In which context does Apple use this guideline, and why did they do so? I can only see the logic in the context of having an input field where you search. It just seems a bit odd since it has so many negative consequences, sure it reduces clutter and shortens the length of the form but I think that does not make up for the consequences.
Here is the source for your quote.
You can split data entries into two kinds:
- Fill and forget
- Fill and persist
Fill and Forget
In this scenario:
- The user lands upon some data entry interface
- The form is empty
- Once filled by the user there is an action to progress to the next "stage"
- The data entered is not persisted - it only enables progressing to the next "stage"
- Once submitted, the user cannot return to data entry interface (meaning there is no "previous" options, or no cyclic paths returning to that data entry in the user journey).
Gatekeepers are good examples for this data entries type. Examples:
- Login forms
- Captcha forms
- 2FA forms
- Contact request form
- Newsletter signup form
In these sort of cases, you could replace a label with a placeholder:
- The form is empty to begin with
- The place holder denotes the purpose of each field
- Forms are generally short so either the nature of each field is in the user's memory (eg, username), or there is little need to edit previously entered values.
- Once submitted there is no way back, so no issue with no labels or placeholders (due to the filled form).
Fill and Persist
Unlike the previous kind, once filled data is persisted and users can return to these data entry interfaces for editing purposes.
Examples are "delivery address" in a checkout wizard.
Here, it makes sense to put label or users will not be able to infer the nature of each field. (Users can infer from the values, but this is error-prone).