Retire the tier your own numbers argue against

Stop taking a format when its take-home per hour, on your own recorded timings, stays below the rest of your card, and withdraw it slowly enough that its repeat buyers move to a named replacement. Most rate cards carry one such tier, kept because nobody ever divided its take-home by its hours.

The number that decides it

Take-home per hour, per format, on your own recorded timings. Not price, not gross, not what the platform says the job was worth.

Take-home means after commission, processing and conversion. Hours mean everything the format consumes: reading the brief, chasing missing material, the extra round of questions a certain tier always generates, and the setup you pay again every time. Unpaid time is where effective rates collapse: Hara et al. (2018), analysing 3.8 million tasks by 2,676 Mechanical Turk workers, estimated a median of about $2 an hour, and the authors note the figure turns on how unpaid searching, rejected and unsubmitted work is counted.

Thirty jobs is enough to act on and ten is enough to notice. The comparison only means anything across your own formats.

The cheapest tier usually has the worst ratio, by more than the price difference explains, for the reasons in why the bottom tier attracts the hardest buyers.

Retire on the ratio, not the volume

Volume is not the test. A format that books constantly and pays two-thirds of your other work per hour is displacing the better work, because every slot it fills is a slot the good format needed.

Any two of these three signals are enough:

• Take-home per hour below your other formats and not improving with practice.
• A revision or clarification rate above your average, which is the same cost arriving late.
• The format you would drop first if you were ill, a surprisingly honest instrument.

The exception: a cheap fast tier that reliably converts buyers into a more expensive second job is doing acquisition, and is judged on what follows rather than on itself.

All of this rests on tracking take-home per format, which is three columns and a minute per job.

Withdrawing it without losing anyone

A format vanishing without notice reads to a repeat buyer as you having stopped working. The sequence that works:

Stop advertising it, keep honouring it. Remove it from the listing while still accepting it from people who know it exists. Existing buyers notice nothing.

Tell the repeat buyers directly, once, with a replacement named. One message: this is retired at the end of the month, the nearest thing is X, here is what it costs. Naming a replacement is the whole trick, because most of them wanted the outcome rather than the format.

Give a date and honour it. An indefinite wind-down never finishes.

Price the exception rather than refusing it. If someone insists, quote a rate that makes the hour worth spending. Some will pay, which tells you the tier was underpriced rather than unviable.

Check for that outcome first. Poor take-home per hour with steady demand is a pricing failure rather than a product failure, and a rise is a smaller move than a withdrawal - the timing is in when to raise a rate and by how much.

What to do with the freed hours

The hours a retired tier releases are the only slack most earners get.

Move toward the formats fewest people offer, because a format ten people supply is priced by ten people. The floor has already collapsed under anything a tool produces instantly, and what an automated score does in two seconds maps which tiers are structurally doomed rather than merely underpriced. The same goes for anything a buyer can self-serve; the packaged tools that made that easy are what your cheapest tier competes with. Recorded formats are the usual direction of travel, and the equipment question is smaller than people assume - the gear reference covers it without the upgrade spiral.

If a tier exists because a platform's order types push you toward it, that is worth checking rather than assuming: Rate Cock's judges page sets out which formats its listings support, which decides whether your tiers are choices or constraints.