How Data is Affecting Media, Advertising, and Entertainment Careers
Insights from some of the nation’s leading data scientists in the media, advertising, and entertainment industries
In advance of our upcoming event — Data Science Salon: Applying AI and ML to Media, Advertising, and Entertainment, we asked our speakers, who are some of nation’s leading data scientists in the media, advertising, and entertainment industries, to answer a few of our most pressing questions about the future of their industries. Read on for their insights — there’s some great advice in there!
What are some reasons a data scientist would want to move from another field into media/ad/entertainment?
“I’ve really enjoyed working in media because there are so many aspects of the company that data science can help with. I’ve been able to work on forecasting, operations research, user segmentation, natural language processing, content recommendations. Data science improves our readers’ experience with the Times but also helps with business concerns ranging from newspaper distribution to advertising sales. As the newspaper business continues to evolve with readers’ changing habits, I’m sure that the scope of our work will only increase.” -Anne Bauer, Director of Data Science, NY Times
“I think most data scientists are looking for a few key things in the roles they take and those are: interesting problems to work on, an abundance of data, and the ability to grow and learn new things. The media industry has more data available to it now than ever before and with that comes incredible opportunities to develop innovative ways to leverage that data for business impact. On top of that, the industry is changing at an accelerating pace as people’s media consumption habits evolve with the advent of new media platforms and technologies. In an industry that is changing as quickly as the media space, data scientists have to stay current with the latest advances in machine learning, analytics, and computing platforms to be competitive. This has created an exciting environment where someone with great analytical skills who is willing to learn the industry can have a tremendous impact.” -Bob Bress, Head of Data Science for Freewheel, a Comcast Company.
“It’s a quick changing field with constant evolution of user media habits that require research, creative thinking, and persuasion. The media space is a great place for a data scientist or analyst who enjoys a constantly changing environment that demands out of the box thinking.” -Wes Shockley, Senior Manager — Audience & Analytics, Meredith Corporation
“I think most people join journalism because they believe in the mission and potential of the media to do good. When working in this space you have the potential to create or support the institutions holding power to account and driving meaningful conversations and change. You have the opportunity to be of service to a variety of people looking for information and answers. It’s unlike anything else.
If that doesn’t capture your attention, it is also worth mentioning working in media is an NLP data scientist’s dream (to paraphrase Tess Jeffers, a data scientist in the WSJ newsroom). Media also provides any number of interesting challenges to solve: propensity, churn, revenue, topic modeling, audience clustering, and more.” -Alyssa Zeisler, Research & Development Chief, Senior Product Manager, Editorial Tools, Wall Street Journal.
“Depending on the research domain, data has many modalities: speech, acoustics, images, signals, point clouds, graphs, words, and more. Although there are specific visualization techniques for each domain, I especially enjoy the data rooting from visual content, using geometric priors, and its underlying high-dimensional nature. For anyone interested in vision and graphics applications of machine learning, media/entertainment industry is strongly suggested.” -Ilke Demir, Senior Research Scientist, Intel
“There are many unspoken and novel applications of data science in the entertainment industry today, but the plethora of opportunities yet to be discovered are what’s really exciting. It’s an industry that’s over 100 years old, and the chance to modernize and scale it lies in the contributions of data scientists. The next innovation in entertaining and bringing joy beyond the way we currently consume and produce movies, TV, and music is just around the corner, waiting for data scientists to unleash.” -Kim Martin, Data Science Manager at Netflix
“Growth. The Marketing Analytics Market is expected to reach USD 4.68 billion by 2025, at a CAGR of 14% over the forecast period 2020–2025. This is further boosted by the adoption of cloud technology and Big Data which will further increase the growth of the marketing analytics market.” -Denver Serrao, Sr. Software Development Engineer at WPEngine
“I think having a passion for the industry is key. Unlike industries such as biotech or pharmaceuticals, media and entertainment are inherently relatable to the vast majority of us, simply due to their prevalence in our daily lives. I myself began my data science career at Paramount Pictures (Viacom) due to my love for movies. I believe this level of familiarity and fondness for the subject matter is hard to cultivate otherwise, and it translates to better motivation at work.” -Daryl Kang, Lead Data Scientist at Forbes
“There is a strong component of human psychology and behavior that is part of most decisions in media/advertising/entertainment. While data and algorithms can be automated and learn a lot, there is a strong human element that requires diverse voices and thinking in order to truly connect users to content well.” -Amit Bahattacharayya, Head of Data Science at VOX Media
What advice do you have for new entrants to the field? (aka, what do you wish someone had told you?)
“First and foremost: the importance of clearly communicating is often underappreciated, but can mean the difference in an analysis or body of work being used or not. New entrants should work on how to articulate ideas and communicate them in ways that a stakeholder is likely to understand, whether that individual relies more on numbers or anecdotes. Learning what is a valuable problem to solve, how to ask good questions with data and solve problems creatively are similar and adjacent skills.
