How much time firms lose to email, measured

The two most careful measurements of an ordinary working day put email at somewhere between one hour twenty minutes and two hours per person, per weekday. For a seven-person firm, that’s between one and one and a half full-time people, every week, forever. Nobody can hand you your own firm’s number. The published figures differ by a factor of four, because each one measures something different.

That range is worth more than a single figure. You can see its edges, and you can check where your own firm sits.

• The number is between an hour and two hours a day, per person. One logging study clocked 1 hour 23 minutes a day. One diary study of a much heavier mailbox clocked about two. Anything you read above three hours is almost certainly self-reported.
• One person’s hours times headcount gives the wrong total. Internal mail is paid for at both ends. The load is lopsided too: a small number of people carry most of it, and in a small firm that person is usually the owner.
• A large share of the elapsed time sits outside the mailbox. It’s a client waiting for a reply while the thread stays open across two days.
• You can measure your own firm in a week. Three counts, no timesheet. The counting matters more than the benchmark.

This piece is about the size of the number and how to get yours. What to do once you have it sits next door. Clearing a backlog, the filing that doesn’t repay itself, and the finding costs are all in getting time back from your inbox. Turning hours into dollars is a separate and slipperier exercise, and it’s the real cost of email overload. The costs that show up as something other than hours belong to the hidden cost of a cluttered inbox.

The two numbers somebody actually measured

Most of the figures in circulation come from surveys, where people are asked to estimate. Two studies did something better. Neither ran on a small accounting or consulting practice, and that limit is worth carrying through everything below.

The logged number: 1 hour 23 minutes a day. In 2016, a team led by Gloria Mark at UC Irvine, with Shamsi Iqbal, Mary Czerwinski and Paul Johns at Microsoft Research and Akane Sano at the MIT Media Lab, published Email Duration, Batching and Self-interruption at the CHI conference. They put activity logging software and a chest-worn heart rate monitor on 40 information workers. Then they left them alone at their normal desks for 12 workdays. The software recorded which window was in the foreground, second by second.

Two things in that table matter more than the headline. The first is the ratio. 1 hour 23 minutes out of a logged computer day of 4 hours 34 minutes is about 30 percent of the time these people spent at a keyboard. The second is the spread. The range runs from a person who never opened email to a person who spent nearly eight hours in it, on the same day, in the same building. An average describes a group that varies that much poorly, and that’s the first reason to count your own.

The third figure is the one people find hardest to believe about themselves. A median of 58 separate visits to the mailbox is a visit every eight or nine minutes across a normal working day. That’s the shape of the time as much as the size of it. How often to look at the inbox is its own decision, worked through in when to check your email.

The diary number: about two hours a day. Stephen Barley and Debra Meyerson of Stanford, with Stine Grodal of Boston University, published E-mail as a Source and Symbol of Stress in Organization Science in 2011, from data collected between October 2001 and March 2002. Seventy-nine employees of a large workstation and server company kept a contemporaneous log of every communication event over two full workdays. They recorded start time, end time and how many messages were involved.

Their average respondent worked 9.4 hours a day and spent 6.4 of those communicating. Email took 31 percent of the communicating. Across the two logged days the mean was 4.0 hours doing email and 185 messages handled: about two hours and about 92 messages a day.

That study is old, and it predates the smartphone entirely, which cuts both ways. Volume has gone up a great deal since. Against that, its participants logged every event by hand. That catches email done away from the machine, which the foreground-window method misses. The two numbers sit together rather than in conflict. They’re the floor and the ceiling of the same estimate, measured from opposite ends.

Barley’s team found something else an owner should read twice. The more email a person handled, the longer their working day ran. The feeling of being overloaded had no statistical relationship to hours worked at all. Hours and strain are two different quantities. This page is about the hours, and the other one is why your inbox makes you anxious.

Why the published figures disagree by a factor of four

Go looking and you’ll find confident numbers between half an hour and four hours a day. They can’t all be right, and the researchers say so plainly. The CHI 2016 authors reviewed the prior work. They list an average of 87 emails received per day from one study and 34.5 minutes a day on email from another. Their conclusion: “these disparate estimates could be due to a variety of factors, such as culture, workplace, and measurement technique.”

Three of those choices move the number more than anything about the people being measured.

What counts as email time. Foreground-window logging counts the seconds the client is the active window. It leaves out reading on a phone. It leaves out thinking about a message while you look at something else. It leaves out every minute of the work a message caused. A diary counts email done anywhere, and relies on the person to notice and write it down. These differences are large. They’re most of the gap between 34 minutes and two hours.

