You glance at a Slack notification. It takes three seconds to read. You go back to what you were doing. Simple, right?

Not quite. According to a widely cited figure in attention research, commonly attributed to Gloria Mark at UC Irvine and reported by sources including TCTEC Innovation, the average time it takes to fully regain deep focus after an interruption is 23 minutes and 15 seconds. A three-second glance can cost nearly half a workday's worth of recovery time if it happens often enough.

That's the core of the 23-minute rule.

What the 23-minute rule actually says

The 23-minute rule is not a claim about how long your attention span lasts. It's a statement about recovery cost: how long it takes, on average, to return to the same depth of focus you had before an interruption pulled you away.

The distinction matters. Your attention span isn't limited to 23 minutes. What's limited is your ability to snap back instantly after a context switch. The rule describes the drag, not the duration.

Where the number comes from

The 23 minutes and 15 seconds figure is commonly attributed to Gloria Mark's research into workplace interruptions at the University of California, Irvine. Secondary sources, including TCTEC Innovation (September 2025), consistently reproduce this number in the context of knowledge worker productivity and deep work interruptions.

A few caveats worth noting:

The figure represents an average, not a universal constant. Recovery time varies by person, task complexity, and the nature of the interruption.

The same stat gets reprinted across hundreds of productivity articles, sometimes without full methodological detail. Treat it as a well-supported directional estimate, not a precise timer for every worker.

What the research consistently shows, regardless of the exact number, is that interruption recovery takes significantly longer than most people expect.

Why interruptions are expensive: attention residue

The mechanism behind the 23-minute rule is what researchers call attention residue. When you leave task A to respond to an interruption (a Slack message, an email pop-up, a "quick question" dropped in a doc comment), part of your cognitive load stays attached to task A even after you nominally return to it. You're physically back at your desk, cursor on the right file, but your mental state is still partially elsewhere.

Reloading full context takes time. The deeper the original task (writing, coding, analysis, research), the more context there is to reload, and the longer recovery takes. This is why context switching cost hits knowledge workers harder than it hits roles with naturally fragmented, transactional work.

The hidden time budget leak

Consider a simple estimate. Suppose you're interrupted twice per hour during a four-hour focused work block. That's eight interruptions. At 23 minutes of recovery cost each, the theoretical total recovery time exceeds the entire block.

Reality is more forgiving than that calculation (recovery costs overlap and compound rather than stack cleanly), but the directional point holds: repeated interruptions don't add linear costs, they multiply them. Each new interruption resets a recovery clock that may not have finished running from the previous one.

This is why many people feel like they worked a full day but produced very little. They were busy. They just weren't in deep focus for long enough to do the work that requires it.

How to reduce the damage

The research on interruption cost points toward a few practical responses:

Batch your communication.Rather than monitoring Slack, email, or messages continuously, schedule two or three windows per day to process them. Notification fatigue is a real productivity drain; reducing interrupt frequency is more effective than trying to recover faster.

Protect focus blocks.Designate stretches of time (90 minutes is a commonly used interval, though the right length varies) where communication tools are closed and interruptions are structurally prevented rather than willpower-dependent.

Use timeboxing frameworks as scaffolding.Pomodoro-style sprints are a structured work/break cycle. That's different from the 23-minute rule, which describes an unplanned interruption cost. But timed sprints do help you establish rhythm and make interruption windows predictable rather than random.

Apply environment controls.Do Not Disturb modes, device focus settings, and system-level website blockers all reduce the surface area for interruptions. The stronger the enforcement, the less willpower the system requires.

How LockIn MCP operationalizes this

Most distraction blockers address one layer: they block a list of sites in a browser. The problem is that a motivated, distracted brain will open a new tab, switch browsers, or disable the extension. The interruption still happens.

LockIn MCP takes a different approach by operating at three levels simultaneously.

First, system-level blocking via hosts file edits. Sites like YouTube, Reddit, and X are redirected at the OS layer to 127.0.0.1, which means they're blocked across every browser and every new tab. There's no browser toggle to flip. This is the enforcement layer that makes focus blocks structurally reliable rather than volitional.

Second, a Chrome extension with task-aware filtering. Rather than applying a static blocklist, the extension evaluates whether a page relates to the active focus task. Pages that don't connect to the task are blocked. Distraction attempts show up on a dashboard so you have visibility into patterns, without exposing full browsing history.

Third, MCP connectivity for AI-initiated sessions. This is where LockIn fits into workflows built around AI assistants. You can connect ChatGPT, Claude, Perplexity, Gemini, or Poke to LockIn's MCP server. A practical workflow looks like this: you tell Claude "I'm working on the Q3 report for the next 90 minutes" and LockIn activates focus blocking automatically. If you genuinely need to visit an unrelated site mid-session, you can request a temporary unblock with accountability through the AI rather than breaking the session entirely.

The accountability mechanism matters here. Impulsive tab-switching is often automatic. Having to explicitly ask an AI assistant for a temporary unblock introduces enough friction to interrupt the habit loop. You can read more about connecting your AI assistant to LockIn's controls.

Installation runs from a single command: npx -y lockin-mcp install. It works on macOS, Linux, and Windows. LockIn fits into your workflow, not the other way around.

Frequently asked questions

Is it always exactly 23 minutes?No. The 23 minutes and 15 seconds figure is an average drawn from workplace research. Recovery time depends on how deep the original task was, what type of interruption occurred, and individual cognitive differences. The useful takeaway is that recovery takes substantially longer than most people assume.

Does this only apply to phone notifications?Any context switch counts: email, a colleague dropping a question in a document comment, switching browser tabs to check something "quickly," or opening a new chat thread. The interruption doesn't need to come from a device notification.

What if my job requires constant context switching?Some roles genuinely require it. The practical response is to define "detour windows" (scheduled periods for reactive work) and narrower focus targets for protected blocks, rather than eliminating switching entirely. Batching reactive work into defined periods reduces the interrupt frequency during focused periods.

How is the 23-minute rule different from Pomodoro?Pomodoro is a planned work/break structure (typically 25 minutes on, 5 minutes off). The 23-minute rule describes an unplanned cost: the time it takes to recover from an interruption that wasn't scheduled. Using Pomodoro can reduce the opportunity for unplanned interruptions, but the two concepts address different things.

A practical starting point

The core implication of the 23-minute rule is simple: interruptions cost more than they appear to, and reducing their frequency matters more than trying to recover faster.

Two concrete steps that follow from this: set two or three fixed communication windows per day and treat the time between them as structurally protected. Then consider whether your current setup makes it easy enough to actually enforce that protection. If willpower is doing all the work, explore how AI-connected blocking can take some of that load off you. LockIn MCP's pricing starts at $2.99/month after a 3-day free trial, with a $59.99 lifetime option if you prefer a one-time purchase.