Q: How can we prioritize something if we don’t know the solution? How can we estimate it?

A: That is like asking whether we can prioritize going to the grocery store without a shopping list. Sometimes the list helps — sometimes recipes help, even — but I have confidence in your ability to figure it out. If I am super hungry (high cost of delay), I’ll be even less concerned with a list.

Having/not having a prescribed solution does interesting things to the “bet”. On one hand, having a spec improves confidence. If we receive cash on delivery when we deliver “Exactly This”, then great!

But software product development (often) doesn’t benefit from this approach. I may be MORE confident to ask a team to run lots of experiments to move an important metric, than I am that one pre-determined solution will eventually deliver an outcome. I also know the traps inherent in only trying one thing. It is a balancing act.

Now, about estimation. Say I have…

1. A confident estimate for a given solution (2–4 months), and
2. A loose estimate on moving a metric (2–8 months), with an escape clause after 2 months if we find moving the metric to be too hard, and various pivot/proceed points throughout. Tendency is to gravitate to the confident estimate, even when the outcome driven approach is probably (but not necessarily) the better economic / game-play decision.