Founders, and the hours that were never the business.

A founder note, with the evidence I wanted before I wrote it. Most founders can feel that the week gets eaten by small jobs, but few have counted what those jobs cost beyond the minutes they take. This post goes through what the research says about interruptions, switching between tasks and tired decision making, including where a famous result has been challenged, and what that means for how you arrange a week in a store.

I have kept the claims to what the papers report. None of them was run on Shopify merchants, so treat each as evidence about people doing knowledge work, which is what running a store mostly is, and not as a measurement of your business.

Where a founder's week goes

American Express and Small Business Saturday UK surveyed 1,000 UK micro, small and medium sized business owners, in research reported by Startups Magazine in 2026. Owners said they spend an average of 11 hours a week on administrative or finance related tasks, about six working days a month. Time on sales and growth came to just over half of that. These are self reported estimates from a mixed set of sectors, so they are not a measure of a Shopify brand, but they match what most founders say when asked: the admin is larger than they planned for.

The admin in a store has a recognisable shape. It is product details that need updating, a helpdesk inbox that never quite empties, stock questions that arrive in the same words every week, and a catalogue whose tags and notes drift out of line. Each job is small. The trouble is how they arrive: one at a time, at unpredictable moments, and most of them need you to stop what you were doing.

• 11 hours

  a week on admin or finance tasks, as reported by UK business owners

  American Express and Small Business Saturday UK, via Startups Magazine, 2026

• 23 min 15 s

  average time to get back to an interrupted piece of work

  Gloria Mark, field study of 36 professionals, reported by Gallup

• 44%

  of interruptions in that study were caused by the person themselves

  Gloria Mark, reported by Gallup

What an interruption costs

Gloria Mark, a researcher in informatics at the University of California, Irvine, spent years watching how people work. In an interview published by Gallup she described a study in which she and her colleagues shadowed 36 professionals in the workplace for three days each. Her summary was that people do not have working days, they have working minutes that last all day. Workers switched between what she calls working spheres, meaning distinct pieces of work, on average every 10 minutes and 29 seconds.

Most interrupted work was picked up again the same day. In the study, 81.9% of interrupted work was resumed that day, and on average it was resumed after 23 minutes and 15 seconds. That is a gap between leaving a task and returning to it, and it includes the other work done in between. It does not mean that every interruption costs 23 minutes of lost time, a point that is often lost when the figure is quoted. What it does mean is that interrupted work often does not come back at once, and when it does, you have to pick up the thread again.

The same study found that people are about as likely to interrupt themselves as to be interrupted by someone else. In the Gallup write up, 44% of interruptions were self interruptions. That matters for a founder, because the inbox, the dashboard and the store admin pages are all one click away. The fix is not only to keep other people out of your day. It is also to reduce the number of reasons you have to go and check.

Mark and two colleagues, Daniela Gudith and Ulrich Klocke, ran a separate laboratory experiment on what interruptions do to the person doing the work. They had 48 participants play a human resources manager answering emails, in blocks of 12 emails, with interruptions arriving about every two minutes by phone or instant message in some blocks and none in others. The surprising result is that interrupted work was done faster, with no difference in quality as measured by errors and politeness.

The cost showed up somewhere else. People in the interrupted conditions reported higher workload, more stress, more frustration, more time pressure and more effort. The authors' reading is that people compensate for the time they know they will lose by working faster and writing less, and that this comes at a price. They add that it is possible people could cope over time or that the measures would only increase, because the experiment could not tell. It also used mostly university students in a lab, not owners in a real job, so use it as an indication of the mechanism, not as a measure of what happens in your shop.

Of the 48 participants, 81% were German university students with a mean age of 26, paid 12 euros each, and 29.2% had never used instant messaging, so the group looks little like a shop owner with a full inbox. The sample is one reason to read the result as a pointer. Put together, the two studies say something useful for a founder. Interruptions make work resume late and make it feel worse while it is done, even when the output looks fine. A week of small interruptions can be a tiring week with a full to do list and few of the long stretches you wanted.

Why unfinished work follows you

Sophie Leroy, a business school researcher at the University of Washington Bothell, studies what she calls attention residue. She describes it as the mental leftover from incomplete tasks. In her words, like having too many browser tabs open, unfinished tasks stay active in our minds. Her 2009 paper is the source of the idea, and the University of Washington magazine reports her view that multitasking does not exist: people switch back and forth very quickly, and that is the problem. The brain, she says, likes to have things closed, or in good standing, before moving to something else.

