Where small businesses actually lose time

Small business owners spend roughly a third of the working week on administration — about 16 hours a week, or around 70 hours a month. It rarely disappears in one big block. It leaks away in small, repeated tasks that each feel too minor to fix, which is exactly why they survive for years.

How much time are we actually talking about?

The numbers are consistent enough across surveys of small business owners to be worth taking seriously:

• Roughly a third of the working week goes to admin — about 16 hours a week, or 70 hours a month.
• The average owner works about 49 hours a week and says they should be working about 42.

Put those together and the picture is uncomfortable. The overwork is roughly seven hours a week. The admin burden is more than double that. The owner is not working too hard because the business is too busy — they are working too hard because a large share of their week is spent on work that does not need them.

Why doesn't it feel like 16 hours?

Because it never arrives as a 16-hour block. It arrives as four minutes here, twenty minutes there. Nobody experiences that as a problem worth solving. You experience it as "the day got away from me."

The tasks that make it up are almost always the same shapes:

• Moving the same information between two systems that don't talk to each other
• Filing, renaming and finding documents
• Writing the same kind of message for the hundredth time — quotes, follow-ups, confirmations
• Chasing people who owe you money or an answer
• Assembling a report by copying numbers out of somewhere else
• Re-reading things to check for the one detail that matters

What they have in common: they are recurring, they are structured, and they do not require real human judgement. That combination is the single most reliable indicator that work can be removed rather than merely endured.

The arithmetic of one repeated task

Take one ordinary example. A quote takes 25 minutes to put together. You do six a week.

One task. Not a transformation programme, not a new system for the whole business — one task, handled properly, gives back the equivalent of about two and a half working weeks a year.

Now notice the trap: at the scale of a single instance, the saving is 20 minutes. Nobody rearranges their business for 20 minutes. The value only becomes visible when you multiply by frequency, and almost nobody does that multiplication in their head. This is the main reason obviously-fixable work survives for years.

Acceleration vs elimination — the distinction that decides the size of the number

This is the most important idea in this article, and it is where most AI spending goes wrong.

Acceleration means a person still does the task, but faster, with an assistant helping. Research from the Federal Reserve on generative AI adoption found time savings in the region of 2.2 hours per week for users. That is real and worth having. But the person is still in the loop, still switching context, still spending attention on the task.

Elimination means the task no longer arrives at a human at all. The invoice is read, renamed and filed without anyone opening it. The follow-up goes out without anyone remembering to send it. A person reviews exceptions, not every instance.

Both are legitimate. But they produce different-sized numbers, and the difference compounds. Acceleration makes a good week slightly better. Elimination removes a category of work permanently — and it keeps paying every month without anyone maintaining the habit.

The failure I see most often: a business buys AI tool subscriptions for the whole team, everyone uses them enthusiastically for a fortnight, and nothing structural changes. They bought acceleration and expected elimination.

The cost nobody counts: switching

The stopwatch figure understates the real cost, because a four-minute task rarely costs four minutes. It interrupts something else. You stop, change context, do the small thing, then spend a while getting back to where you were.

This is why owners routinely report that admin "took the whole morning" when the tasks themselves would add up to well under an hour. The tasks are not the expensive part — the fragmentation is. A morning broken into nine pieces produces far less than an unbroken one, and no time-tracking exercise captures that.

It also explains a common frustration: people delegate the small tasks, feel the relief, and are surprised by how much better the week gets relative to the hours actually saved. They did not just buy back the minutes. They bought back the continuity.

Where should I look first in my own business?

You do not need a consultant to do the first pass. For one week, keep a note every time you do something that makes you think "again?". Then sort the list by two questions:

1. How often does this happen? Daily beats weekly. Weekly beats monthly.
2. Does it need real judgement, or does it just need doing? Be honest. A lot of work that feels like judgement is actually a rule you have never written down.

The things that are frequent and low-judgement are your candidates. Start with exactly one of them — the highest-frequency one you can describe in a few sentences. Businesses that try to fix everything at once usually finish nothing.

What does "40+ hours a month" actually look like?

Phantom Ops targets 40 or more hours a month returned to a typical client. Against a roughly 70-hour monthly admin burden, that is a bit over half of it — and it is not achieved by one clever trick. It is usually three or four eliminated tasks stacked together, each individually unremarkable.

That is deliberately unglamorous. The interesting part is not the technology. It is that the same arithmetic, applied to four ordinary tasks instead of one, is the difference between a 49-hour week and a 42-hour one.

Teeanu "Teejay" Thomas

Founder of Phantom Ops (Pty) Ltd, an AI transformation partner in Johannesburg, South Africa. Registered POPIA Information Officer with South Africa's Information Regulator.