Insights · 27 Jan 2026 · 5 min read · Iordanis Passas

Your research expired the day you shipped

In three points

Evidence describes the product that existed when it was gathered — not the one you have now.
RUCF decays confidence by ×0.8 per release touching a measured surface, not by the calendar.
The decay closes the loop by itself: shipping a fix automatically queues the re-measure.

Ask a team how current their UX research is and they will answer with a date. “We tested this in Q2.” The date is the wrong unit. The question is not how many months have passed — it is how many releases have touched the thing you measured.

A usability test describes the product that existed on the day of the test. Ship a change to that flow and the description now refers to something that no longer exists. Ship three and you are navigating with a map of a different city that happens to share the street names.

Decay on deploy, not on the calendar

RUCF multiplies a surface's confidence by 0.8 for every release that touches it. Three releases and your research is at half strength no matter how recently you did it. A slower time half-life runs alongside — twelve months for reported evidence, eighteen for observed — because user expectations drift even when your product does not. But the dominant term is the deploy, because you control the deploy, and it is the thing actually invalidating the evidence.

This inverts a comfortable habit. Teams treat research as an asset that persists until someone decides to refresh it. Decay-on-deploy treats it as a perishable whose freshness is a function of your own release velocity. A team shipping weekly cannot also claim its Q2 test still holds. One of those two facts has to give.

The loop closes itself

The useful consequence sits one step further on. In RUCF, priority is recoverable cost × confidence ÷ effort, and re-measurement is prioritised by how low confidence has fallen on surfaces that matter. So shipping a fix lowers the fixed surface's confidence — which pushes that surface back up the measurement queue automatically.

You do not need discipline to re-measure. You do not need a research-ops calendar or a quarterly ritual. The arithmetic asks for it, every time, in proportion to how much you have been changing the thing. That is what a framework is for: replacing virtue with structure.

Signals this affects

All twelve — decay applies to any measured signal. It bites hardest on task completion and activation, the surfaces teams change most often.

Find your gap