Roopak Gupta surely isn't the only data technologist who credits a fascination with computer games for his life-long interest in computer science, but not all gamers have gone so far as to build their own digital diversions. For Mr. Gupta, VP of data at Krux, it was a handheld Tetris-like parachuting game that propelled him towards learning basic programming language.
"I was so intrigued by the parachuting game that I taught myself the basic programming language in an effort to build the game myself," he said, reminiscing about his age-13 obsession. The game's mission was to save people falling from helicopters.
"I think gaming in general helped me focus, concentrate and get excited about computer science. I knew the best way to achieve this at that time was to study hard and pursue a career in the sciences," added Mr. Gupta. "The academic system in India required students in high school to pick between Commerce, Arts and Science and for me the choice was obvious."
He went on to study computer science and management at the Indian Institute of Technology in Delhi.
Before 2010, when he joined Krux, which provides cloud-based data management platform services, Mr. Gupta spent around a decade with Rapt, a firm that helped publishers manage ad inventory; Rapt was acquired by Microsoft in 2008.
Today, working from Krux's San Francisco offices, he helps site publishers create and process audience segments, and provides reports showing steps to help clients take action on their CRM and third-party data.
Ad Age: How did your experience building games inform your work today?
Mr. Gupta: That experience started an exciting journey of learning, knowledge and innovation that has continued to this day and has allowed me to build several complex software systems that are used by folks all over the world. I am quite proud to say that I still pursue every project regardless of size with the same amount of passion as my 13-year-old self. In my own way I'm still trying to save all those people jumping from the helicopters.
Ad Age: What's the difference between simply showing data and actually helping clients take action on that information?
Mr. Gupta: The key to making data actionable is building application programming interfaces -- APIs -- and services that allow the data to flow into any system that can make use of the information in real-time. We have a lot of clients who are putting that data to work on two fronts. First, they're recognizing a user as they come to their sites and then immediately delivering a unique, personalized experience. Second, they're finding users of interest to deliver relevant brand advertising to both on- and off-property.
It takes real-time web service APIs to customize content on websites based on audience segments, and to deploy marketing campaigns for increasing brand awareness (while also driving additional subscriptions and registrations). With the right technology and data-management platform, publishers can surface insights into which users are more likely to subscribe, for instance, and then turn those insights into action by reaching them via direct media campaigns informed by the publishers' user data.
Ad Age: There's a lot of talk about the distinctions between first- and third-party data. But how does second-party data fit in?
Mr. Gupta: The market is now tilting heavily towards first-party data, or data collected on business owner's properties, but with augmentation from second- and third-party data to accomplish scale and deeper insights. Second-party data is the data shared on peer-to-peer connections between trusted partners. Today's data-management technology allows publishers and advertisers to add, remove or update any data sources with no changes to core infrastructure.
It's important for advertisers and publishers to choose a technology partner that enables them to capture data from different sources and streams without limitation and lets them analyze that data using supervised and unsupervised machine learning techniques. The core idea for the business operator is to adapt easy-to-use technology so that you don't have to hire an army of data scientists to make the data understandable, scalable and actionable.
Ad Age: What's one of the most interesting data sets you've worked with? Why?
Mr. Gupta: It's a hard question to answer because I am really a data technologist, not a data scientist, and I don't spend a lot of time analyzing segments. But our clients bring us great stories all the time. For example, one of our clients recently hypothesized that a particular segment of their users was generating the highest number of clicks across all of their campaigns and advertisers. After they dug deeper it turned out that the high-clicking category did indeed exist, but it was comprised of users with very different attributes than the young male audience segment they expected. The new insights enabled the client to build an entirely new segment of users that was in fact much larger, more relevant and delivered a higher CPM. This relatively new ability to discover and make actionable audience behavior and intentions in real-time is at the core of what DMPs are bringing to marketers and publishers.
Ad Age: You mentioned that you're a data technologist rather than a data scientist. How do you distinguish between the two?
Mr. Gupta: A data technologist designs and builds the plumbing so that the various technology pieces communicate with each other via APIs and are able to collect, manage and activate the data. A data scientist focuses on what the data is saying in the areas of audience discovery, intent and value.
Ad Age: Since the NSA scandal there are concerns emerging among overseas companies and governments regarding use of U.S.-based cloud services. Has that affected your work at Krux?
Mr. Gupta: Yes -- we have heard concerns from our global clients, particularly in the EU but also in Asia as well. Fortunately we built our data-management platform with an initial feature that focused on consumer web-data protection. We saw early on that the digital ad ecosystem had become awash in questionable sources of consumer data, data that was collected by violating the privacy of the user, or the business model of the site owner, or both.
The key for success in this area is to have a flexible technology stack, and to be able to deploy it appropriately in each country or region that you do business. For example, German law deems IP addresses personally identifiable information, so in all of our European data centers, we scrub IP addresses from log files. In this way we're able to leverage the cloud services pioneered in the U.S. to serve our international clients, but in a way that adheres to their standards for privacy and data protection.