Multiple submissions, one fee, three times the work

A comparison is the most reliably underpriced job in this trade, and the reason is a framing error rather than a negotiating failure. The buyer sent one brief and received one deliverable, so it feels like one job to them - and unless your rate card says otherwise, it becomes one job to you too.

It is not one job. It is n assessments, plus a synthesis that only exists because there are n of them.

The actual work

Assessing three submissions and ranking them is not three thirds of your standard job. Each subject requires the full attention pass, because a comparison in which one of the three was skimmed is worthless and, more to the point, visibly worthless.

Then there is a fourth piece of work that has no equivalent in a single job. You have to hold all three in mind at once, decide what separates them, and produce a ranking you can justify in a way that survives being read by someone who disagrees with it. That synthesis is the hardest writing in the trade, because a comparison is the one deliverable where the buyer already has a hypothesis and will notice immediately if yours contradicts theirs without explaining why.

So the cost curve looks roughly like this.

Those multipliers are an illustration of shape rather than measured figures, and the shape is the point: the synthesis grows faster than the count, because the number of pairwise distinctions you have to be able to defend grows faster than the number of subjects. Your gut estimate will undershoot it too: in Buehler, Griffin and Ross (1994), only 29.7% of students finished a large task by their own best estimate.

The one thing that genuinely does not repeat is the setup - reading the brief, opening the tools, writing the delivery message. That is a single fixed cost across the whole job, and it is the entire justification for a base fee rather than a flat per-subject price.

Price it as base plus per subject

The structure that matches the cost is not complicated, and it is the same structure trades with variable-quantity work have used forever.

A base fee covering the brief, the setup, the synthesis and the delivery. A per-subject fee covering each assessment. A stated maximum, because there is a number above which the job stops being a comparison and starts being a project you should quote separately.

Concretely, if a single assessment sells at 100, a comparison priced at a base of 60 plus 80 per subject gives you 220 for two and 300 for three. That is cheaper per subject than buying three separate assessments, which is a real discount the buyer can see, and it is roughly three and a half times a single job, which is what the work actually costs you.

The buyer gets a genuine volume discount because your setup cost really is shared. You get paid for the synthesis, which under a flat "comparison of up to five" price you would be doing for free.

The general principle - that a rate card exists so no individual conversation needs to be negotiated - is worked through in the piece on building a rate card you can defend, and comparisons are the case that tests it hardest.

Say it before they commission, not after

The expensive version of this is discovering the subject count after the money has changed hands.

"Can you rate these" with four attachments, booked at your single-assessment price, is not a bad-faith move. It is someone who read your listing, saw a price, and did not encounter anything telling them that quantity affects it. The fix is upstream of the conversation.

Put the structure in the listing itself, in one line: comparisons priced as a base fee plus a per-subject fee, with the two numbers visible. A buyer who can compute their own total before messaging you arrives already agreeing to it, which is the only negotiation strategy that reliably works.

Confirm the count in the first reply, in the same sentence as the price, before starting. "Four subjects, so that is base plus four - total X, and I will have it back to you by Thursday" takes eight seconds and closes the entire category of dispute.

When a subject arrives mid-job that was not in the count, the answer is a price rather than a refusal, for the reasons set out in the account of what a follow-up job is worth. Adding a fifth to a four-subject comparison also invalidates part of the synthesis you have already written, which is worth saying out loud, because buyers genuinely do not know that.

Why buyers want them anyway

None of this is an argument against offering comparisons. They are one of the highest-value jobs available, and they exist because the buyer has a question that a single assessment structurally cannot answer.

A comparison is the format where a human is least replaceable. Automated scoring will happily produce a number for each of five submissions, and it produces those numbers independently, which is exactly the thing that makes the output useless for ranking - the data-side treatment of why independently produced scores do not compose into a ranking explains this better than a pricing post can. What buyers cannot get from the scoring tools they have already tried is a defended judgement about which one, and why, from someone who looked at all of them in the same sitting.

That is also what makes the deliverable good. The ranking is the product; the individual assessments are the evidence for it. A comparison that reads as several separate reviews stapled together has skipped the part that was actually bought, and it is the single most common way these jobs go wrong even when the pricing was right.

Buyers approach a multi-subject brief with expectations that are not always stated - what they intend to do with the ranking, and how much of a gap between subjects counts as meaningful - and the commissioning-side view of that is on Rate Penis's judges pages.

If you are setting a comparison tier up on a platform for the first time, Rate Cock's judge listings are where you can see how other people have structured the base-plus-subject split and what caps they have set. Look at the caps in particular: the number people stop at is usually four or five, and that is not arbitrary.