Dr. JT Kostman is a Data Scientist, Mathematician and Psychologist – and the Chief Data Officer for Time Inc. Over the past 20+ years JT has provided data-driven insights into human behavior for organizations ranging from Fortune 500 companies to U.S. Intelligence Agencies.
At Time Inc. he leads a rapidly growing global organization of Data Scientists, Data Miners, Strategic Analysts, and Digital Media Marketers who are developing cutting-edge capabilities that are revolutionizing Digital Media Marketing, Data Mining, Predictive Analytics and Consumer Insights.
We talk to JT on how a leading publishing house has successfully adopted analytics and how is India being leveraged within this initiative.
AIMAnalytics India Magazine: Thanks JT for this interview with us. Tell us more about Analytics at Time Inc.?
Over 100,000 people subscribe to our newsletter.
See stories of Analytics and AI in your inbox.
JTJT Kostman: Our CEO, Joe Ripp, has been steadfast in his championing of data and analytics at Time Inc. He has made quite clear to our executives, employees, and investors that data and analytics are central to our value proposition – and that data and analytics should be at the core of every decision we make and every action we take. Data is considered to be both one of our most valuable assets and also a catalyst for creating value in all areas of our business.
To avoid the frustrations and obstacles that are most common to cohering, integrating, and maximally leveraging data, our CEO likewise made clear to the organization that the CDO’s team owns all the data at Time Inc., including all data generated or provided by our various business units, partners, and vendors. Where in most organizations there is a constant battle for the CDO to be able to gain access to the data analysts need to succeed, at Time Inc. the CEO has resolved that challenge definitively.
AIM: Analytics at a leading publishing organization appears like a novel idea. What is the need of analytics at Time?
JT: Publishing has historically been a somewhat polarized profession, with an artificial demarcation between journalists on the one hand, and the business folks on the other; a structure we lovingly refer to in publishing as “Church and State”. When Joe Ripp became CEO of Time Inc. he changed that model – and as a result, we are fundamentally changing our industry. The only metrics publishers used to care about was circulation and profit on the one hand, and the number of Pulitzer prizes and similar awards on the other.
AIM: ok, and what are some of the analytics solutions that you work on?
JT: Our group is intimately involved in nearly every critical aspect of the business. The areas we support include, but are not limited to:
- Evolving our AdTech capabilities through sophisticated audience analyses, segmentation models, programmatic capabilities, and the ability to better understand the needs and interests of our audiences;
- We are presently developing a Data Driven Editorial capability that will help our editors and journalists better understand their audience and evaluate the extent to which our content resonates with various audience segments;
- Our team has done considerable work in modeling the behaviors and preferences of subscribers, newsstand purchasers, and online audiences to develop audience insights that transcend any that are otherwise available in the market;
- We have likewise developed metrics to help inform numerous operational aspects of the business for the purpose of increasing profitability and decreasing expenditures.
AIM: Can you brief us about a specific use case in analytics that has brought significant value to Time Inc.?
JT: We recently developed a rather sophisticated Audience Segmentation Analysis tool that we were able to beta test with a major advertising client. By incorporating some of the client company’s data into our analyses, we were able to develop models that proved extremely valuable in identifying previously unknown (and unexpected) insights into their market. These findings led to an increased advertising commitment to us of over a million dollars (US) – and this proof-of-concept beta test, which took us less than three months to develop, and only a few hours to implement – has already also served as the foundation for other similar, and even more lucrative, opportunities.
JT: I am a Data Scientist, Mathematician and Psychologist by training – and the work I have done has been at the nexus of those fields. The work my group does is similarly highly influenced by this perspective. We consider not just the data patterns, but their implications for understanding behaviors. Similarly, in developing target models we go beyond focusing on simple superficial behaviors like clicks and transactions and take more of a cognitive-behavioral approach that allows us to more effectively describe, understand, predict and influence behaviors. We are in the process of developing patentable solutions that will give us the ability to gain unprecedented insights into our audiences.
AIM: Please brief us about the size of your analytics group and what is hierarchal alignment, both depth and breadth.
JT: I am relatively new to Time Inc., having joined as their first Chief Data Officer only a few months ago. The team I am building consists of both existent resources and new hires. We have groups that are focused, respectively on: AdTech; DB Management; Data Aggregation and Augmentation; Data Governance; Traditional Analytics/Modeling; and a Data Science team focused on advanced analytics. The teams are located at several locations in the US (primarily in New York) and in Bangalore. The overall headcount is presently in flux; we are continuing to incorporate additional small teams across Time Inc., but the targeted size will be approximately 150-200 people, which we should reach within the year.
AIM: What kind of knowledge worker do you recruit and what is the selection methodology? What skill sets do you look at while recruiting in analytics?
JT: My primary criteria are pretty simple: I want people who are (very) smart, highly numerate, technologically facile, extremely curious, and easy to work with. I have very little patience for poor attitudes or arrogance. Our team members are already some of the best in the business – and if they can maintain some sense of humility, so should anyone who is new to the team. I firmly believe we should all be able to learn from one another and continue to expand our capabilities.
