Art, Data Science and Product Strategy

Photo by freestocks on Unsplash

“ The ancient teachers of science”, said he “promised impossibilities and performed nothing. The modern masters promise very little; …..but they ascend into the heavens; they have discovered how the blood circulates, and the nature of the air we breathe. They have acquired new and almost unlimited powers; they can command the thunders of heaven, mimic the earthquake, and even mock the invisible world with its own shadows.

-Excerpt From Frankenstein , Mary Shelley

The above lines are spoken by Professor Waldman to Victor Frankenstein touting the achievements of modern science and the futility of promises made by ancient alchemists. I reread Frankenstein over winter break and it resonated with me beyond its themes of human ambition and fallibility. As a data science product manager, to me it was a tale of falling in love with building and unleashing a product of your design without understanding the market. I doubt it would have been so tragic if Victor Frankenstein had armed himself with product management techniques, as he combined natural philosophy with modern science to create a life form. It was also analogous to my own efforts in bridging data science with art; redesigning old ways of working to leverage statistical models.

One of the best jobs I had was when I was leading data science product management for Walmart US apparel; some of the most fascinating and challenging requirements revolved around implementing data driven recommendations in processes primarily driven by art. Challenging, not because there are no machine learning models that can prescribe close to accurate trends but gaining trust from your clients, especially in a world where planning is closely tied to artistic sensibilities and type 1 thinking can be tricky. Everyone is aware of all the borders data science has conquered. The all chimerical world of love and its Petrarchan sonnets have been quantified as a compatibility score and the LSTM neural network models are making the magic of music composition more approachable. But how we action, monetize and most importantly partner with our clients to execute automation of decisions through machine learning models is what differentiates a viable business model from Sisyphean efforts around AI investment that don't go beyond a proof of concept.

Embrace the art : Typically in data science projects, once an opportunity has been identified for enabling data science capabilities, product managers work concurrently on customer and product development. But in an art-driven mileu, extra effort must be put into unravelling years of esoteric knowledge and reifying definitions of abstract ideas, like trendy, presentable, aesthetic… It always irks me when I hear product managers mention they don't understand art and prefer to focus on technical minutia. Art is not just affiliated with museums and aesthetes, its a fundamental human activity. Working at the intersection of art and science helps you understand the association between the two. Leaving behind the promptings of technical wisdom, data science product managers should instead open their minds to getting an in depth understanding of the art of the problem and process.

“I will pioneer a new way, explore unknown powers, and unfold to the world the deepest mysteries of creation.”

-Excerpt From Frankenstein, Mary Shelley.

Understand the intersections of science and art : Professor Waldman in Frankenstein believes it a waste to spend time on natural philosophy. But as evident in the book, Frankenstein combined his early knowledge of old world natural philosophy and modern science to accomplish the impossible. With fashion, time spent auditing historical fashion forecasting processes helped identify the right threads to focus on and enhance with data science models. Its true, some fashion decisions were made on intuition. But as I focused on listening and building trust with our business customer, I realized the outcome was not decision automation but structured empowerment to determine future trends, colors, styles and outfit pairings. I built detailed requirements for the data science model to predict trends based on history, global design trends from trend forecasting companies, competitive analysis conflated with an aspect of heuristics from a human perspective. The end result is a perfect balance of AI models that learns from complex patterns generated from historical transactions, attributes extracted from trending images/texts and business users who ingest the AI signals as guidance to plan the assortment. Adoption of this methodology ultimately depended on the alchemy of the trust we build with our business users, I covered it my previous blog.

Never underestimate user inertia : One of the saddest moments in Frankenstein is when Frankenstein’s extremely intelligent and strong creature secretly helps the De Lacey family in hopes of being accepted. But the family attacks him because they find the creature hideous; they are scared of something that does not fit into what is considered “normal” in a social context. Sometimes business teams are so used to performing the quotidian tasks like second nature however cumbersome, that they don’t look to enhance the routine. Its easy to trust an old and tried methodology they control versus an AI algorithm that promises to be effective but is a black box. In an effort to build next generation functionalities , never lose focus on changing an existing customer perspective through frequent engagement. Underestimating customer inertia could have your product end up an expensive experiment that never reached production.

Build confidence with pilot : When you are transforming art driven domains, remember your business customers are highly susceptible to atavism. During discovery phase, when we interview business users to understand unmet needs and challenges, all the problems we hear are about tough unmanageable processes or tools that take up time; no one complains about the art driven process. Truth is, no business user is ever looking to upgrade existing art driven processes. In apparel, most of the key requests we got were on inventory optimization, space planning or pack configuration. Fashion forecasting while foundational to apparel planning is something that the apparel business team perceives to have a good handle on and is not prioritizing to enhance. So its important to get it right the first time or the users will quickly revert back to the old ways. Building confidence with pilot business champions will be crucial to a wider buy-in during product roll out .

Don’t complicate : Finally focus on simplifying your products and creating transparency on the algorithmic recommendations. This could mean using the business acumen of your users to determine features, rolling launches, frequent outreach activities leading up to implementation and surfacing insights in the form of storytelling to highlight correlations, outliers and trends. Finding a way to infuse art with data analysis in your presentations can reveal a picture that strikes plangent chords with your clientele, enabling trust.

“Sir Isaac Newton is said to have avowed that he felt like a child picking up shells beside the great and unexplored oceans of truth”

-Excerpt From Frankenstein , Mary Shelley.

As with any product strategy, its always about keeping an open mind, starting from tabula rasa and reason. The business client possesses the wisdom of the past and present working of the process you are trying to enhance, so continue to stay present and curious. You will be rewarded with a good business partnership and more importantly a seamless integration of AI recommendations and human information processing. I thoroughly enjoyed reading Frankenstein and pondering upon the lessons that still apply today. But if only Victor Frankenstein had worked with a product manager, a powerful literature of ego, fear of isolation and the tragedy that ensued could have ended up as a sci-fi about a boldly creative polymath who amalgamated natural philosophy and modern science to create a super intelligent, strong and generally approved ancien régime AI, albeit one made with human parts.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store