Somebody quotes you a figure. Twenty-eight percent of the workweek. Thirteen hours. Two and a half hours a day. You do the multiplication in your head and land on an annual number with four or five digits in it. You feel briefly alarmed. Then the day carries on. You can spend against a number you worked out yourself, and that one came off somebody else’s slide.
So this piece builds one you can check.
Start with the honest headline. The hours are the easiest line on the bill and the smallest. Most articles on this subject count that line alone, which is why the total reads as scary and slightly unreal at the same time. The full account has six lines on it.
- The hours you spend inside the mailbox.
- The hours either side of it, getting in and getting out.
- The version of you that does the rest of the work afterward, which the measurements say is a different version.
- The judgments, which cost more than their duration and which nobody counts.
- The work that never happened because those hours were already spoken for.
- The occasional expensive miss, which is rare, lumpy, and larger than everything above put together.
Only the first takes a stopwatch. The rest are real anyway, and three of them estimate well enough to decide on. That’s the goal here: a number you’d defend out loud, and a sense of which parts of it a change to your setup could plausibly touch.
If it’s the feeling you’re after rather than the bill, why your inbox makes you anxious is the piece next door. It goes at the mechanism rather than the money.
The figure you have already seen
Almost every claim about the cost of email traces back to one source. Here’s the one.
In July 2012 the McKinsey Global Institute published The social economy, a report on what social technologies might be worth to business. Buried in it is the sentence that’s been quoted ever since: “The average interaction worker spends an estimated 28 percent of the workweek managing e-mail and nearly 20 percent looking for internal information or tracking down colleagues who can help with specific tasks.”
Four things about that sentence. The retelling usually drops all four.
It’s an estimate, and the report says so. It covers “interaction workers”, meaning managers, professionals and salespeople whose job is largely dealing with other people, so the estimate describes that group rather than the whole workforce. It says managing email, which is broader than reading and answering, and broader than most people picture. And it’s from 2012. That’s before Slack, before Teams reached most desks, and before the phone in your pocket became the second place your mail arrives.
The famous “13 hours a week” appears nowhere in the report. It’s what you get when you apply 28 percent to a 47-hour week. That’s a fair thing to do, and it’s arithmetic rather than a measurement. Apply it to your own week and you get your own figure: 11 hours if you work 40, 14 if you work 50. For that same arithmetic run against the price of software, what an AI email client costs does it there.
The contemporary reading is volume rather than hours. Breaking down the infinite workday, published by Microsoft on June 17, 2025, draws on anonymized Microsoft 365 signals up to February 15, 2025, plus survey responses from 31,000 knowledge workers in 31 markets. It puts the average at “117 emails daily”, most of them “skimmed in under 60 seconds”. That’s telemetry from one company’s customers rather than a national statistic, and the 31 markets reach well beyond the United States. Take it as an order of magnitude. How to skim a full inbox in minutes works that number into a time budget properly.
Both figures are useful for orientation and useless for a decision. Both are about somebody else.
Your own estimate is wrong in an unknown direction
The next move is to estimate your own hours. Most people produce a figure in about four seconds. An hour a day, maybe ninety minutes, more in a bad week.
That figure comes out wrong, and the interesting part is which way.
In 2021, Douglas Parry and colleagues published a systematic review and meta-analysis in Nature Human Behaviour. They compared what people said about their own digital media use against what device logs recorded. Across 106 effect sizes, the correlation between the two was r = 0.38. That’s a real relationship and a long way short of a substitute. Of 49 direct comparisons of means, only three, about six percent, landed within five percent of the logged figure.
Then the part that matters for a cost estimate. Twenty-three of those comparisons over-reported and twenty-three under-reported. The authors conclude that the evidence is “insufficient to conclude whether estimates are typically under- or over-reported”. The error runs both ways, person by person. So you’re in ordinary company here. People are simply bad at this, each in whichever direction they happen to be bad.
The study covers phones, computers and social media rather than email in particular. Its authors sit in South Africa, the United Kingdom, the United States and Norway rather than in any one country’s workforce. What it establishes is a fact about self-report, and your own inbox stays unmeasured. It kills the shortcut all the same. A guess can’t carry a defensible cost figure, and a margin can’t rescue the guess, because you’d have to know which way to add it.
