Predictions are what we anticipate for the future. They shape our plans across work and life. We execute these plans according to what we predict would be optimal outcomes.

When making a prediction, there is a tension between accuracy and precision. A prediction must first be accurate to be of any value; a wrong guess is just that - a miss. However, the value of a prediction is dependent on both accuracy and precision. Saying "The market will eventually go up" might be accurate, but it's so vague that it's hardly useful. "Google's stock will rise in 2023" has a higher precision and lower accuracy. "Google will go up 3% by tomorrow" would be precise and useful, if accurate.

To predict accurately and precisely, an understanding of the cause and effect, variables, weights and mechanisms are required. Product decisions are guided by understanding market segments and needs, competitive forces, budget constraints, available technologies, and innumerable other variables and weights. Successful decisions create compelling product offerings that are indispensable and a step function from the status quo.

This process is effort intensive for complex systems. Reducing that effort generally leads to lower accuracy or precision. In a world of finite resources, effort is fungible across different initiatives. One may spend their time doing user research or data analysis, or building and shipping the product and measuring its performance.

It requires less effort to measure than to predict. There is also a certainty with measurement that guarantees accurate and precise conclusion (hindsight). This measurement becomes as an incremental step for the next cycle of predictions (t0 → t1). This is the essence of the build-measure-learn loop. Instead of investing heavily in the planning phase, you can build a minimal viable product (MVP), measure its performance, and iterate based on real-world data and feedback.

The obvious answer is it should be a balance. Only talking to users and never shipping is meaningless. Shipping a product without doing any preparation is also futile. Some industries may "move fast and break things" while others like healthcare or aerospace require more intense preparation due to the high stakes involved.

Don't make blind predictions and don't build in a vacuum. Listen to what users want, then listen if they want what you built.