Per-job pay hides the hours from you

An employee knows they worked fifty hours because someone counted. Someone paid per job knows they did eleven assessments, which is a number about output and tells them nothing at all about time.

That missing measurement is the mechanism. Overwork in this trade is not usually a decision anyone made; it is a decision nobody was in a position to make, because the information was never collected.

Why the unit of pay hides the load

Per-job pricing converts every extra twenty minutes into a discount rather than into overtime. Take longer on a job and the fee does not move, so the cost of a hard week shows up as a lower effective hourly rate, which you only see if you were calculating one.

Three effects compound on top of that.

Time per job varies enormously by format and buyer, so a week of eleven jobs can be six hours or sixteen. The unbilled work - enquiries, waiting, admin, the buyer who asks four questions and books nothing - scales with enquiries rather than with orders, so a busy week with poor conversion is the most tiring kind and the worst paid. And demand arrives in the evenings, so the load lands on the hours that were already doing the recovering.

The result is a trade that feels light per job and heavy per month, with no single moment where it obviously became too much.

The early signs, before it is a problem

None of these are dramatic, which is why they get past people. The WHO's description of burn-out in ICD-11 names three dimensions: exhaustion, increased mental distance or cynicism about the job, and reduced professional efficacy. It classes burn-out as an occupational phenomenon rather than a medical condition, and anything persistent is a conversation for a doctor as well as a spreadsheet.

• Turnaround creeping. Jobs that used to go out early now go out on the deadline. This is the earliest reliable signal and it appears in your own delivery timestamps.
• The rubric getting thinner. Assessments shortening, axes covered more briefly, the specific observation replaced by a general one. Buyers notice this later than you do, and then it appears in reviews.
• Dread at the notification. A booking producing irritation rather than nothing is a straightforward capacity signal.
• Batching collapsing. Doing jobs one at a time as they arrive, because you no longer have the concentration to sit down to five.
• Revisions rising. Rushed work produces the requests that cost most, which makes the next week worse. This is the loop that turns a busy month into a bad quarter.

Two of those five are visible in numbers you already have, which is the argument for tracking anything at all.

Capacity as a business decision

The decision is not "how much can I stand". It is "what is the weekly job count at which my delivery stays good and my effective rate stays acceptable", and it is a number you can calculate rather than sense.

Take your median minutes per job for each format, add roughly a third for unbilled time, and divide the hours you actually want to work. The answer is your ceiling at today's prices.

Once it exists, everything downstream gets easier. A week at the ceiling with demand still arriving is evidence for a price rise rather than for more hours, because this work has no leverage in it - every job needs you personally, so growth past the ceiling is price and format, never volume. That constraint is the whole subject of what scaling actually means here, and it is the reason a capacity number is more useful in this trade than in most.

Setting the ceiling also converts declining work from a failure into an operating rule, which is the part that makes it psychologically possible.

Use your own records rather than your mood

The reason to time jobs is not curiosity. It is that the trend in minutes per job is the earliest quantitative warning available, and it moves before anything else does.

Three columns per job is enough: format, minutes, take-home. Reviewed monthly, on a date rather than when you feel worried, they answer questions your recollection cannot - whether this month was actually busier or merely worse, which format is quietly consuming the week, and whether your effective rate has been falling while your income held steady.

That last combination is the classic pre-burnout state and it is invisible without the numbers, because gross income going sideways feels like stability. The habit costs about a minute a job, and the case for keeping those records at all is mostly this one.

Tracking a number over time to see a trend that is invisible day to day is not a novel idea, and the general discipline of it is set out well on the measurement side - the same reasoning applies to your minutes as to anything else somebody logs weekly.

The structural fixes, in order of effect

Raise the price before adding hours. A ten percent rise costs no time at all and does the same thing as a longer week. Retire the worst format. One tier usually accounts for a disproportionate share of the hours; identifying it is arithmetic and the case for withdrawing it is usually stronger than it feels. Batch, and publish a turnaround rather than a response time. This buys back the fragmentation, which is the tiring part. Take a stated break rather than a silent one. A paused listing costs visibility; a late delivery costs a review, and the review outlives the fee.

One thing worth saying plainly. This work is bounded, and part-time forever is a perfectly good version of it - the ceiling is not a failure to grow, it is the shape of the trade, and pretending otherwise is where a lot of the exhaustion comes from. Stopping altogether is a legitimate outcome rather than a failure, and knowing when to stop treats it as the decision it is.

Two outside notes. Buyers are not the source of the pressure; the commissioning side generally expects a considered piece of work within a day or two, and the buyer-side account of how a job is placed is a useful check on the urgency you have imagined for them. And the fatigue of producing the same scored breakdown for the fiftieth time is real, which is why what a score is actually communicating is worth re-reading occasionally - the parts a buyer values are narrower than the parts you have been exhausting yourself over.

If the platform-side controls matter - pausing, capping concurrent orders, hiding a listing without losing it - Rate Cock's judges page is where its own are documented.