How to Be More Productive When the Advice Is Made Up

Ask how to be more productive and you will be handed the same three numbers within a minute. It takes 23 minutes and 15 seconds to recover from an interruption. Multitasking costs you 40 percent of your productive time. The average attention span is now 47 seconds. Two of those three cannot be traced to a study, and the third comes from a book rather than a paper.

That matters more than it sounds. A field that repeats unverifiable figures for a decade is a field where nobody is checking, and the advice built on top of them is guesswork wearing a lab coat.

So here is what the research actually measured, and the small number of changes that survive contact with it.

The Most Quoted Number in Productivity Has No Source

The 23 minutes and 15 seconds figure is usually credited to Gloria Mark at the University of California, Irvine. Her papers are real and worth reading. The number is not in them. It appears in interviews, then in a newspaper, then in every productivity article since, and the 2008 paper it is attributed to does not contain the phrase anywhere.

What her earlier work did measure is more interesting. In a study of 24 information workers, people returned to an interrupted task the same day after an average of 25 minutes and 26 seconds, with two other pieces of work in between. The standard deviation was 54 minutes. That spread is the real finding, and it is the part that gets dropped.

The 40 percent multitasking figure has a similar history. It comes from a psychologist’s verbal estimate in an American Psychological Association article, and the same article puts the measured switching cost at a few tenths of a second per switch. One of those is a headline. The other is what the experiment found.

The Attention Span Number Is Half a Story

The 47 second figure is real research, and it is worth saying where it lives. It comes from Gloria Mark’s book rather than from a peer-reviewed paper, and two independent replications she cites landed at 50 and 44 seconds. The direction is solid. The decimal-point confidence around it is not.

Her published work gives a sharper picture anyway. Across 32 employees observed for five working days, people opened their email an average of 74 times a day, spending 32 seconds a visit, for a total of about 34 minutes.

Read those two numbers together. Half an hour of email spread across 74 separate entrances into it. The time was never the problem. The 74 doors were.

That reframes the usual advice. Batching email is not about the minutes it saves, because the minutes are already small. It is about how many times a day you re-enter a context and have to find your place again in the thing you left.

What Being Interrupted Actually Costs

Now the uncomfortable part. In the controlled experiment everyone cites, interrupted work was finished faster.

Forty-eight subjects worked on a task either undisturbed or interrupted. Undisturbed took an average of 22.8 minutes. Interrupted took 20.3 and 20.6 minutes depending on the type of interruption. People compensated by working quicker.

The cost showed up somewhere else entirely. Stress rose from 6.9 to 9.5 on a twenty-point scale. Frustration rose, time pressure rose, and effort rose. The interruption did not steal the minutes. It charged for them in a different currency, which is why a day full of interruptions leaves you flattened without leaving obvious gaps in your output.

This is why “protect your focus” as advice keeps failing. You are not defending your speed, which holds up fine. You are defending the thing that runs out by four in the afternoon.

The Load Is Real, Even If the Numbers Are Not

Microsoft has the one dataset here nobody else can match, because it sits inside the tools. Its 2025 Work Trend Index combined anonymised telemetry with a survey of 31,000 knowledge workers across 31 markets. The finding: an interruption from a meeting, an email or a chat every two minutes during core hours. That is 275 a day.

The same data put the average at 117 emails and 153 chat messages a day, with more than fifty arriving outside working hours. Fifty-seven percent of meetings are unscheduled calls with no invite behind them.

Eighty percent of the global workforce told the same survey they lack the time or energy to do their job. That is not a motivation problem and no morning routine touches it.

Where the Week Actually Goes

Two older surveys are worth knowing, with their age attached. In 2023 Asana found knowledge workers spending 58 percent of the day on coordination rather than on the skilled work they were hired for, including 2.8 hours a week in meetings they called unnecessary. A 2021 survey by APQC, a benchmarking nonprofit, put genuinely productive work at about 30 hours of a 40-hour week.

Treat both as indicative. Asana sells software that promises to fix the thing it measured, and self-reported time is generous to the person reporting it. The reason to cite them anyway is that they agree with the telemetry, which has no opinion and no product to sell.

How to Be More Productive, in Three Changes That Have Evidence

Start with the calendar, because it is the only lever that moves 275 interruptions. Individual habits operate on the margin. The interruption rate is set by how your team agrees to reach each other, and it is the one thing on this list that needs somebody else’s agreement to change. Ad hoc calls with no invite are the biggest single category, and a rule that says urgent means a call and everything else waits until eleven costs nothing to adopt.

Assume your time estimates are wrong by about two thirds. The canonical study on this asked students to predict when they would finish a thesis. They predicted 33.9 days and took 55.5. Only 29.7 percent finished by their own date. Asked for a worst-case estimate, they said 48.6 days, and reality beat even that half the time. Your planning error is structural, so correct it arithmetically: take the estimate and add two thirds, then plan the week against that number.

Cut the hours before you optimise them. The largest trial of a four-day week, published in Nature Human Behaviour, followed 2,896 employees across 141 organisations in six countries for six months. Burnout, job satisfaction, mental health and physical health all improved against control companies where nothing changed. The earlier UK trial of 61 organisations found sick days down 65 percent and resignations down 57 percent, with revenue up slightly across the 23 firms that shared figures.

Treat that revenue number carefully, since it comes from self-selected companies that volunteered their data. The wellbeing results are the solid part, and they point the same way: the compressed week did not reduce output, which means the extra day was never where the output was coming from.

Fewer Hours, Same Output

The finding underneath all three is the same one, and it is unglamorous. Most of what looks like a productivity problem is a volume problem. You are being reached too often, you plan as though you will not be, and you buy the difference with hours at the end of the day where your judgement is worst.

None of that is fixed by working harder inside the same structure. It is fixed by changing what reaches you and by being honest about how long things take, both of which are boring and both of which hold.

The same pattern shows up wherever people are exhausted by work that is technically going fine. It is worth reading this alongside what the stages of burnout actually look like, because the early ones are almost always described as a productivity problem by the person in them.

What Does Not Work, and Keeps Being Sold

Tools first. A subscription cannot reduce the number of people who need something from you, and buying one at the moment you feel overwhelmed is a purchase made in the worst possible state. The same test applies here as to any recurring cost: name the task it unblocks, in one sentence, before you pay. That is the whole of the argument in whether the top AI tier is worth it, and it generalises to every tool on the list.

Tracking everything. Measurement has a cost and the cost is attention, which is the resource in question.

Routines borrowed from someone whose constraints you do not share. The five in the morning story is a schedule that works for a person with no school run and control of their own calendar. Copying the wake-up time does not copy the conditions.

The Honest Summary

Knowing how to be more productive turns out to require less than the industry around it implies. Protect the first two hours, because that is where the hard thinking has to go. Add two thirds to every estimate. Reduce the hours and watch what actually breaks, which is usually less than you expect.

And be suspicious of anyone who hands you a number to three decimal places. Twenty-three minutes and fifteen seconds is precise enough to sound measured and it was never measured at all. The precision is the tell.