How often you open the inbox is the largest single decision anybody makes about email. It’s also the one that usually gets made for you. It gets made by a notification, or by a habit that formed in 2014, and then it just keeps running for years.
This piece is about making it on purpose. What the schedule should be, what has to be true for it to survive a bad week, and the short list of things that genuinely justify breaking it.
- Fewer trips is the quantity that matters. Tidy blocks are one way of getting there. That’s been measured, and the measuring is next door.
- The expensive check is the one somebody else started. After an email alert, workers took nearly ten minutes on the detour and another ten to fifteen before they were back on the disrupted task. At four alerts an hour, they were rarely outside one of those windows.
- Three working passes and a short scan is a schedule a small firm can hold. Each pass needs a written job, and one of them needs to be late enough in the afternoon to still reach a client today.
- Write down what actually cannot wait. In most small firms it’s three to six things. If you file returns, part of that list is published by the IRS with the days counted.
- A checking rule for other people is a management decision, not a productivity tip. If any of your staff are non-exempt, telling them to look at 9:00 at night creates hours you owe them.
Two neighbors carry the parts this page hands off. The pull toward the inbox when it’s already empty, and the evidence behind checking less, is why you check your email constantly. Evenings and weekends, and your rights as somebody else’s employee, is protecting your evenings from email. Within this set, what the hours add up to across a firm is how much time firms lose to email. What to do inside a pass once you’re in one is getting time back from your inbox.
What the evidence settles before you start
Two findings are worth having in hand. They decide what kind of schedule is worth building. Both are established properly next door, and the useful residue is short.
Checking less works, and it shows up as lower pressure rather than more output. The one field experiment on it, Kushlev and Dunn in 2015, put 124 adults through a week of checking three times a day and a week of checking freely. Daily stress came out lower in the limited week, Cohen’s d = .37. The study found no difference at all in how many messages people received or answered. The same work got done in fewer sittings.
The number underneath that is the one to set your target by. Told to check three times a day, participants reported a mean of 4.70. In the free week they reported 12.54, and before the study started, on an ordinary workday, 15.48. So the treatment that produced the result was roughly a two-thirds cut from wherever somebody started. If you’re at fifteen and you land on six, you’ve had the benefit.
The second finding says where that cut has to come from. Mark and colleagues at CHI 2016 watched 40 workers’ natural patterns for twelve days with heart monitors, instead of instructing them, and batching showed no relationship with stress. What their batchers had was fewer trips. People who clustered their email into a few heavy hours spent significantly less total time in the mailbox than the people spread evenly through the day, and checked significantly less often. The block is a device for getting the number down. A schedule of two beautiful sittings with a phone being swept in between is still a lot of trips.
One boundary comes from the 2015 authors themselves, who say their result may not hold for people with no choice about how often they look. If the mail is the work rather than an interruption to it, none of what follows is aimed at you.
So aim at trips. Everything below is about making a smaller number of trips actually hold.
The check you did not choose
The strongest practical argument for a schedule is arithmetic, and it’s worse than it looks.
In 2007, Shamsi Iqbal and Eric Horvitz published Disruption and recovery of computing tasks, a two-week log of what actually happened on the machines of 27 people at the authors’ own organization, in roles from program manager and administrator to software developer. Their primary work was interrupted by an average of 4.28 email alerts an hour. Just over 40 percent of those alerts drew a response inside 15 seconds. That’s faster than a considered decision.
Then the detour. After an email alert, the average time before returning to any suspended application was 9 minutes and 33 seconds. The paper’s own summary counts the whole thing: participants spent “on average nearly 10 minutes on switches caused by alerts, and spent on average another 10 to 15 minutes (depending on the type of interruption) before returning to focused activity on the disrupted task.” In 27 percent of cases the interrupted task was picked up again more than two hours later.
Call the whole event twenty to twenty-five minutes. Don’t multiply that by 4.28, because the windows overlap and the day wouldn’t fit. The honest reading is worse than a multiplication. At that rate a person is hardly ever outside a recovery window, so what’s left of the ordinary working hour is the gaps between detours. All of it for messages that would have been sitting there perfectly well at 11:00.