Whether the work email creates is inside the number or outside it. A message that says “can you confirm the depreciation schedule before Thursday” takes forty seconds to read and forty minutes to answer properly. Nearly every published figure puts the forty seconds in and the forty minutes out. That’s defensible, because the forty minutes is the work you’re paid for. It also means the headline understates what the inbox governs. What the inbox governs and what it consumes are two different quantities.

Whose day it is. In the logged study the range on a single day ran from zero to nearly eight hours. Job role explains more of that than any national or generational story. A small firm has a wider spread of roles than a large department does.

There’s another reason to treat a headline carefully. Most of them rest on asking people to estimate their own use, and people estimate badly. That failure has been measured, and the real cost of email overload handles it. The practical consequence is simple enough. Your own guess about your own inbox is a guess, and so is your team’s.

A firm’s number is not one person’s times headcount

This is where a firm-level count comes apart from a personal one. It’s also the part almost nothing written on this subject addresses.

Internal mail is paid for twice. You write to your senior on Tuesday. The firm buys your composing time and their reading time, and a reply buys two more. A message that travels between two colleagues costs the firm at both ends. The industry benchmarks show it as two separate people’s ordinary usage. In a practice where most correspondence is with clients, this is a small correction. In a practice that runs its internal coordination through email, it’s a large one. It’s also the one part of the total your firm can cut on its own.

The load is lopsided, and predictably so. Rob Cross, Reb Rebele and Adam Grant wrote Collaborative Overload for the January to February 2016 issue of Harvard Business Review, drawing on two decades of organizational network analysis. They report that “the time spent by managers and employees in collaborative activities has ballooned by 50% or more”. At many companies, “people now spend about 80% of their time in meetings or answering colleagues’ requests”. The most useful finding here is the distribution, which runs extremely uneven: “in most cases, 20% to 35% of value-added collaborations come from only 3% to 5% of employees.”

Three to five percent of a seven-person firm is one person. You already know which one. Their name is on the engagement letters, and they get copied on everything as a courtesy. An average across the firm hides that entirely, so the count below runs per role rather than per head. What that concentration does to the person carrying it, and what can be handed off, is email overload when you own the firm.

Part-time and seasonal staff break the arithmetic. A three-day-a-week bookkeeper doesn’t carry three fifths of a full mailbox. They carry a full mailbox that arrives five days a week and gets touched on three. That’s why their Monday is the worst one in the building.

The hours that are spent waiting

There’s a second clock running. For a professional practice, it’s often the one that matters to the client.

The largest measurement of email conversations available is Evolution of Conversations in the Age of Email Overload, presented at the World Wide Web conference in 2015 by Farshad Kooti and Kristina Lerman of the USC Information Sciences Institute with three colleagues at Yahoo Labs. They analyzed more than 2 million users exchanging 16 billion messages. It’s consumer Yahoo Mail rather than business mail, so read the shapes rather than the exact values. The shapes are striking.

Individual replies are fast and very short. The median reply is 43 words. The single most common reply length is five words. Half of all replies go out within 47 minutes of the message arriving. How to read that distribution when you’re the one waiting, and when a quiet thread has actually gone quiet, is the email nobody answered.

Threads are slow. A thread averages 3.76 messages and the median thread runs 3.5 hours. The mean time span is 53.2 hours, and 22 percent of threads last longer than a day.

Put those together and you have the arithmetic a per-person time figure misses. Four messages, at five to forty-three words each, is perhaps six minutes of anybody’s typing. The exchange occupies two days of the client’s calendar. Say that exchange stands between you and starting a return. The firm lost two days of a slot on the schedule, and the six minutes is the only part that shows up in any of the studies above.

The same paper holds a second finding worth keeping. As daily load rises, people reply to a steadily smaller share of what arrives. It runs from about 25 percent of everything received on a light day down to under 5 percent at around 100 messages a day. The messages you stop answering stay where they are. They become the backlog, and a backlog is carried rather than spent, which is exactly why it never appears in a time study.

Seven people, counted

Here’s the arithmetic laid out for an ordinary small practice, so you can substitute your own numbers. It’s arithmetic on the ranges above rather than a research finding. The per-role hours are estimates, and you should replace them within a week.

Fifty-five hours a week is about 1.4 full-time equivalents. Over a 46-week working year, allowing for vacation and holidays, it’s roughly 2,500 hours. If you prefer the flat version, seven people at the logged mean of 1 hour 23 minutes each comes to 48 hours a week, and seven people at the diary mean of two hours comes to 70. The weighted total sits where you’d expect. The weighting is the honest part, because it puts nearly half of the firm’s email time on two of the seven desks.

