Before we got back into the case, I gave my MSc. in Data Science class a hypothetical. A team walks into your boardroom and pitches ₱1.5 billion for an enterprise data platform, with a five-year TCO. Imagine you’re on the board. “What’s your biggest question? Then, give me two more to back it up.”
It’s a small exercise, but it does its job well. It forces the class to treat data science, AI, and tech investment as a business value question first, instead of just a tech one.
From there we went back to the main case, a Harvard Business Review case I’ve been using since 2019: a brick-and-mortar retailer that, because of a financial crisis, had to pivot into a company built around data and AI. That pivot isn’t an easy decision, especially when you’re navigating unfamiliar territory. So there has to be an actual business case behind it, and we spent real time building one out. Each student gave their own justification, and we put them in four buckets: context, strategic choices, opportunities for learning, operational benefits. Once that case felt solid, I threw in a curveball. “What are the risks a board would actually raise?” Back through resources, the approach to retailing and culture, organizational issues, technical issues. I thought it was brilliant when one of the students asked, “What if it doesn’t work?” I forced the class to further dissect the question; and it did surface different dimensions of risk.
I like “teaching” data and AI strategy this way instead of just handing over a framework. Students get to be the proponent building the case and the board member judging it, in the same session.
I always remind them that frameworks help to frame our thinking, but they need to be broken down, dissected, and contextualized before they can be truly useful.
A few things I wanted them to walk away with. The infrastructure question has to get tackled early. Measurement discipline is not optional; you need a clear success metric and a clear roadmap, so champions and stakeholders know what they’re benchmarking against, especially on a five-year TCO. The organizational problem is more difficult than the technical problem, every time. Generative AI investment is tough on ROI right now, and GenAI is not a shortcut to having an actual AI strategy. Take the long view, but find the quick wins, as they let you keep funding the long game.
None of it works without the organization (people and structure) actually being able to do it. The board can approve the ₱1.5 billion, the technology can be genuinely good, and the whole thing still stalls out if the organization doesn't have the capacity to make the change or the capability to run it once it's live. I've seen good strategies get stuck at the pilot stage for exactly that reason.
We did all of this on the blackboard! No slides for the whole session, and that raw discussion is refreshing in a way a deck with a clicker never is.
In the second half of the three-hour session, I walked the class through a framework for mapping data and tech strategy at the organizational level. You start with business value, even with the mission and vision of the organization, and work down step by step to the technology: data capability, infrastructure, the rest. It’s the most practical.
Buying the technology is the easy part. Whether the organization can actually use it is the more challenging question, and no five-year TCO answers that for us.