That reframes the question. The question is how many times a day your mail comes and gets you, and how much of the cost is incurred before you’ve read a word. A firm owner who looks eight times because eight moments suited them is in materially better shape than one who looks five times because five things pinged.
Which is why the alerts go before the schedule goes in, on the phone as much as the desk. Gmail’s help documentation lists three settings for desktop mail notifications, “New mail notifications on”, “Important mail notifications on” and “Mail notifications off”, delivered through Chrome, Firefox or Safari while Gmail is open in the browser (checked September 6, 2026). The middle one is the trap. It pulls you in less often, and somebody else still chooses when. The full account of badges, previews and what silencing them does and doesn’t fix is in why you check your email constantly.
A schedule that holds up in March
Willpower does a small share of the work here. Here’s a version to start from, built for a small professional firm and deliberately unambitious. Email advice mostly fails by being abandoned in the first hard week, and this one is shaped to survive that.
Three working passes at fixed times, plus one short scan.
A scan at 8:40. Ninety seconds, against a list. Answering, filing and reading through all belong to the 11:00 pass. You’re looking for the small number of things that would change what your morning is for, and you know what those look like because you wrote them down (that’s the next section). If the list stays clear, close it and start the real work. It’s a scan because the first hour is the only hour of the day you fully own, and handing it to whoever wrote to you at 11:00 last night is an expensive way to begin.
A working pass at 11:00. The substantial one. One rule holds it together: every message you open gets decided. The four endings a message is allowed to have, and the destinations that make that rule survive a heavy day, are set out in getting time back from your inbox.
A working pass at 2:30. The most valuable slot in the day, and the one most often skipped. A reply sent at 2:30 still reaches a client inside their working day, so it can still produce an answer today. The same reply at 5:15 arrives after their day is over, and it costs the firm a day of elapsed time on a thread that was only ever six minutes of typing. That gap between typing time and calendar time is counted properly in how much time firms lose to email.
A close-out at 4:45. Ten minutes, one question. Is anybody blocked overnight by something sitting in here? If so, send the two lines that unblock them. Otherwise, close it. This pass empties the evening check, which is the only reliable way anybody stops making it.
Three things separate a schedule from an intention.
Write down what each pass is for. A pass without a job turns into browsing, and browsing is what makes people conclude they need more passes. The scan looks for fires. The 11:00 clears. The 2:30 sends what other people need today. The 4:45 unblocks. Four sentences on a card.
Say the times out loud to the people who write to you. A good deal of the pull toward the inbox is the worry that somebody is waiting, unsure when you’re coming. A published window turns that worry into a fact. It works better as a firm norm than as a personal reflex: setting client response-time expectations.
On a bad day, drop to one pass and keep it. A practice that needs a calm Tuesday goes first on the week you needed it most. That test, and what makes any email habit survive, is email habits worth keeping.
One question this leaves open on purpose. Whether to look before you leave the house has no measurement behind it in either direction, so settle it on the one thing you can observe. Does what you find at 7:00 ever change what you do before 9:00? If it doesn’t, the early check is a ritual, and it’s paid for out of the part of the day where a ritual is most expensive.
What actually cannot wait
Every schedule meets the same objection: what if something urgent arrives. The answer is a list. Write down what genuinely cannot wait two hours in your practice. In most small firms the honest list runs to three to six items, and seeing it that short on paper is what makes the rest of this possible.
Part of the list may be decided for you. If your firm files returns, the clocks are published and counted in days.
For individual returns, IRS Publication 1345 (Rev. 12-2025) states that when the IRS rejects the electronic portion of a return and the preparer cannot fix the cause, the preparer “must take reasonable steps to inform the taxpayer of the rejection within 24 hours” (checked September 6, 2026). The same publication says a rejected individual return submitted on or before the due date is still timely “if the taxpayer corrects the electronic portion of the return and resubmits the return by the fifth calendar day after the due date of the return.”