As to the particular skillset, I am pretty flexible. We have as much room for traditional SQL and DBM skills as we do for those who are comfortable with the Apache suite of solutions. While I expect the MapReduce paradigm will continue to predominate – and Hadoop (et al) skills will quickly become required simply to enter the profession, I expect we will also see the need for analysts to increasingly improve their mathematical abilities. I’ve found that even a basic knowledge of logic, set theory, and discrete mathematics can turn an otherwise simply competent SQL analyst into a highly valuable asset.
As to analytic tools, I tend to let analysts pick their own, to the greatest extent reasonable. I have an entire group who works exclusively with SAS, though I personally tend to prefer the flexibility of R and Python; these are skills that tend to impress me more than the ability to simply point-and-click through an off-the-shelf solution.
Our selection methodology doesn’t presently involve any coding challenges, analytic problems, or psychometric assessments; though I intend to eventually evolve to using more objective evaluation tools. Presently we rely on an assessment of our candidate’s background and a series of Behavioral Event Interviews conducted by peers, supervisors, and internal clients. During the course of these conversations we ask candidates to describe particular experiences they have had working with various datasets and solving real-world challenges.
AIM: What according to you are some of the most significant challenges you face within the analytics space?
JT: The greatest challenge has been a lack of qualified talent; there simply are not enough truly capable analysts to get the work done. This situation has, unfortunately, been exacerbated by several strategies that were potentially intended to alleviate the problem.
In the US and UK we have seen a cottage industry spring up to mass-produce Analysts and Data Scientists. Luring kids in with the promise of big salaries and sexy jobs (a cover story of the Harvard Business Review actually called Data Science the “Sexiest Job of the 21st Century”), they are taking what are often potentially promising people and giving them questionable skills that leave them under-qualified – and wastes both their time and ours as we end up having to sort through increasing piles of resumes to find anyone with real skills.
Many of the data solution companies are focusing their efforts on building products that can be operated by “Business Users”; the polite euphemism they use to describe anyone who doesn’t really know what they’re doing analytically. Packaging their solutions with claims like it being a “Data Scientist in a box” and contending that companies will have “no need for expensive analysts” once they purchase their solutions, both misrepresents the importance of hiring experts who are able to meaningfully interpret analytic results – and ultimately leads companies frustrated, believing it is an analytic approach, and not the tools, that are failing them.
AIM: How do you see Analytics evolving today in the industry as a whole? What are the most important contemporary trends that you see emerging in the Analytics space across the globe?
- The MapReduce paradigm and NoSQL approaches will continue to predominate as traditional RDB thinking continues to drive toward obsolescence.
- We will increasingly see what we now consider to be “advanced analytics” become fundamental to virtually all medium and large businesses that want to remain competitive. What is now seen as an analytic advantage will increasingly become a matter of survival.
- As much of an impact as Moore’s law will continue to have on processing power and speed over the next few years, those benefits will increasingly become asymptotic; and I suspect those benefits will continue to be dwarfed by the advances that are simultaneously being made in algorithm development. Solving a complex problem more simply will always trump the ability to race to a less elegant solution. Many of the techniques that are now part of the analyst’s basic toolkit would have been impossible even a few years ago, but for the speed at which the analyses can be conducted – but as fast as processing power continues to increase, the best tool in the analyst’s arsenal will always be that grey squishy wetware they keep between their ears.
AIM: What are your thoughts about Analytics in India? Where does it figure in your whole analytics strategy?
JT: I have spoken at length in various venues across the US, from Silicon Valley to New York, about the pervasive sense of “Digital Colonialism” that has come to taint the capabilities and relationship the West is becoming increasingly reliant on India to provide. Presumably out of some sense of frustration and desperation at not having sufficient talent in the US and UK, firms have increasingly been outsourcing their analytic needs to India; but they frequently have been reserving the most interesting, challenging, and frankly fun parts of the work for their teams “back home”. They tend to forget that some of the most talented analysts in the world are coming from India.
IIT’s was recently ranked third in the world among the best technology universities; their graduates are among the most talented analysts I’ve ever worked with, as are so many of the other Indian analysts I have had the privilege of working with over the past 20+ years. Relegating such remarkable talent to the less interesting parts of the work – and the arrogance of thinking that high-level analytic thinking should be reserved for the West – treats India as an “analytic sweatshop” and only serves to drive away the best and the brightest. While several firms have built their business model on precisely this approach, I believe this sort of shortsightedness inevitably harms the profession, our interests, and the market.
AIM: Anything else you wish to add?
JT: The future of analytics is still up for grabs. Given the remarkable progress India has made over the past few decades as a digital and data powerhouse, I firmly believe that future can be based in Bangalore and Hyderabad. Whether or not that comes to be will ultimately be decided by the next generation of Indian analysts – and the extent to which they accept the challenge and come to believe India is the future of analytics: Tat tvam asi. Though art that.