So measure it. That takes a week.
Log one ordinary week. A time tracker that records which application is in front of you does the whole job on its own, and there are several. If you’d rather keep your machine as it is, use a scrap of paper. Write the clock time each time you go into the mail and each time you leave it. Paper is the rougher method. The gap is small, and paper starts today.
Pick a week that’s boring on purpose. A week with no filing deadline in it, no return from vacation, nothing launched. Those weeks are real, and they price differently. If two ordinary weeks are within reach, take two. A single week carries more noise than most people expect.
Count going in and coming out, along with being in. A forty-second drop into the inbox costs more than forty seconds. Where the rest of it goes is the subject of email overload and why important mail slips past. That piece puts a measured number on the reorientation, which is why this one leaves it there.
Write down the count of separate visits, too. How many times you went in matters more than the total. The next section is the reason.
At the end you have two figures. Hours in the mailbox per week, and visits per week. Those are the only two inputs the rest of this needs.
Interrupted work gets done faster and costs more
Here’s the finding that makes the stopwatch method understate the bill. It’s the most surprising result in this whole area.
At CHI 2008, Gloria Mark of the University of California, Irvine, working with Daniela Gudith and Ulrich Klocke at the Institute of Psychology at Humboldt University in Berlin, published The Cost of Interrupted Work: More Speed and Stress. Forty-eight participants, mostly German university students, were given an email task to complete. Some were interrupted partway through, by phone or by instant message. The interruption was either about the subject they were working on or about something unrelated. Some worked straight through.
The expected result was that the interrupted groups would be slower. They were faster. Mean time to finish was 22.77 minutes with no interruptions, against 20.31 and 20.60 minutes in the two interrupted conditions, and the difference was statistically significant. Errors held steady. So did politeness.
What moved was everything a clock misses. On a scale of 1 to 20, rated stress went from 6.92 uninterrupted to 9.46 and 9.13. Effort went from 9.50 to 11.04 and 11.52. Frustration, time pressure and mental workload all moved the same way. The emails themselves got shorter, too: a mean of 31.49 words when nobody interrupted, 29.17 and 30.16 when somebody did.
The authors’ reading is the useful one. “People compensate for interruptions by working faster, but this comes at a price: experiencing more stress, higher frustration, time pressure and effort.”
Two caveats before anyone builds a policy on it. It’s a laboratory task with a defined finish line, and most real work runs looser than that. The participants were students in Germany rather than American professionals with clients. The finding describes how people respond to being interrupted, and it prices no particular job. Nobody has run the equivalent experiment on a working week.
Take the direction seriously, because it changes what your measured hours mean. You time yourself, you find forty-five minutes a day, and the natural conclusion is that email is fine. This study suggests a short measured time can be the symptom rather than the reassurance. You got fast because you had to. The cost came out somewhere the stopwatch was pointing away from: in effort, in strain, and in work that came out slightly thinner than it would have.
That’s the third line on the bill. It’s also why a page offering you one dollar figure for email overload is selling you something. Nobody can price a shorter email.
The lines that never reach a timesheet
Three more. They rise in what they’re worth and fall in how easily you can count them.
The judgments. Each message you rank is a small decision. You make it fast, on partial information, and it draws down something the work later needs. The count of those decisions is what matters, which is why an inbox of 300 routine newsletters is easier than 40 messages from clients. Why the ranking lands on you at all, and why folders leave it with you, is set out in email overload and why important mail slips past. For the ledger, note the number and leave the column blank. It’s real, and it carries no price.
The work that got displaced. This one is countable, and almost everyone skips it. The hours in your mailbox came out of somewhere. They rarely come out of client work, because client work has a deadline attached and email feels like it has one too. They come out of the thing with no deadline: the follow-up to the proposal, the case study you were going to write, the two calls a week that keep the pipeline from thinning in March. Ask what you left undone this week, and how long it’s been sitting there. The question has a specific answer. The answer usually carries a dollar value you already know.