For business returns, Publication 4163 (Rev. 12-2025) gives a Transmission Perfection Period of 10 calendar days from the rejection, and 5 calendar days for Forms 7004 and 8868, the extension applications. It’s emphatic about that window: it “is not an extension of time to file; it is a period to correct errors in the file,” and “the Transmission Perfection Period is never extended” (checked September 6, 2026). Weekends leave it the same length. So does the year-end cutover.
The practical consequence is specific. On a filing day, an acknowledgment for something you transmitted runs on a clock measured in hours, and it reaches you through your filing software, which in most practices means an email. So during a filing week you put one watch on one sender. That’s a small commitment, and a durable one. The version of this question where the answer changes for ten weeks and then changes back is inbox zero during busy season.
One thing stays off the list, and it’s worth naming, because a lot of low-grade checking is done in fear of it. The IRS opens a conversation with you on paper. Its current guidance on recognizing scams says flatly, “We never email without your permission,” and that it never sends direct messages by social media (checked September 6, 2026). The notice you’re afraid of missing is arriving in the post. It’ll still be there at 11:00, at 4:00, and on Thursday.
For the rest of the list, sort by what keeps rather than by what matters. A message can matter enormously and still keep until 2:30, and mixing the two is where most triage rules quietly go wrong: important versus urgent separates them properly. Once you have your three to six, cover each one with a standing watch on a named sender or condition, and keep the number of watches small. The test for what earns one, and why a person running nine has already given the game away, is setting your first watch in plain words.
The arrangement that makes it safe
A checking schedule is an arrangement, and it holds or fails on whether the arrangement covers the exception. The self-control part is small.
The clearest evidence for that comes from the largest dose anybody has administered. In 2012, Gloria Mark and Stephen Voida of UC Irvine, with Armand Cardello, published A Pace Not Dictated by Electrons at CHI. In it, 13 information workers were cut off from email entirely for five workdays. Their machines were logged, and researchers sat with them.
The focus effect was large. Mean window switches fell from 37.1 an hour to 18.2. The mean time a window was left open before switching away rose from 75.5 seconds to 131.9. People stayed on things nearly twice as long. What that did to their measured stress, and why the answer is more careful than it’s usually reported, belongs to why your inbox makes you anxious.
The part that matters here is what made five days without email survivable at all. Participants stayed reachable by phone, in person, and through colleagues walking over to tell them about anything critical that had gone to the whole group. Those colleagues, surveyed at the end of every day, reported that it was harder to get information out of the person on the cutoff, while their own satisfaction, productivity and stress were unaffected. The cost landed exactly where you’d expect and nowhere else.
So a schedule needs one of three fallbacks, and it needs it named out loud rather than assumed: a phone number your clients know they can use, a colleague who’ll interrupt you, or a watch that will. With one of them in place, this is a discipline. Without one, it’s a bet.
Four situations where the arrangement breaks down, and the schedule is the wrong instrument:
Your response time is the service. If clients hired you partly because you answer fast, or you’re the person covering a live matter today, a general schedule is fighting the product. Schedule the parts of the day that sit outside the live matter, and leave the rest alone.
Several of you work the same stream. If two or three people all handle the firm’s mail, personal schedules can make things worse: everyone’s own inbox looks clean while a client waits six days for a message each of them assumed somebody else had. That’s a question of visibility, and it wants a shared surface: one inbox for a two-person company, and at practice scale running the firm inbox.
You’re the only cover. In a one-person firm, the person who walks over to say this one is on fire is you. A schedule moves that risk around, and it stays with you. Worth knowing in advance rather than discovering.
The volume is a staffing problem wearing an email costume. A schedule redistributes attention, and the total stays the same. If what’s in there is mostly work that needs your judgment, spacing out the visits makes you later. That’s a capacity question rather than a timing one. Getting time back from your inbox has the honest test for telling the two apart, and Point versus hiring an assistant weighs the usual remedies, confidentiality included.
Setting one for the whole firm
The moment a checking rule stops being yours and becomes something you expect of staff, it changes category. It’s a wage and hour question now, and a small firm where HR is one more hat the owner wears is exactly the kind that finds this out late.