Two cautions about that number before it goes into a slide.

It’s the size of the activity, not a saving. That contrast is worth spending, because a firm that halved it would gain less than 27 hours of client work. Some of what’s in there is client work already.

The second caution is about dollars. Multiplying 2,500 hours by a billing rate produces something between misleading and absurd, for reasons worth understanding before you do it. That conversion is worked through properly in the real cost of email overload.

Counting your own firm in one week

Three counts, one week, no timesheet. Take them per person and keep the roles separate, because the average will lie to you.

Count one: volume, from the sent folder. The messages you send are the cleanest proxy for the messages you handled, and unlike received mail they’re all yours. Gmail’s help documentation lists the operators you need (checked September 6, 2026):

• from: finds mail from a specific person, with from:me given as an example.
• after: and before: set a date range.

One search per person over one full week gives you a number nobody has to remember or estimate. Any client that can sort a sent folder by date does the same job.

Count two: duration, by sampling rather than stopwatch. Sampling wins here, and it asks less of you. Timing yourself fails twice: you forget, and the act of timing changes the thing. So set a phone alarm to go off eight times across a working day, at times you didn’t choose. Each time it goes, write down one letter: E if you were in email at that moment, W if you were doing other work, O for anything else. Do it for three days. The share of E marks, multiplied by your working hours, is your email time. It beats any recollection, because it never asks you to recall anything.

Count three: latency, on threads you care about. Take the last ten client threads that ended in a decision. For each one, note the hours between the first message and the last, and the number of messages it took. The medians of those two numbers are the counts almost nobody keeps, and they’re the ones a client would recognize. They’re also the counts that move when you change how you work rather than how fast you type.

Then leave them alone. Take the same three counts again in eight weeks. A single reading is a benchmark, and it’s worth very little. Two readings eight weeks apart are an instrument.

Which of these hours can come back

The useful question is which parts of the number are addressable, and the honest answer separates into four.

The reading and the judging are the work. A partner deciding whether a client’s question needs a call or a paragraph is doing the job. That time is the job, and it should stay where it is. No tool and no habit should aim at it. Sort two weeks of mail honestly, and if this is most of the pile, what you have is a capacity problem rather than an email problem. The comparison you actually want is Point versus hiring an assistant.

Working out which messages those are is the largest recoverable slice. In a mailbox taking 90 to 120 messages a day, most of the time in front of the client goes on working out which messages need you. This is the part a ranked, summarized queue changes. It’s also the part that most of the published time figures are actually measuring.

Finding things again, and filing so you can, is recoverable and well studied. It has its own evidence and its own section, in getting time back from your inbox, and re-deriving it here would only cost you time.

The arrival rate stays where it is, whatever you install. How much mail comes in answers to a decision about who writes to you, about what, and through which channel. Nothing installed in the mailbox touches it.

One caution about the interventions with a good reputation. The same 2016 study that produced the duration figures above also tested batching, the practice of gathering email into set sittings. It reported that “despite widespread claims, we found no evidence that batching email leads to lower stress”, even where it did move the productivity rating. A habit can shift one of these quantities and leave another untouched. That’s the argument for keeping the counts apart, rather than collapsing them into a single feeling about the week. That result in full, and what it means for how often to open the mailbox, is why you check your email constantly. Which habits survive contact with a real practice is the email habits worth keeping.

Anyone quoting you a percentage that software will give back, including us, is guessing. What you can establish for your own firm in twenty minutes is the ceiling. Take your last 40 messages and mark each one as needing a decision from you or needing only to be seen. The share that only needed to be seen is the largest that any change to your setup could possibly touch. Everything below that is a real estimate, and everything above it is marketing.

What Point changes about the count

Point is an email app that runs on top of the mailbox you already have, Gmail or Microsoft 365. You keep your address, and the mail stays where it is. Point changes the second of those four: the time spent working out which of today’s messages need you.

You sit down to a queue that’s already ranked for what needs you today. Each thread carries a plain summary, so the establishing happens before you open anything. On count two, the duration sample, that should show up as fewer visits and shorter ones. On count three, an ask inside a message becomes a dated item, and the scheduling back-and-forth is run for you rather than by you. That should show up as fewer round trips inside a thread, which is the part of the latency you control. Only what needs you, at the top and nothing falls through the cracks describe those two jobs in full, and the whole inventory sits on the benefits page.

You set autonomy per kind of work, and there are three positions: suggest only, review, or handled without you. Everything ships on review. The preparing happens, and the doing waits until you raise it, one kind of work at a time. What Point did sits on an activity log, and you can put it back. One exception applies, and no email software escapes it: a message that has been delivered is on somebody else’s server.

