The Learning Curve Design Problem

Getting people to use your product requires more than onboarding, it demands a transition plan.

A New Edge

Tiger Woods is one of the most successful golfers of all time. In his prime, he was able to combine power with accuracy and solid putting to win more money on the PGA tour than any other golfer. He has the 2nd most career wins in the history of the PGA and will likely have the most by the time he retires. He has also won the 2nd most major tournaments (these bring out the best golfers) and still has a chance to claim the most here, too. His list of accomplishments is extensive.

Despite his success, he was never satisfied. He always looked for a new edge. Even when he was at the top of his game, he saw that he had room for improvement. He frequently tinkered with his swing, sometimes overhauling it. Every time he did, people would question why he would mess with successful results. He was already the best golfer after all.

His first few tournaments back after refining his swing, he would struggle a bit; he did not look like his usual self. His shots would be a bit more erratic. He might lose a little bit of distance. His performance would suffer and he would not be playing in tournaments like we had come to expect. Commentators would begin to wonder if he needed to revert back to his old swing. This would go on for several weeks.

Then, suddenly, the training finally clicked. He would get comfortable with his new swing. He would get the timing down and his body adjusted. He would find a bit more power or a bit more accuracy than he had before. Already the best player, he became even better, increasing his advantage over everyone else.

Local Minima and Maxima

What Tiger’s swing change shows us is that learning is not always linear nor is it a step function (staying level, and then jumping up when the skill is achieved). As we learn new skills, or in Tiger’s case enhance current ones, our results do not improve incrementally along with the amount of effort put in.

We learn new skills to break out of a local maximum. With our current skills, we plateau and we need to do something to advance to the next local maximum. However, doing this often dips us into a local minimum from some time before we reach that new local maximum.

As we learn we must unlearn some old habits and break others, causing our performance to temporarily decrease as we try to build back up with the new pieces that we are adding. If the task is difficult, we may have more to unlearn, which will require even more effort to get back to and above where we once were. This is important to consider when we add new tools to help our users.