a generic overview of what's going on. So, I started that with the traditional way of doing Excel sheets, and some graphs, the amount of storage, roughly what things look like in production. So, after a little while, I got bored, I knew we should be able to automate this, so I basically shopped around for something that could take just data and display it in a fashion that I can show to people in the staff meeting every week. And this is how we found Geckoboard.
How a tool like Geckoboard is helping with your decision-making?
Since Rubicon Project is completely metrics driven, there is no decision making being made by the gut. We spend a lot of time on just analytics. Geckoboard is a tool that can give a very quick overview of how the business is actually doing, and quickly what's going on, if anything is happening. The big difference is, if you look at something like Microstrategy, we actually mail out about 60 or 70 reports everyday. I'm very, very deep in terms of everything that's happening inside of the company. Geckoboard is not the deep analytics tool. But it can give a quick overview about what's going on in terms of our overall QPS rate across all data centers. It's also a good tool to aggregate different data sources. I have everything from bash scripts running on certain machines, like one bash script is collecting how many cores are in all our data centers. And that just runs as a nightly job that collects like, basically /proc CPU info and creates a JSON and posts the JSON into Geckoboard. And then I have things like graphite that where it's using the graphite API to actually go into pull some of our more generic graphite graphs and post them into Geckoboard. I can show all this.
What are the challenges in building a data driven culture like yours?
In many companies you have these silos. You have the database team, you have the business intelligence team, you have the analytics team, and then you have all the executives that are counting on weekly or daily reports and they look at the data. What's required is that you need somebody that basically can just go in and say, “Okay, we need a quick overview of what's going on?” As soon as you show this, especially to the executives, they start to appreciate the value. Our CEO basically took this and basically had it on his iPad every morning, so he can see what's going on in terms of the overall QPS rate. It's basically a quick overview of how the business is doing. If you are a large organization you may need somebody that is familiar enough with SQL, or can use APIs and create the very first dashboard. It's not like companies don't have dashboards, it's just the way they are generated is very traditional, because data is considered to be complicated, it's all a database team, and there's an analytics team, so you need somebody that has a little bit of outside the box kind of thinking.
What would be your advice based on your experience, to any organisation that reads this blog post and says wow these guys have built a very robust structure, but how can I do this? How can I start?
Frank Addante is the CEO of the company, and at some point, the old mayor of Los Angeles was coming into the office, and he tasked the marketing team to display data all over the office. He was like “you guys, put up some nice data on these TV monitors”. The marketing team came over to me, and I actually created specific Geckoboard dashboards that were across the whole facility, that actually looks really awesome. I think that gave a lot of curiosity among other teams, on how do we get access to this data. You can always start with a very simple dashboard and help people get excited about data by making that data accessible and visible in the office. That is a good start.
Download the case study as PDF