The underlying rule is old and simple. Work suffered or permitted is work time, and it counts wherever it’s done, including a couch at 9:00 at night. Those regulations read from the employee’s side, along with what they do and don’t give a person the right to refuse, are in protecting your evenings from email. What follows is the same rule read from the desk of the person who signs the checks, which is a different set of obligations.
The burden sits with the firm. 29 CFR 785.13 says management “cannot sit back and accept the benefits without compensating for them.” Then comes the sentence that catches most small firms: “the mere promulgation of a rule against such work is not enough.” A policy against after-hours email leaves the problem in place if the firm knows the work is happening and takes the benefit of it.
Two things narrow the exposure, and both are narrower than they sound.
The first is the de minimis rule at 29 CFR 785.47, which allows insubstantial periods beyond scheduled hours to be disregarded where they “cannot as a practical administrative matter be precisely recorded for payroll purposes.” It applies “only where there are uncertain and indefinite periods of time involved of a few seconds or minutes duration,” and the regulation itself cites authority that ten minutes a day does not qualify. Glancing at a lock screen is plausibly de minimis. Answering three client emails from the sofa sits outside it.
The second is a reporting route. The Department of Labor’s Field Assistance Bulletin No. 2020-5, issued August 24, 2020, says one way an employer exercises reasonable diligence about unscheduled hours is “by establishing a reasonable process for an employee to report uncompensated work time.” Where that process exists and is not used, the bulletin says reasonable diligence generally does not require the firm to go hunting, including through “records showing employees accessing their work-issued electronic devices outside of reported hours.” The conditions are real. The process counts only if the firm leaves it free to use, and only if the firm pays what gets reported.
What that adds up to for a seven-person practice:
For exempt staff, the schedule is advice. A partner choosing 11:00 and 2:30 is making a decision about their own day.
For non-exempt staff, it’s a boundary with a door in it. Their checking window ends when their day ends. If something genuinely needs them after that, there’s a phone call, a way to record the time, and the time gets paid.
Don’t write a rule you won’t enforce. A no-after-hours-email policy alongside a partner who sends at 10:00 at night and expects an answer is the exact fact pattern 785.13 describes.
None of this is legal advice. State law can be stricter than the federal floor, and a firm that’s been quietly relying on evening coverage for years should ask its own employment counsel rather than take a guide’s word for it. It belongs here because it changes the shape of the answer. A checking schedule is usually sold as self-discipline. For an employer it’s also a control, and a cheaper one than the alternative.
What changes when something else is watching
The objection that actually keeps people looking is a fair one. Knowing what’s in there means going in, so the only safe interval feels like always.
Point is an email app layered over the Gmail or Microsoft 365 account you already have, and Point answers that objection rather than arguing with the habit. Mail is weighed as it arrives, so what you sit down to at 11:00 is a feed led by what needs you today, with a plain summary on each thread. You decide from a line instead of opening things to find out whether they matter. The pass gets shorter, and a shorter pass is most of what makes a smaller number of passes hold.
For the short list from earlier, you say what you’re waiting for in one sentence, and Point tells you when it lands. Every standing watch is visible, and you can stop it. The rest waits for a briefing. The day arrives as one read, on a cadence you choose, and what’s worth an interruption is something you decide rather than something a notification decides. Nothing falls through the cracks and only what needs you, at the top are those two jobs in full, and the whole inventory is on the benefits page.
How much Point does unaided is set separately for each kind of work, on a range running from suggest-only through review to fully handled. Review is where every one of them starts, so the preparing happens and the doing waits until you move that particular kind of work up. What Point did is logged, and you can put it back.
The limits, plainly, because this is a piece about a schedule and you’re the one who sets it. How much mail turns up answers to a decision about who writes to you and through which channel, which sits outside the mailbox entirely. What your firm treats as urgent is yours to write, and a watch pointed at the wrong thing stays quiet in exactly the way an uneventful week does. Point can’t stop you opening the inbox at 10:40 at night. And a non-exempt employee’s after-hours reading is still hours you owe them.
Common questions
What time of day should I check email?
