The edge of the friction map

Analytics tools are very good at finding feature level friction. Click patterns, drop-off rates, funnels, all stack neatly into a backlog and give everyone something appropriately measurable to discuss.

Almost too neatly.

Follow the logic far enough and we mostly arrive at a higher resolution confirmation of what we already suspected. Yes, users are dropping here. Yes, this workflow takes too long. Yes, people would like fewer steps. The maths is sound; the novelty is less clear.

The more interesting signal is usually somewhere at the edge of the friction map. Like when someone exports query results into an LLM and asks, “What’s driving this spike?” That is not necessarily a user leaving the product. It is a user awkwardly assembling the next part of the it from what is already available.

Most valuable opportunities may begin as an unintended path: copied data, a workaround, a spreadsheet with too many tabs.

The task is to notice when that path is becoming a product in its own right, then build it on purpose.