This explains a feeling founders know well. You answer a customer, glance at a pending catalogue edit, start on a supplier email, and by midday you have touched twelve things and finished three. Nothing was hard. The jobs were just left open, and each open job takes some attention away from the next.

The practical advice Leroy gives is plain. Work offline when you need to concentrate, take breaks between tasks and meetings, do demanding work early in the day when your mind is sharp, and write things down so your brain can let go of them. The magazine piece says to tackle demanding work before about 9:30 in the morning. That is her advice and is not a rule from a trial, so test it against your own week before you rearrange it.

The decision fatigue claim, and its critics

Many articles about founder time cite one study as proof that people make worse decisions as the day goes on. It deserves a closer look, because the first reading is stronger than the evidence supports.

In 2011 Shai Danziger, Jonathan Levav and Liora Avnaim Pesso reported in the Proceedings of the National Academy of Sciences on 1,112 parole rulings made over 50 days by eight Israeli judges across four prisons. The proportion of favourable rulings fell from about 65% at the start of each decision session to nearly zero before a food break, and returned to about 65% after it. Overall, 64.2% of requests were rejected. The authors read this as evidence that extraneous factors, such as tiredness or hunger, shape what judges decide. The judges averaged 22.5 years of experience, heard between 14 and 35 cases a day, and spent about six minutes on each deliberation, so this was skilled, repeated, rule bound decision making at speed, which is why it was read as relevant to anyone with a long list of small calls to make.

It was challenged in the same journal. Keren Weinshall Margel and John Shapard looked at 227 decisions across 12 hearing days and interviewed lawyers, a panel judge and prison service staff. They found that the order of cases was not random. The board tries to finish all the cases from one prison before a break and starts with a different prison afterwards, and prisoners without a lawyer usually come last. Unrepresented prisoners were about a third of cases and succeeded 15% of the time, against 35% for those with a lawyer. Lawyers also said they put their strongest cases first and kept their weakest for last.

Their conclusion was that the drop within a session is an artefact of how cases were ordered, not proof that judges tire. I am not claiming that the original authors are wrong, only that the finding is disputed and cannot be treated as settled. A founder should not plan a week around a result like that. The better lesson is narrower: if a decision is routine and rule bound, asking a tired person to make it forty times is a poor use of that person, whatever the effect size turns out to be.

What to protect and what to hand off

None of this says you should never be interrupted. Some interruptions are the business: a wholesale buyer asking a question, a customer with a damaged parcel. Mark's interview records that people call an interruption on the same topic as their current work an interaction, and a different topic a distraction. The aim is to know which kind you are letting in.

From the evidence above, three things look worth acting on, and each is cheap to try.

• Count before you change. Write down for one week what you did and how many times you switched. Your own count beats any headline figure, including the 11 hours above.
• Batch the checking. Since 44% of interruptions in Mark's study were self caused, set fixed times to open the helpdesk, the order screen and the catalogue, and keep them shut between.
• Finish what you start, or write it down. Leroy's advice on closing loops costs nothing: note where you stopped so the task does not keep running in the background.

Then decide which work you hand to someone or something else. A useful test is whether the job needs your judgement or only your typing. A new range, a supplier negotiation and a hard customer call need you. A product description that needs a small edit, a tag that is inconsistent and a stock question you have answered before many times need a decision from you at most, and the typing before that decision can be done by someone else.

That is the line I would draw for any helper, human or not. It should take the repeatable work in a form you can check in a minute, and hand the decision back to you. A helper that does the work and leaves you to find out later is just another source of worry, and a helper that asks you about everything is just another interruption.

Where the hours go instead

The point of getting hours back is not to fill them with more admin. Mark's finding that interrupted work gets done faster but feels worse is a reason to protect long stretches for the work that only you can do: the next collection, the partner you have been meaning to approach, the customer you want to call. Those jobs are slow, they are hard to resume after a break, and they are the reason you started the brand.

If you try only one thing this month, make it the count. Spend a week writing down each switch and what caused it, then sort the list into jobs that needed you and jobs that did not. The second list is the one to shrink.

Where BYOM fits

Kina reads your store and your helpdesk. When something needs changing, Kina drafts the change and brings it to you for approval, so you can run more of the business yourself.