The Right Answer

In the tech corridors of Cambridge, Massachusetts, Sundar, a brilliant data scientist at a rising AI startup, sat frustrated in a conference room surrounded by dazzling dashboards and intricate machine learning models.

His team had just finished building a high-precision model that could predict the likelihood of online users clicking ads—down to decimals. But the company’s growth was stagnant. Something didn’t add up.

Later that evening, he met Dr. Nancy Carter, his old MIT professor, over coffee. After listening to Sundar rant about precision and models, she gently asked, “What was the original problem your team set out to solve?”

Sundar paused. “To help our clients grow their businesses.”

Nancy smiled. “So why are you solving for clicks?”

The question hit Sundar like a bolt. His team had been optimizing for the wrong metric all along—chasing accuracy in a proxy problem instead of understanding what truly mattered to the users.

The next day, Sundar shifted focus. Instead of predicting clicks, the team explored customer lifetime value. It was harder, messier, and required assumptions—but even an approximate answer to that real question began yielding results.

Months later, their strategy transformed not only their product but also client success. The quote from Dr. Carter echoed in his mind:

"An approximate answer to the right problem is worth a good deal more than an exact answer to an approximate problem."