Advance warning of a switch only partly reduces the cost, never eliminating it

What it is

Telling a user in advance "you're about to switch to another task" and giving them time to prepare does lower switch cost — but not to zero. Even with a very long preparation window, some cost always remains that no amount of warning or preparation removes; this is the residual switch cost. This corrects a common overly optimistic expectation: adding an "about to switch" cue to a design does not solve the switching-cost problem, it only solves the portion that preparation can compress.

Why it happens

Advance warning and preparation time work by letting the active goal-reset process begin before the new rules are actually needed, ahead of the moment the switch happens. That work would otherwise have to squeeze into the instant of the switch itself; doing it early leaves less to do at switch time, so the cost drops accordingly. The residual portion is different — it comes from the continuing activation of the old task's rules, and that activation's decay depends mainly on elapsed time and whatever else happens in the meantime, not on whether the user has mentally decided to switch. Even when a user is fully prepared psychologically, the lingering activation of the old rules decays at its own pace; if the new task starts before that decay finishes, the interference still happens. This is why the cost curve, no matter how long the preparation window, drops with time and then flattens into a plateau rather than reaching zero.

Studying it

The same manipulation used to decompose switch cost applies here: systematically lengthen the preparation interval and watch where the declining switch-cost curve plateaus. Whatever remains at the plateau is the residual quantity this entry is about.

Common independent variables: length of the preparation interval, from nearly none to generous. Common dependent variables: the magnitude of the drop in switch cost as a function of interval, and the plateau value.

This result is often used to gauge the ceiling on "advance notification / preloading" design interventions — telling a product team there is an upper bound on how much benefit this class of intervention can deliver, and that investing further beyond a certain preparation length yields no additional return.

Methodological note: the absolute size of the residual cost varies widely across different task pairs, so a specific number measured in one study should not be treated as a universal constant transferable to a different setting. Only the qualitative conclusion — a floor exists and it never reaches zero — is stable.

Where it stops holding

• If two tasks share almost no rules (little task-set-inertia competition to begin with), the residual cost is already small, and advance warning's marginal benefit will look close to "solving the entire cost" — but that's because the residual was low to begin with, not because warning became more effective.
• This applies to switching performed continuously by the same individual; it says nothing about organizational solutions such as dividing work across multiple people to avoid switching altogether — that is a different level of solution.
• The conclusion rests on short-interval lab switching paradigms. If a user has a longer period of free activity between two switches (tens of seconds or more), the lingering activation may have already decayed naturally, narrowing the difference between warning and no warning.

Applying it

• Don't treat "give advance warning of an upcoming switch" as a complete solution to switch cost when signing off on a design; expect it to deliver only partial improvement, and plan for the rest with other measures — reducing rule overlap, widening tolerance for errors right after a switch.
• When deciding how much lead time an "upcoming switch" warning should give, determine empirically where the benefit curve plateaus; increasing the warning lead time beyond that point yields little further improvement, so there's no need to stretch the wait indefinitely.
• Verification: test at least three different warning lead times on the same switching scenario and plot switch cost against lead time. If the curve flattens past some point, the residual-cost floor has been reached, and further optimization should shift toward reducing task overlap or adding tolerance for errors, rather than lengthening the warning further.

Related

• Same group: A9.03.1 Switching tasks requires reloading rules and goals · A9.03.2 Switch cost is higher when tasks are similar · A9.03.3 Apparent parallel processing is mostly rapid switching · A9.03.4 Switch cost has an active goal-reset component and a passive carryover-interference component · A9.03.6 Switching from a simple to a complex task and the reverse carry asymmetric costs · A9.03.7 Frequent small switches can accumulate more total cost than a few large ones
• Nearby: A5.08 Interruption cost and task resumption
• Search terms: residual switch cost · preparation interval · task-set reconfiguration