Sell Me This .. Model

As data scientists, we often spend extensive time developing machine learning models, aiming to enhance business processes with by offering accuracy, efficiency, and capabilities beyond human reach (yes, because we teach machines)!

🤔 However, our deep immersion in the technical aspects can lead us to misjudge our models’ ‘immediate’ appeal to others, which often does not occur (and I realize that even saying ‘often’ is optimistic for many, as the adoption of these models might be rare!).

💡 One key issue is the difference in perspective between us, the developers, and the business experts who use our models.
We are often traped into being biased regarding the inherent value of our models! We expect quick acceptance and success, but must acknowledge that these experts may not see the immediate benefits.
Their skepticism, though initially surprising, is a normal response to unfamiliar and complex technological tools.

😎 Let me for a moment borrow something another domain, from my experience with petrochemicals marketing. Convincing customers to adopt a new engine oil, in the face of their loyalty to existing brands, was a significant task.
Even when highlighting our product’s advantages, there was noticeable hesitation.
This experience is quite similar to data science. We can become so focused on our models’ capabilities that we overlook the essential task of building trust and understanding among users.

💡 An effective strategy for overcoming this challenge is to involve potential users in the development process from the outset.
Reflecting on my time in the petrochemicals industry, involving clients in decisions like naming and specifications was not just helpful — it transformed them into product champions.

💡 In data science, treating models as products and viewing business experts as customers is critical.
Involving them in shaping the model’s scope, choosing features, and more importantly, through ongoing interactions where we address their concerns and integrate their feedback, is extremely valuable.
This approach ensures the model not only meets technical standards but also resonates with the practical needs and queries of the users.

Ovearall, it’s crucial for us as data scientists to extend our view beyond the technical development of models.
Embracing a sales mindset and actively engaging business users in the creation process can effectively narrow the gap between development and adoption.