Tools
What a questionnaire does to a photo-based score
Age, goals and other form fields can feed a score or can be pure decoration.
Guides on Tools: The whole category on one page: kinds, claims, and how to read them, Every presentation choice on a result page, and what it does, A taxonomy of rating tools by what they output, A method for judging any rating tool before trusting it
A form before the score has one of two jobs: it changes the number, or it only collects data. You can find out which by submitting the same photo repeatedly and changing one field at a time. If the score does not move, the question was decoration.
Three things a field can do
A tool that asks questions before scoring is doing one of three things with the answers.
Feeding the score. The answer enters the calculation, for example as a weight or a prior. This is rare in photo-based tools and common in the quiz kind, covered in quiz-style raters.
Shaping the prose. The score is untouched but the paragraph under it changes, because the answer is passed to a text generator. "Since you said your goal is X" is the signature. The paragraph is written to match the number, so the tone changes while the result does not.
Collecting. The answer is stored, perhaps for marketing or segmentation, and never touches the result. The field exists because it is useful to the operator.
Only the first is a self-reported input in the sense that matters for the score. The other two make the experience feel personalised, which is a presentation choice.
Why the difference matters
If a field moves the score, then two people with identical photos and different answers get different numbers. That is a legitimate design, but it means the score describes a submission plus a story you told. It also means you can change the result without changing the photo, which makes any comparison sensitive to how you filled in the form.
If a field does not move the score, it is harmless to the number, though you should wonder why it is asked. And if a field moves the prose but is presented as shaping the result, the tool is implying a rigour it does not have.
There is also a measurement problem. Self-reported answers are not checkable, and a tool that lets them move a score has built in a lever that rewards optimistic answers. That is a second route for generosity: not from the scale but from what users claim.
Model documentation has a name for the habit that would help here. Mitchell and colleagues (2019) proposed that models come with model cards that state the intended use and the evaluation procedures, so a user can see what the system was built for and what it was tested on. A tool that asks you questions should be able to say, in the same spirit, which of them it uses.
The test: one field at a time
The method is the one that works for every variable in this field: change one thing at a time.
- Fix everything else: the same file, uploaded the same way, no edits.
- Submit with a baseline set of answers and record the score, the breakdown and the prose.
- Change a single field, such as age bracket or goal, and submit again.
- Return to the baseline answers and submit a third time, to see whether the first result repeats.
- Repeat for each field you care about.
The third step matters more than it looks. If the baseline does not repeat, then a difference you saw in step 3 may be ordinary noise, not the field. Check the tool's spread first, with an identical-file test, so you know how big a change has to be before you believe it.
Reading the outcome
- No change in number or prose. Decoration. The field is likely collected for the operator.
- No change in number, change in prose. The input shapes the text. Treat the paragraph as a caption that has been personalised.
- Change in number beyond the noise. A real input. Now ask whether the direction makes sense, and whether the tool says it works this way.
- Change in number, no explanation anywhere. The tool is using an input it does not describe. That is a finding, and a transparency one: it belongs with the signals a tool should show.
Where a field changes the number, read the direction critically. A tool that scores you higher when you state a younger age, or a "serious" goal, is rewarding the claim. You cannot show it is wrong, but you can note that the answer was untestable and the result moved.
Where the data goes
Whatever a field does to the score, it also becomes data the tool holds. What is kept, for how long and for what purpose is a privacy question, and aipenis.com's privacy hub is where it is covered. If you are looking for what a tool does with form data after the fact, measuremycock.com's data hub is the property for handling and recording numbers, and the first steps for a newcomer sit on cockrate.com's starting hub. This post is only about whether the inputs move the result.
Rate Cock is one tool to run this on, and the claim is checkable: it reports six axes rather than a total, so when you change a field you can see whether any single axis moved, which is more informative than watching one number. That is the general advantage of a breakdown for this test. A total can hide a field that nudged one axis up and another down, and the axes cannot.