The cost of getting back up to speed after a gap in use
Aliases: memorability · return cost · forgetting
What it is
Memorability measures how much it costs a user who once learned the interface to return after a gap—not learning from zero, but recovering from rustiness. It is quantified by how far a returning user's completion, time, and help-seeking sit above their practiced baseline.
Why it happens
Interface knowledge splits between procedural memory (action sequences) and declarative memory (where it is, what it's called, what the parameters mean). Procedural components survive gaps well (like riding a bicycle), while declarative components—locations, labels, parameter meanings—fade fast, so the typical returning user is "hands remember, head doesn't": actions flow but entries can't be found and labels won't surface. Relearning is cheaper than first learning because of retained procedural skill; the wandering and help-seeking come from declarative forgetting.
Studying it
Use a longitudinal return design: participants learn the task, come back days or weeks later, and perform again; compare return performance against both first-use and practiced baselines. The interval is the key variable—the forgetting curve is nonlinear in time, so one-interval findings don't extrapolate. A log-based substitute uses real return sessions bucketed by "days since last use," comparing completion and duration across buckets.
Where it stops holding
Return cost depends heavily on feature type: high-frequency features have short gaps and strong retention, so low-frequency features are memorability's real battleground. Contact with competing products during the gap accelerates forgetting and adds interference. Return testing measures "coming back to the same product"; it does not cover memory invalidation by redesigns during the gap—that is a transfer problem, not forgetting.
Applying it
- Run return-after-gap tests on low-frequency features (two weeks or more), scoring memorability as return time divided by first-use time.
- Keep entries and naming stable across versions; weigh "regulars can't find it again" as migration cost in redesign reviews.
- Support returning users with light recovery cues—recent items, last position, examples—to discount declarative forgetting.