The expensive miss. Once in a while a message that mattered goes past. Nine days later a client is gone, or a deadline is, or a supplier has gone quiet. This is the largest line on the ledger and the hardest to plan against, because a two-second oversight and a five-figure loss are the same event. The email that cost you a client follows one all the way through, including the part where a lost referral source turns out to be worth more than the lost fee. An honest expected value is beyond anybody. What you can say is that once it’s happened to you, it counts for more than a rounding error.
Two more costs sit outside this piece on purpose. Mail that follows you past six and takes an evening with it is protecting your evenings. The version of all this that lands on somebody with a payroll to meet, where the same hours come out of everyone else’s week too, is email overload when you own the firm.
Turning hours into dollars without kidding yourself
Now the arithmetic, and the two places it goes wrong.
The first mistake is multiplying every mailbox hour by a billing rate. Ten hours a week at a full rate produces an annual figure with six digits in it, and it’s nonsense. None of those hours were going to be billable, and some of that email is the client work. Nobody hands you back ten sellable hours.
The second mistake is the overcorrection, which is to decide the whole exercise is unserious and stop. The exercise holds up. It just needs two honest inputs rather than one.
Price the hour at what the next one is worth. The marginal hour, in other words, rather than the average one. Fill a recovered hour with billable work and the marginal rate is your billing rate. Fill it with the business development that’s been sliding and the price is what that work returns, which usually runs higher and less certain. Fill it with going home and the answer stops being a dollar figure. That’s a legitimate answer, and it ends this section for you.
Apply a conversion fraction, and pick one you’d defend. Of the hours a better setup gives back, some fraction becomes work that earns. The rest becomes slack, which has a value of its own and stays off the money line. A quarter to a half is the range most people can argue for with a straight face. A figure built on 100 percent conversion is a sales pitch rather than arithmetic.
So the formula is short:
weekly hours in the mailbox × the fraction you would realistically convert × the rate the converted hour earns = the weekly cost of the first line only
The first line only. That’s the point of writing it out. What comes back is an underestimate by construction. It leaves out the reorientation, the compression, the judgments and the tail risk. If the underestimate is already big enough to act on, stop there and act. If it comes out small, your week is going somewhere other than email, and you’ve just saved yourself a purchase.
One ordinary week, priced
Take a solo consultant. She works about 47 hours in a normal week and bills 22 of them at $185. She logs a week and finds 8.4 hours in her mail client, across 31 separate visits.
Her first instinct is 8.4 × $185. That’s $1,554 a week, and a little over $71,000 across a 46-week working year. She knows the number is wrong, because 47 billable hours was always out of reach.
So she uses the formula. Of the 8.4 hours, she reckons a third turns into work that earns. Her constraint is the two proposals a month she keeps putting off, rather than desk time. A third of 8.4 is 2.8 hours. At $185 that’s $518 a week, and across 46 working weeks, about $23,800 a year.
Then the lines the formula leaves out, which she writes down anyway rather than pricing:
- Thirty-one visits a week means thirty-one departures and thirty-one returns, and the reorientation on each one sits outside the 8.4.
- Somewhere in those 8.4 hours are several hundred small rankings, and no column holds them.
- The two proposals a month have been two proposals a quarter since February. She can name the client she thinks she lost that way, which makes it a real cost rather than a theoretical one.
- She has never yet had the expensive miss. She has come close twice.
Twenty-three thousand eight hundred dollars is the conservative, defensible figure, and it’s the one to decide against. Notice where it came from: her own log and her own rate. The 28 percent did no work at all.
Run yours. It takes ten minutes once you have the week.
What the number is for
A cost figure earns its keep by deciding something. Here’s what it decides.
It sets a ceiling on what a fix is worth. A defensible figure of a few hundred dollars a week makes software cheap against it, or an assistant, or a different working pattern, or all three. A small figure makes the same purchase a bad idea, and you’d be turning it down for a reason rather than a feeling.
It tells you which fix. Each remedy lands on a different line. Sorting rules and folders move mail around and leave the ranking with you, so they touch line one a little and line four not at all. Turning off notifications cuts the visit count, which is line two, and leaves line one where it is. Answering faster raises the visit count. Getting to zero resets a number that was never on the bill, which is the argument in a calm inbox versus inbox zero. Something that ranks the mail and tells you what’s in a thread before you open it goes at lines one, four and six together. That’s the change with the largest reach, and it sounds like the smallest.
