The real cost of one cross-tool interruption

A cross-tool lookup rarely ends at the lookup. Someone asks where a ticket stands, which policy applies, or what a customer said last week. The search starts in one app, continues in another, and usually triggers a message to a person who already knew the answer.

That handoff is expensive because the work stopped is not only the search. Attention has to return to the original task.

What the research suggests

UC Irvine interruption research is often cited for the claim that it takes about 23 minutes to resume an interrupted task after attention moves elsewhere. Using that as a planning assumption—not as a universal law—one avoided detour per workday returns roughly:

• 1.9 hours of focus time per person each week across five workdays
• 38 hours across a 20-person team—nearly a full workweek of capacity

Those numbers are estimates. They are useful because they give a pilot something concrete to test against, not because every team will see the same result.

What to measure instead of vibes

If you run an operational AI pilot around answers and action, baseline before you expand:

1. Time-to-answer for a recurring question type
2. How often people still ping the human who “just knows”
3. Throughput for the workflow that was supposed to get faster
4. Whether answers stay linked to sources people trust

Why this belongs on a product site

Pipeer’s homepage uses this interruption framing because the product is built for the hunt across Jira, Slack, Gmail, Drive, Calendar, and related systems. The point is not that AI magically deletes interruptions. The point is that a focused pilot should prove whether fewer detours show up in the metrics your team already cares about.

If you want the research PDF behind the 23-minute figure, start with Gloria Mark’s UC Irvine work on interrupted task resumption. Then design your pilot so the claim is falsifiable for your stack.