Fixed times, with a written job each. A workable default for a small firm:
- a ninety-second scan around 8:40, against a list of what would change your morning
- a full working pass at 11:00
- a second at 2:30
- a ten-minute close-out at 4:45, whose only question is whether anybody is blocked overnight
The 2:30 slot is the one to protect. A reply sent then can still get an answer inside the client’s day, and the same reply at 5:15 costs the firm a day of elapsed time on the thread.
Should I check email first thing in the morning?
Scan it, and leave the working pass for 11:00. A ninety-second look for the two or three things that would change what the morning is for is cheap and worth having. A full pass at 8:00 hands the only hour you completely control to whoever wrote to you overnight. The honest test is whether what you find early ever changes what you actually do before 9:00. For most people it doesn’t, and there’s no research settling this in either direction, so use your own week as the evidence.
What if something urgent arrives while I am not looking?
Write the list of what genuinely cannot wait two hours. In most small firms it’s three to six items. Then cover each one with something other than looking: a standing watch on that sender or condition, a phone number clients know, or a colleague who’ll interrupt you. In the study where 13 workers went five days with no email at all, that last one is what made it survivable, and their colleagues reported no effect on their own work beyond finding it harder to get information. A schedule with a named fallback is a discipline. Without one, it’s a bet.
Do I still need to check constantly on a filing deadline day?
No, but you do need one watch. IRS Publication 1345 (Rev. 12-2025) requires a preparer who cannot fix a rejection to take reasonable steps to tell the taxpayer within 24 hours, and gives a corrected individual return until the fifth calendar day after the due date to be timely. Publication 4163 (Rev. 12-2025) gives business returns a 10 calendar day perfection period, 5 days for Forms 7004 and 8868, and says it “is never extended” (both checked September 6, 2026). Those are clocks measured in hours and days, and an alert on the one sender that carries acknowledgments covers them while the rest of the inbox stays shut.
When should I not check email?
In the first hour of real work. In the ninety seconds before a meeting, when you can’t act on anything you find. And after your close-out has confirmed the overnight is clear. The meeting one is the most expensive and looks the cheapest: all you can do with what you find is carry it into the room unresolved.
Can I tell my staff not to check email after hours?
You can, and if any of them are non-exempt you have a reason to. 29 CFR 785.13 says management “cannot sit back and accept the benefits without compensating for them,” and that “the mere promulgation of a rule against such work is not enough.” So a written policy alone settles nothing if the firm knows the work is happening. The workable arrangement is the one the Department of Labor points at in Field Assistance Bulletin No. 2020-5, from August 24, 2020: a reasonable process for reporting unscheduled time, which the firm neither discourages nor ignores, and which it actually pays. This is not legal advice, state law can be stricter, and a firm that’s been relying on evening coverage should ask its own counsel.
How do I tell clients I am only reading email three times a day?
As a service commitment rather than a personal preference, and with a faster route alongside it. One window, the same for everybody in the firm, published, plus a phone number for the case that has to move today. That does more for the schedule than any amount of resolve, because most of the pressure to check is the fear that somebody is waiting, unsure when to expect you. Arriving at the number, and what to say when you’re going to miss it, is setting client response-time expectations.
The short version
Aim at the number of trips. Checking less has been tested, and it lowers daily pressure while the same amount of mail gets handled. Batching arranged for its own sake showed no relationship with stress. A two-thirds cut from wherever you are now is the size of move that produced the result.
Take the alerts out first, because the arithmetic on an interruption is worse than it looks: nearly ten minutes on the detour and another ten to fifteen before real focus comes back, four times an hour. A schedule with notifications running is the same day with guilt added.
Then run three working passes and a short scan, each with a written job, and protect the mid-afternoon one, because it’s the only pass whose replies can still produce an answer today. On a bad day, drop to one pass and keep it.
Build the whole thing around a written list of what actually cannot wait, which is shorter than the fear suggests, and some of which is published in days by the IRS if you file returns. Much of what you’re afraid of missing arrives by post. Cover the real list with a watch, a phone number or a colleague, because the exception is what makes the schedule safe rather than lucky.
And when you set it for anybody other than yourself, it stops being a productivity tip. For non-exempt staff, after-hours email is hours, and the rule has to be one you’re willing to enforce in both directions.