You cannot price what you have not timed

Every pricing decision in this trade runs off one number, which is your minutes per job. Almost nobody measures it before setting a price, and the guess is wrong in a consistent direction.

Run one whole job, not the interesting part

The practice run has to include everything a real job includes, because the parts people skip are the parts that take the time.

Reading the brief and deciding what is being asked for. Any clarifying message you would have sent. The assessment itself. Formatting it into whatever a buyer actually receives. The delivery message. The admin afterwards: logging the job, filing the record, the review request.

Time the whole sequence on a clock, in one sitting, and write down the total. Timing only the writing gives you a number that is comfortably wrong and will underprice you for months.

The guess is wrong in a known direction. In the study that named the planning fallacy, Buehler, Griffin and Ross (1994) found students predicted their theses would take 33.9 days and actually took 55.5, and fewer than a third finished by their own best estimate.

Use generic material you own or that nobody has a claim to. A real submission is not practice material, and practice material and where to get it covers the sourcing problem, which is less obvious than it sounds.

What the number does

It converts a price into an effective hourly rate, which is the only figure that lets you compare a written job against a recorded one, or this trade against anything else you could do with the evening.

And it tells you which side of the pricing risk you are on. Per-job pricing means every extra twenty minutes is a discount you did not agree to, and you cannot judge whether that risk is acceptable without knowing the baseline - hourly versus per-job pricing is the piece on who carries it.

Then take the honest total, multiply by what an hour of your attention is worth to you, and check the result against the visible market band before you write a rate card. Setting a rate card you can defend picks up from there.

Expect two corrections

The first run is slower than a real job for one reason and faster for another, and both need adjusting for.

It is slower because it is the first. Expect the figure to fall substantially over the first couple of dozen jobs, and expect it to keep drifting down slowly after that.

It is faster because nothing went wrong. Real jobs contain a thin brief that needs a question, a buyer who replies at midnight, a file that will not export. Add a margin for that rather than pretending your median job is your clean job.

The other omission is the time that never attaches to a job at all: enquiries that did not convert, messages, profile upkeep. That overhead is a real fraction of a working session and it is invisible in a per-job timing - the arithmetic of a part-time week is where it gets counted.

Re-run it when the work changes

A new format, a new rubric, a longer deliverable, a move from written to recorded: each of these resets the number, and pricing the new thing off the old timing is how a format ends up quietly unprofitable. Twice a year is a reasonable rhythm even when nothing has changed, because your own speed has.

Two things worth knowing while you are timing. Anything a model produces instantly has already had its price set by that fact, so the minutes worth measuring are the ones going into the part a model cannot do - what automated scoring reliably does maps where that boundary sits. Scoring conventions belong to Penis Rater's scores material and measurement conventions to Measure My Cock's method pages; both decide what your practice job contains, which is what makes it long or short.

For the platform-side account of what a job looks like when one arrives, Rate Cock's judges page describes the shape of the work. Time yourself against that shape rather than an imagined one.