Counting the week before trying to fix it

An owner cannot hand away any part of the week until somebody has counted it, and the counting turns out to be the harder half of the job.

What can be set out is the arithmetic, and it starts with a number almost nobody has: where the owner's week actually goes.

There is no good public number for this, and that is worth saying out loud

Search for how many hours a small business owner works and you will find confident figures between 50 and 70 hours a week.

Trace them back and most lead to the same place: a company that sells software, insurance or payroll services polled a self selected panel of a few hundred owners, published the result, and the number travelled for a decade with the survey's methodology left behind.

Those numbers are not worth repeating, because the method behind them cannot be checked.

The serious public dataset on how people in the United States spend time is the American Time Use Survey, run by the Bureau of Labor Statistics (bls.gov/tus).

It is a large, government run diary study, and its 2025 results report that people employed full time worked an average of 8.1 hours on days they worked (news release).

What it does not do, and no public dataset appears to do, is isolate the owner of a small ecommerce brand and report what share of those hours were decisions only that person could have made.

So the honest position is that nobody can tell you where your week goes. You have to count it yourself, and the counting is cheap.

The only question that matters about an hour

Two hours can look identical on a calendar and be entirely different work. The useful distinction is not urgent against important, and it is not billable against unbillable. It is this:

Did this need the owner, or did it merely reach the owner?

Work needs the owner when the owner is the only person holding the authority, the accountability or the information required to finish it. Signing a supplier agreement needs the owner. Deciding to discount the flagship product needs the owner. Choosing to open a new channel needs the owner.

Work merely reached the owner when it arrived because the owner is the default address. A customer emailed the address on the contact page. The 3PL replied to the last person in the thread. A listing was wrong and the owner happened to open the page. None of that required the owner's judgement.

It required somebody, and the owner was standing there.

Almost every founder underestimates the second category, because it does not feel like waste while it is happening. It feels like being useful.

Four categories, and only one of them is owner work

When you count a week, put every block into one of four buckets.

Decide. A judgement was made that changed what the business will do. Price, scope, spend, hire, direction, risk.

Do. Production work with your own hands. Building the campaign, fixing the collection page, writing the email.

Check. Looking at something to confirm it is still fine. Dashboards, pacing, inbox sweeps, opening the store to see if it looks right.

Receive. Being the place things land. Reading, acknowledging, forwarding, answering questions somebody else could have answered from a written rule.

Most owners who count honestly find a similar shape: decide is the smallest column, usually a few hours. Receive is the largest and it appears nowhere on any calendar, which is why it survives year after year without ever being examined.

Why everything routes to the owner by default

This is a mechanism, not a personality trait, and that matters because mechanisms can be changed and personalities mostly cannot.

Routing a piece of work requires a rule. Where a written rule exists, the item goes where the rule sends it.

Where none exists, the item routes to whoever holds the most context, and in a small company that person is the founder by construction, because the founder built the thing and therefore knows why every decision was originally made.

So every missing written rule is a small permanent tax on the founder's attention, charged every time that situation recurs. A refund threshold that has never been written down is not one decision. It is one decision per week, forever, until somebody writes it down.

This is also why handovers have to be documents. A conversation transfers the answer to one instance. A written rule transfers the routing.

What the interruption research actually says

The number everybody quotes is that it takes about 23 minutes to return to a task after an interruption. It traces to work by Gloria Mark and colleagues at the University of California, Irvine, and it is cited far more loosely than the research supports.

The most cited paper, The Cost of Interrupted Work: More Speed and Stress (CHI 2008, ACM Digital Library), found something more interesting than a stopwatch figure. People who were interrupted completed their tasks in less time than people who were not, with no measurable drop in quality.

What they reported was more stress, more frustration, higher time pressure and more effort.

Read that carefully, because it cuts against the usual advice. In that experiment, interruption did not make the work slower. It made the person absorb the cost.

That is precisely the failure mode of an owner who is the default address for everything. The work still gets done. The orders still ship. Nothing in any report goes red. The price is paid somewhere that no report measures, which is exactly why the situation can continue for years.

Long hours, with the caveats attached

Two pieces of real evidence, both with limits worth stating plainly.

The first is the joint WHO and ILO estimate published in 2021: 745,000 deaths in 2016 from stroke and ischaemic heart disease attributable to long working hours, with working 55 hours or more per week associated with roughly a 35 percent higher risk of stroke and a 17 percent higher risk of dying from ischaemic heart disease, compared with 35 to 40 hours (WHO).

The caveat is real and the authors state it: this is pooled observational evidence across many studies and countries, so it establishes association at population scale, not what will happen to any one person.

The second is John Pencavel's study of the relationship between hours and output (IZA discussion paper 8129, published in the Economic Journal). Below a threshold, output rises roughly in proportion to hours. Above it, output rises at a decreasing rate.

The caveat here is bigger than most people who quote it admit: the data come from British munitions workers during the First World War doing repetitive manual production, in small samples, a century ago.

It does not transfer cleanly to a founder answering supplier emails, and anyone using it as proof that founders should work forty hours is overreaching.

What both sources genuinely support is narrow and still useful. Hours are not linear in output, and beyond some point the additional hour is bought at a price that is paid somewhere other than the profit and loss.

Counting a week without installing anything

Do not buy a time tracking tool. The tool becomes the project and the project replaces the counting.

Use paper or one note file, three columns: the time, what it was, and which of the four categories it belongs to. Write entries of five words. Do it for one normal week, not your worst week, because the worst week is not what you are trying to fix. At the end, total the four columns.

The result is usually uncomfortable in a specific way. It is not that the total is high. It is that decide is small and receive is large, and receive is entirely made of things that have written answers nobody has written.

If you want the short version, the Founder Score asks eleven questions about a normal week and returns the hours behind the answer and the order to fix them in. It takes two minutes, runs entirely in your browser, shows its result with no email, and sends nothing until you ask for the file.

It is not a substitute for counting a real week. It is a way to find out whether counting a real week is worth your time.

What to do this week

1. Print or open one page and label three columns: time, what, category. Keep it next to you for five working days. Five words per entry. Do not tidy it.

2. At the end of each day, mark every line D, X, C or R for decide, do, check or receive. Do it the same day, because you will not remember on Friday what Tuesday afternoon actually was.

3. Total the four columns on Friday. Write the four numbers down somewhere you will find them in three months.

4. Take the three largest items in the receive column and write the rule that would have routed them elsewhere. One sentence each, with a number in it where a number belongs. A refund limit. A reply window. A named person.

5. Pick the single item that appeared most often across the week and give it a written owner other than you, even if that owner is a written rule rather than a person. One item. Not five.

Next week you will have something almost no owner has: a measured week and one fewer default address.