The limits are worth stating against the counts above. Point leaves the arrival rate alone, and that’s the single biggest term in the whole equation. The forty minutes of real work a client’s question creates stays forty minutes, and it should. A thread waiting on somebody outside your firm waits as long as it ever did, and Point’s contribution is needing fewer of those waits. The counting stays yours. Take the three numbers above before and after, because they’re the only way you’ll know whether anything changed.

Common questions

How many hours a week does the average person spend on email?

Between about 7 and 10 hours, on the two most careful measurements available. A 2016 study logged 40 information workers for 12 workdays and measured a mean of 1 hour 23 minutes a day with the email client in front, or roughly 7 hours a week. A diary study of a heavier mailbox recorded about 2 hours a day, or 10 hours a week. Figures much above that are usually self-reported estimates. Figures much below usually count only the seconds the email window was active. The spread within any one group is wider than the gap between the studies: on a single logged day the range ran from zero to nearly eight hours.

How much time does a small firm lose to email in a year?

For a seven-person professional-services firm, something in the region of 2,500 hours a year, or about 1.4 full-time equivalents. That comes from weighting per role rather than averaging: an owner and an administrator at around 2.5 hours a day, seniors at 1.5, staff at 1. Treat it as the size of the activity rather than a saving, because a good part of those hours is client work being done.

Why do the published numbers disagree so much?

Because they measure different things, and they say so. The authors of the 2016 logging study reviewed the prior work. They found published figures as far apart as 34.5 minutes a day spent on email and 87 messages a day received, and attributed the disparity to “culture, workplace, and measurement technique”. Three choices move the number most. Whether email done away from the computer is counted, whether the work a message causes is inside the figure or outside it, and whose job is being measured. Before quoting any figure, check which of those three choices it made.

Does a firm lose more time to email than to meetings?

Not usually, taken across a whole organization. Harvard Business Review’s “Collaborative Overload” reported that at many companies people now spend about 80 percent of their time in meetings or answering colleagues’ requests, without splitting the two. The 2011 diary study did split them, and put email at 31 percent of communication time against 34 percent for meetings and unplanned encounters. Email is distinctive because it arrives in dozens of small pieces spread across the day, which is a different problem from a two-hour meeting.

How do I measure how much time my team spends on email?

Sample rather than time. Have each person set a phone alarm to go off eight times a day at times they didn’t choose, and note whether they were in email at that instant. The share of email marks times their working hours is the estimate, and it beats recollection because it never asks anyone to remember anything. Pair it with a count of messages sent per person over one full week, taken from the sent folder rather than from memory. Repeat both in eight weeks, because one reading is a benchmark and two are an instrument.

How many emails a day is too many for one person?

Around 100 messages a day is where the share you reply to collapses. The largest study of email conversations found people replying to about 25 percent of what arrives on a light day and under 5 percent at roughly 100 messages a day. Something does change at that threshold: the unanswered share stops being a queue and starts being a backlog, which is carried rather than worked. If your daily arrivals are in three figures, expect the problem to show up as things quietly going unanswered rather than as long hours in the mailbox.

How much of this time can software actually give back?

Nobody has measured it honestly, so treat every published percentage as a guess. What you can establish for your own firm is the ceiling. Take your last 40 messages and mark each one as either needing a decision from you or needing only to be seen. The second group is the largest share any change to your setup could possibly reach, and the realistic gain is some fraction of that. In a mailbox at 90 to 120 messages a day the seen-only group is usually the majority, which is why ranking is worth more than typing faster.

What the count comes to

Two careful measurements bracket the personal number at between one hour twenty minutes and two hours a day. Everything wider than that bracket is either a survey or a definition you have yet to check. For a seven-person firm, weighted by role rather than averaged, that comes to about 55 hours a week and something near 2,500 hours a year: close to one and a half people, doing email.

The firm’s number behaves differently from the person’s in three ways worth remembering. Internal mail costs the firm twice and shows up once. The load concentrates on a few desks, and in a small practice that means one, usually the owner’s. And a large part of the elapsed cost is a thread sitting open for two days over six minutes of typing, which no per-person study will ever show you.

So take three counts rather than one benchmark. Messages sent per person in a week. An eight-times-a-day sample of whether you were in email. The hours and message count on your last ten client threads that ended in a decision. Take them again in eight weeks. Then sort forty messages into the ones that needed a decision from you and the ones that only needed to be seen, because that ratio, and not anybody’s percentage, is the size of what a change to your setup could reach.