The expensive option is doing work even when nobody buys it

A buyer looking at a single price has no way to judge it. A buyer looking at three prices judges the middle one against the other two, which means the number you sell most of is set partly by the number you sell least of.

That is the argument for a top tier, and it is also the argument for making the top tier real rather than a prop, because a prop gets bought eventually and then you have to deliver it.

What anchoring actually does here

Anchoring is the well-documented finding that a judgement about value drags towards whatever number was seen first or seen alongside. Tversky and Kahneman described it in 1974 in Science as "adjustment from an anchor", a heuristic that is usually effective but leads to "systematic and predictable errors". Price also colours the experience itself: in Plassmann and colleagues' 2008 study in PNAS, raising the stated price of a wine increased tasters' reported pleasantness.

The practical version for a rate card is narrower than the pop-psychology version. It does not mean a buyer will pay more than they intended. It means that a buyer with a budget already in mind will read your middle price as either reasonable or steep depending on what sits above it, and will do this in about two seconds, before reading any of the descriptions.

A card that runs 25 / 40 with nothing above makes 40 the expensive one. A card that runs 25 / 40 / 90 makes 40 the sensible one. The work is the same in both cases.

Constructing a top tier that is not a lie

The failure mode is a tier priced high with nothing behind it, described in vague superlatives. Buyers detect this, and the detection is not conscious - the tier just reads as unserious, and a card with an unserious tier in it makes the other tiers look negotiable.

A real top tier needs one of four things, and preferably two.

More output. Longer, more sections, more axes covered, a second document. A different format. A recorded version alongside the written one, which is the cheapest genuine upgrade you have because the assessment is already formed before you press record. A scarce input. A live slot, a same-day turnaround, a call. A structural difference. A comparison across several submissions rather than one, or a follow-up pass a fortnight later.

Notice what is not on that list: "more effort" and "premium quality". You cannot sell effort, because the buyer cannot verify it and would not want to pay for it if they could. Every workable top tier is a countable difference, and the way to test whether yours qualifies is to see if you can describe it in a table without adjectives.

That table is legible in five seconds and every row is a fact. A buyer who wants none of it has still learned that your middle row is not the ceiling.

Where to put the top price

Roughly two and a half to three times the middle tier is the range that does the anchoring work without reading as a joke. Below double, the tiers blur and the buyer treats it as a rounding decision. Beyond about four times, the top tier stops being a comparison point and starts being evidence that you are not serious about the other two, which is the opposite of what you wanted.

The middle tier should be the one you actually want to sell, priced where your rate card says it belongs. Anchoring does not license inflating it - the reasons a rate card holds up under questioning are all about the middle number being defensible on its own terms, and a defensible middle is what the top tier is protecting.

When somebody buys it

They will, and the first time is unnerving because you priced it expecting not to have to.

Two things follow. The first is that you have to deliver it exactly as specified, at the quality the price implies, and if you cannot then the tier was decorative after all and should come down. The second is more useful: a sold top tier is information. One sale in twenty at the top is roughly the expected shape. One in five means your middle tier is underpriced and the market has been telling you so quietly for a while, which is a better prompt for a first rate rise than any amount of introspection.

Sales at the top also come disproportionately from repeat buyers, who have already established that the work is worth what it costs and are now buying against a different question - not "is this person any good" but "what is the most of this I can get". That is the buyer worth designing a tier for.

What the top tier cannot fix

It cannot rescue a card whose middle price is wrong for the market band. Anchoring shifts a judgement at the margin; it does not overturn a buyer's sense of what this work costs, and a buyer who has commissioned before is carrying a fairly accurate band in their head - the buyer-side account of how people choose a judge reads like someone comparing risk, not someone being led by numbers.

It also cannot compensate for a floor that has moved underneath you. Automated scoring made generic assessment free, so tiers that differ only in how much generic output arrives are being compared against zero at the bottom; the technical account of what automated scoring produces is the clearest guide to which differences a buyer will still pay for. Where precision genuinely is the product, the conventions that make a measurement comparable are what a top tier would have to be built on, and being explicit about them is a countable difference rather than an adjective.

On a platform where tiers are structured for you, the constraint is what the listing format allows rather than what you can imagine - Rate Cock's judge listings show the shape currently available, and it is worth designing your card inside that shape rather than against it.

The uncomfortable implication of all this is that the tier you spend least time on is doing measurable work, and the tier you agonise over is mostly being read as "the normal one".