It tells you when to stop. The visit count is the number most people can move in a week, alone and for free. You decide once when mail is worth reaching you, in place of deciding it forty times a day. If that brings your figure somewhere you can live with, you’re done, and the rest of this is a purchase you can skip. Why you check your email constantly is the piece on making that stick.
The number leaves one thing alone: how you feel about the unread count, which was never proportional to anything. Unread email guilt is a separate matter, and arithmetic stays outside it.
What Point takes off the ledger
Point moves the ranking off you before you arrive. So here are the lines of the bill that’s aimed at, and the ones Point leaves where they are.
Line one, the hours in the mailbox, shrinks because you open a short ranked feed. Today’s handful sits first. Each thread carries a plain summary, so you can take in a long conversation without reading it through. Line four, the judgments, is where Point goes hardest. The ranking is done before you sit down, so those several hundred small decisions a week are already made. Line six, the expensive miss, is what the whole arrangement exists for. A message from a client with live work and an unanswered question ranks above the newsletter and the receipt. Anything asking something of you becomes a dated item, and it comes back at the moment chasing it is useful. That’s what nothing falls through the cracks covers.
How far Point goes unaided is yours to set, one type of action at a time. The range begins at suggest-only, passes through review, and ends at fully handled. Every action arrives set to review, meaning Point does the preparation and then stops. Raise any of them and that action stops waiting. Point’s own actions are logged and reversible, short of a message already delivered, which no software can retrieve from the recipient’s server. The whole set of changes across a week is on the benefits page, and a calm inbox that sorts itself is the part that speaks to line one.
Three things stay where they are, and they matter for the arithmetic. The same volume of mail still arrives. A hard reply is still hard to write. And recovered hours become revenue only through your conversion fraction, which remains the number most likely to be optimistic. If your ledger came out small, the sum has told you to keep your money, and that’s a perfectly good outcome of having done it.
Common questions
How much does email overload cost per employee per year?
No credible general figure exists. The ones in circulation nearly all come from the same 2012 McKinsey estimate multiplied by an assumed salary, and that chain has three guesses in it. The figure you can defend is your own logged hours, times a conversion fraction, times the rate a recovered hour actually earns. It comes out smaller than the ones you’ve seen, and it’s worth more.
Is 28 percent of the workweek on email still accurate?
It was an estimate for interaction workers in 2012, from a McKinsey Global Institute report on social technologies. Nobody has re-measured it by the same method since. Message volume has gone up and chat has arrived alongside email, so the figure could still be in range. Treat it as an order of magnitude for a category of worker, and measure your own week for the rest.
I only spend an hour a day on email. Is that fine?
Possibly, and check the visit count before you settle it. An hour in one sitting and an hour in forty pieces are different costs. The Mark study found that people under interruption speed up rather than slow down, so a low measured time can be compression at work. If the hour is one visit, you’re fine.
What is the single biggest cost of email overload?
The one that almost never happens. A missed message that turns into a lost client outweighs a whole year of the hours. It arrives rarely and in lumps, so being more careful in proportion leaves it roughly where it is. That asymmetry is the honest reason to change the setup rather than the effort.
Does answering email faster reduce the cost?
It reduces one thing and raises another. Faster replies shorten the window in which a sender decides you’ve gone quiet, which protects line six. They also raise the number of visits and the number of judgments, which are lines two and four. Speed at the top of the list, on the same number of trips to the list, is the combination that helps.
The bill in one paragraph
The hours are the smallest true line, and the only one anyone counts. The famous number for them is a 2012 estimate about a category of worker rather than a measurement of you. Your own guess will land off, in a direction you can’t predict, so log one ordinary week and note both the hours and the number of separate visits. Price the hours at what a recovered hour would actually earn, times a conversion fraction you’d defend out loud. Treat the answer as a floor. It leaves out the reorientation, the compression that makes interrupted work faster and worse to do, the several hundred rankings nobody counts, and the rare miss that costs more than all of them. Then put the figure to work. It tells you what a change is worth. It tells you which change to make. And now and then it tells you your week is going somewhere other than email.