It’s also worth noting that a variety of backgrounds are relevant, so don’t think you’re missing a specific skill that will keep you from progressing. Our chief of data science is from astrophysics, a lead data scientist on the team comes from biology, and I’ve spent my entire career in newsrooms (and not in data roles). Having an understanding, appreciation and hunger can be just as, if not more important to your ongoing success.” -Alyssa Zeisler, Research & Development Chief, Senior Product Manager, Editorial Tools, Wall Street Journal.
“For anyone looking to get into data science in the media space there are a few pieces of advice I would give:
- Get to know your business in-depth. Technical skills are only half the battle. Data science only provides value when it is applied in a way that solves specific business problems.
- Build momentum. Find small ways that data science can provide business impact to build confidence and garner business investment in larger initiatives.
- Develop skills beyond machine learning. No matter how sophisticated your model is, if you put garbage in, you will get garbage out. Become an expert at exploratory data analysis and ask a lot of questions to know what the data you are working with truly represents. Knowledge of statistical analyses and optimization methods can also yield great benefits.”
-Bob Bress, Head of Data Science for Freewheel, a Comcast Company.
“For me, data science has been largely about learning from other’s expertise. Initially I thought I would study in depth a few different machine learning algorithms and statistical techniques. But, applying these data science tools to problems at the NYTimes has involved learning about a much broader set of topics from a wide range of people. How are we currently addressing the problem, why does that fall short, and what data can we use to improve our approach? How can we work with the rest of the company to improve the data quality so that machine learning can be more effective? How can we present the results of our modeling in a useful way, or integrate our production model into existing company infrastructure?
Coming from academic research, I was used to a paradigm where the analysis and scientific results were paramount and the communication of results was a necessary but secondary task. In my experience with data science, I’ve found that good communication, in both directions, and smooth integration are often just as important to the success of a project as the modeling and analysis. Even straightforward projects can end up quite wide-ranging!” -Anne Bauer, Director of Data Science, NY Times
“Don’t be intimidated by the work at first; a lot of learning happens on the job. This is especially true for those coming from a non-technical background — as a matter of fact, they often bring a diverse set of opinions to the discussion.
Unlike in academia, having a firm grasp of basic programming goes a lot farther than mere theoretical data science skills.
Finally, don’t be disheartened if the work doesn’t seem much like data science at first. Real-world data is messy, and it might take a while to reveal its value.” -Daryl Kang, Lead Data Scientist at Forbes
“There is no right path. Figure out what you are good at and find a way to join that with your job and function.” -Amit Bahattacharayya, Head of Data Science at VOX Media
“Data science can be learned by anyone who has a computer and access to the internet, so there will always be a large pool of data scientists with a set of homogeneous technical skills. The single most important trait that will lift your data science career to a higher level and set you apart from the crowd are your communication skills. Developing a data science solution involves complex techniques starting from acquiring data to training a ML model. The ability to translate analysis outputs into actionable business insights, and communicate them to business stakeholders is the most significant trait of a great data scientist. Logically, the communication of analysis outputs determines the impact of a data science solution as the ability to engage stakeholders. Emotionally, this helps us speak the same language as stakeholders and carve more meaningful alignment. Leveraging the business language to effectively communicate technical results is imperative, as it encourages the stakeholders to participate effectively in the ideation and validation of results. The best data scientists are empathetic in communicating results by crafting a compelling story with clear insights to present facts and figures to facilitate understanding for everyone.” -Upasna Gautam, Manager, Product & Data at CNN/WarnerMedia
“I personally have been in a primarily engineering role in my career. From a data science perspective though, it’s important to have a quantitative bent of mind. Most professionals in this field have an education that combines statistics, maths, programming/computer science along with some domain knowledge in marketing. The ideal person has a strong quantitative orientation as well as a feel for consumer behavior and strategies that affect it.” -Denver Serrao, Sr. Software Development Engineer at WPEngine
“Not all opportunities are created equal. Although you can gain skill and exercise your talent as a data scientist working in a variety of domains, the moment you find that domain you connect with, you’ll see your impact multiply. In that moment, you’ll move beyond just doing what you know, into that space of purpose and drive greater innovation. You’ll wonder what you ever did before”. -Kim Martin, Data Science Manager at Netflix
“SQL and Python are essential — but so is creativity.” -Wes Shockley, Senior Manager — Audience & Analytics, Meredith Corporation
“Always be curious about the facts and the reasoning, and always vocalize your curiosity. When your perspective is built on curiosity, data, and learning; you cannot escape from scientific breakthroughs. This also enables building scientifically grounded products with proper evaluations and theoretical foundations, which are more likely to survive in the longer term.” -Ilke Demir, Senior Research Scientist, Intel
“Attend meetups, even if they’re virtual. Data science is full of jargon, and it gets even more specialized when you move into a subfield such as Media, Entertainment, or Advertising. Meetups provide free exposure to this jargon! Even if you have plenty of technical knowledge, this will allow you to soak up the lingua franca of the field so that you’re ready to talk the talk when you get to an interview.” -Dominick Rocco, Data Scientist at PhData
What does “data science” mean to you? Or, what do you see as the difference between data science, ML, and AI?
