Method
How hard to press
Bone-pressed is a stop, not a force; the reading is repeatable when the ruler stops on bone every time, and it wanders when pressure varies.
"Press to the bone" sounds like an instruction to push hard. It is more precise than that: it is an instruction to push until the ruler stops, and the two are not the same thing.
A stop, not a force
Bone is rigid. Once the ruler's end has actually reached it, more force does not move the ruler any further - it just pushes harder against something that is not going to give. That is the property that makes bone-pressed measurement repeatable at all: bone is a fixed landmark, so a ruler that has genuinely reached it gives the same reading every time, regardless of exactly how many newtons of force got it there. The pressure only matters up to the point where it has done its job of getting the ruler through the compressible tissue in front of the bone. After that point, pressing harder changes nothing about the reading, because there is nothing left to compress.
Where people get it wrong
The failure mode is almost always stopping short, not overshooting. A light touch stops on the surface of the fat pad, not on bone, and the ruler reads long because it has not displaced the tissue it is supposed to press through. This is the single most common reason a self-measurement comes in higher than a clinical one taken on the same person - not dishonesty, just an instinctively gentle touch that never reaches the landmark the method requires.
Pressing to genuine discomfort is not required and is not the fix either. Once the ruler has stopped against bone, any additional force is doing nothing to the reading and only adds unnecessary discomfort. The target is moderate, firm, sustained pressure - enough to compress the tissue fully and feel the ruler stop against something solid, not enough to be straining against it. A person watching cannot judge this from the outside any better than you can from the inside, which is worth remembering if a partner's guess ever gets treated as a check on your own reading - Rate Penis is clear elsewhere that a human's impression of a measurement is not the measurement itself.
The self-check
Take three readings in the same session, resetting the ruler between each one rather than leaving it in place. If the three numbers sit within about half a centimetre of each other, the pressure is landing on the same point each time and the spread you see is closer to what the instrument and your reading precision can resolve. If they spread by more than that, pressure is the first thing to suspect, because it is the input in this method that is under direct voluntary control and therefore the input most likely to vary between attempts. A spread that shows up mostly on the low end - one or two readings noticeably shorter than the rest - usually means those particular attempts stopped on tissue rather than bone.
Consistency matters more here than any single reading being "correct". Even someone pressing slightly too lightly, every time, in the same way, produces a number with a knowable and roughly constant bias. It is pressing inconsistently - sometimes to bone, sometimes not - that turns three readings into three different measurements of three different things, which is the situation the median-of-three convention exists to catch, not fix. Consistent firm pressure can leave a faint temporary mark on the skin at the landmark, and what to do about pressure marks between repeat readings is worth knowing before a mark from the first reading throws off the second. None of this can be recovered after the fact from a photograph - a picture shows the surface at one instant, not how firmly anything pressed against bone underneath it - which is part of why AI Penis is explicit that an image model is inferring rather than measuring.
What this is not about
This is only the pressure question. The reason a fat pad sits in front of the bone at all, and why it varies so much between people, is a separate piece of anatomy covered in why the pubic fat pad is why two length numbers exist - useful background if the "why" behind the pressure rule is what you are after. And pressure is only one of several details that decide the reading alongside landmark, line, and angle; how to measure properly sets out the full method these individual details slot into.
Get the pressure right and it stops being a variable at all - which is the point. A method with one fewer thing that can silently differ between attempts is a method that produces a number worth comparing to anything else, including the figures reported in the published data. That is a different aim from a subjective rating, which weighs an image rather than a landmark - a score from something like Rate Cock is not trying to answer this question and does not need to press anywhere to answer the one it does ask, and a photo-scoring service like Penis Rater is answering a different question in the same way.