Protecting focus while on call: coverage trades, triage rules and recovery time

Deep work · 9 min read

Being on call and doing deep work are usually described as opposites, and for the person holding the pager they often are — but the research on interruptions explains the damage precisely enough to design around it. The widely cited figure traces to UC Irvine researcher Gloria Mark: refocusing after an interruption is slow, with people typically handling intervening tasks before returning to the original one. A careful recent treatment of the deep-work literature is honest about that number's provenance — the famous “23 minutes and 15 seconds” comes from a 2006 Gallup interview rather than a peer-reviewed paper, and is best treated as a vivid illustration, not a physical constant — while Mark's published field studies do establish the direction firmly: the working day is badly fragmented, with average time on a single screen collapsing from about 2.5 minutes in 2004 to roughly 47 seconds in later tracking. For an on-call worker, the conclusion is not “never do deep work.” It is that focus on call has to be structured differently: protected by coverage trades, bounded by explicit triage rules, and honest about recovery. This guide gives that structure.

Accept the shape of the problem

On-call interruptions differ from ordinary interruptions in two ways that matter for planning. First, they are non-negotiable in the moment — an incident does not wait for a focus block to end, and a system that pretends otherwise produces either missed pages or anxious, shallow “focus” spent listening for the phone. Second, the anticipation itself costs focus before any page arrives: part of the mind stays allocated to monitoring, which is why on-call deep work feels thin even on quiet days. Microsoft's 2025 workday research, built on Microsoft 365 telemetry, underlines the ambient scale of fragmentation for everyone — an average of 117 emails and 153 Teams messages a day, half of meetings landing in the 9–11 a.m. and 1–3 p.m. peak-focus windows, and 48 percent of employees describing their work as chaotic and fragmented. The on-call worker carries that baseline plus the pager. Planning that ignores either layer fails.

The coverage trade: focus time is swapped, not found

The deep-work guidance for reactive roles is consistent: negotiate protected blocks rather than discovering them. Concretely, for on-call work that means a reciprocal arrangement — a colleague covers the pager for a defined block (ninety minutes is a workable unit, and deep work guidance puts the meaningful threshold at roughly two hours of uninterrupted time, ideally more), and the cover is returned in kind later in the rotation. Three rules make trades work instead of breeding resentment. The block is scheduled and visible, not grabbed opportunistically, so the covering colleague can plan around it. The definition of “page the focus person anyway” is written down in advance — severity levels that break the block, and everything below them waits or is handled by the cover. And the trade is genuinely reciprocal over the rotation, tracked loosely enough to stay collegial and firmly enough that nobody's focus time is subsidised by the same colleague every week.

Triage rules for the hours outside the block

1. Publish the response-time truth. Most “urgent” requests are not: communication research summarised in deep-work guidance finds only about eleven percent of business contacts expect a response within fifteen minutes, while most expect roughly four hours. Agree explicit response targets per channel (page, chat, email) so neither the on-call worker nor requesters have to guess.
2. Batch the non-page channels. Checking messages at set times — deep-work practice commonly uses two or three batches a day — cuts interruption volume sharply (one summarised figure: up to 68 percent fewer interruptions versus continuous checking) while still beating the response times most people actually expect.
3. Default to asynchronous, escalate by exception. Most unscheduled “quick calls” are convenience for the caller; async handles the large majority of what people think needs real time. Reserve synchronous interruption for the written severity list.
4. Keep a capture pad beside the focus work. Stray obligations that surface mid-block get written down, not acted on — the pad holds them safely until the block ends.
5. Close the block deliberately. End by noting where the work stopped and the first step for resuming. Starting is the hardest part of the next session; a written first step removes the re-entry cost that interruptions otherwise impose twice.

What success looks like: during the traded block, the pager is physically with the covering colleague, the focus worker's status shows the arrangement and its end time, and the severity list — not anxiety — decides what breaks through. Outside the block, responses follow the published targets rather than a permanent state of alert.

Recovery is part of the rota, not a perk

A night of pages is an interruption event measured in hours, and the refocus research scales accordingly: the morning after a broken night is not a normal morning with less sleep, it is a fragmented day that starts fragmented. Teams that protect focus sustainably treat post-incident recovery as scheduling, not indulgence — a later start or a shallow-work morning after a heavy night, agreed in the rotation rules before anyone has to negotiate while exhausted. The same logic applies within the day: after an incident closes, a deliberate reset (notes written while memory is fresh, then a break, then resumption via the written next step) beats drifting back to the desk and discovering the context is gone. Deep work is, in the end, the deliberate refusal to move attention constantly — on call, that refusal has to be organised by the team, because the individual holding the pager cannot organise it alone.

Negotiating the arrangement without apology

Coverage trades fail socially before they fail logistically: someone feels they are asking for a favour, someone else feels imposed upon, and the arrangement quietly dies. Framing prevents that. The trade is not a request for indulgence; it is a scheduling mechanism that raises the team's total deep-work output, because every member rotates through both roles and the focus hours are returned with interest when the covering colleague takes their own block. Put the proposal in those terms, attach the written severity list (which protects the covering colleague from judgement calls), start with one traded block per person per week rather than an ambitious daily scheme, and review after a month with honest numbers: blocks held, blocks broken, and by what. Arrangements that survive are the ones whose terms were boring, explicit and reciprocal from the first week — which is, not coincidentally, also a fair description of every other on-call practice that works.

Frequently asked questions

How long should a protected block be? Deep-work research puts the meaningful threshold at about two hours uninterrupted, ideally up to four; on call, traded blocks of ninety minutes to two hours are the practical unit. Is the 23-minute refocus figure reliable? Treat it as an illustration: it traces to a 2006 interview with Gloria Mark, while her published studies firmly establish slow refocusing and severe fragmentation (average screen time down to roughly 47 seconds). Won't batching messages make me look unresponsive? The evidence points the other way: most contacts expect responses in hours, not minutes, and two to three batches a day typically exceed those expectations. What breaks a focus block? Only what the pre-agreed severity list says — decided in advance, in writing, not negotiated mid-incident.

Sources, accessed October 21, 2026: Carly, “Deep Work: The Hidden Cost of a Fragmented Day — and Where the Advice Breaks” (Gloria Mark provenance, Microsoft 2025 telemetry); Network Perspective, “Context Switching in Developer Experience — Protecting Focus Time”; Dev.to, “How to Schedule Deep Work When Your Calendar Is Already Full” (reactive-role adaptations, coverage trades, batching). Figures are those sources' summaries of the underlying research.