“‘Data Science’ describes the application of analytical methods to data to drive insights. Those analytical methods could include machine learning, statistical analyses, probabilistic modelling, data mining or other methods. ‘Machine learning’ refers to a class of algorithms which generally seek to make a prediction or classification on data while allowing for the algorithm to ‘learn’ and adapt based on training data without explicit code directing it to do so. Machine learning provides a dynamic way of adjusting forecasts or classification methods as underlying data changes. ‘Artificial Intelligence’ more generally describes the simulation of human intelligence by machines. That simulation in many cases uses machine learning algorithms but may also use rule-based expert systems or other probabilistic-based simulation methods. We often see AI and ML used interchangeably today because new applications of AI tend to leverage ML based algorithms” -Bob Bress, Head of Data Science for Freewheel, a Comcast Company.
“Data science is the occupation of extracting value from real-world data; ML and AI are technologies that fall into the data scientists toolkit, along with others such as statistics and data manipulation. AI technologies are those which use programs or machines to mimic cognitive behaviors, while ML is a subfield of AI focused on programs or machines that automatically learn their cognitive behavior from data. Generally, an AI or ML scientist will focus on developing those technologies, often using standard benchmark datasets that are cleaner than real-world data. A data scientist, on the other hand, will take the latest and greatest AI technologies and apply them to messy real-world data to create value for individuals and businesses.” -Dominick Rocco, Data Scientist at PhData
“Data science is preparing, analyzing and deriving meaningful observations from data. It may, or may not be towards AI, and it may or may not be using ML. On the other hand, AI is creating an illusion of human-like intelligence and autonomy in machines, which usually depends on carefully crafted systems and data. Machine learning is the foundation of enabling machines to learn and reason from data and/or observations. As we progress towards deep learning and complex AI applications, the dependency on high quality data becomes crucial, so data science becomes an essential part of AI/ML applications.” -Ilke Demir, Senior Research Scientist, Intel
“Data science is the study of extracting value from data, while AI is the ability of machines to perceive and to adapt to changes in their environment through actions that optimize their objectives. While emblematic of the great technological advances of the present day, neither field is a recent phenomenon. Going by its definition, data science existed for a long as recorded information was available, while the field of AI research began as early as the 1950s. Even the game-changing archetype of modern AI systems, neural networks, was already conceived by the 1980s. What changed was the exponential increase in computing power, coupled with a fall in costs, and the mass proliferation of data in recent years. This enabled data science to alter the paradigm of AI research, supplanting a field that was once logic-based with one that simulates learning through statistical models — we call this machine learning.” -Daryl Kang, Lead Data Scientist at Forbes
“Data science is the application of the scientific process to answering questions with data.” -Wes Shockley, Senior Manager — Audience & Analytics, Meredith Corporation
“I take the science part of data science very seriously. It is not that hard to learn to program or a new language or framework. On the other hand, I can’t teach you math and logical thinking. A true scientist is skeptical, asks the hardest questions of themselves, and has incredible attention to detail. These are the characteristics that we need to be more than analysts.
As for the difference, I don’t really see much difference except an evolving set of words that the world uses to describe ‘How do I use data to model some process and make the most useful predictions that I can?’” -Amit Bahattacharayya, Head of Data Science at VOX Media
What’s next for you, career-wise?
“I consider myself lucky to be working in the media and advertising space at a time when Data Science is playing an increasingly important role in driving value within the industry. I hope to play a leadership role in increasing the adoption of and the investment in data science technologies and personnel across my company and the industry.” -Bob Bress, Head of Data Science for Freewheel, a Comcast Company.
“Unlike in academia, the most effective data scientists in the industry are those that can best productize and sell their data products. In this regard, I believe the greatest opportunity for growth comes in the shift to cloud computing as it allows the data scientist to focus more on the logic and algorithm at hand and less on infrastructure and DevOps. Hence, I expect to see more data scientists take on the role of cloud architect in the future.” -Daryl Kang, Lead Data Scientist at Forbes
“I would like to continue teaching, innovating and mentoring and helping guide small to medium sized organizations be smart w/ their data.”
-Amit Bahattacharayya, Head of Data Science at VOX Media
“Leading my team down the road to high performing predictive insights, so that when an opportunity is missed, it was by choice.” -Wes Shockley, Senior Manager — Audience & Analytics, Meredith Corporation
“It is absolutely amazing to drive the research in the world’s largest volumetric capture stage! My curiosity points to a different research question at every corner of the studio, and we are building unique AI solutions everyday. Having unprecedented amount of visual data and working hand in hand with artists for award winning productions, we are revolutionizing the entertainment industry with AI and data. I feel honored and privileged to have this unique position where my research can actually impact the world through immersive 3D experiences.” -Ilke Demir, Senior Research Scientist, Intel
—
Click here to read Part 1.
Hear from these speakers and more at Data Science Salon: Applying AI and ML to Media, Advertising, and Entertainment, September 22–25, 2020.
