Reverse face search is not reverse image search
Reverse image search, the kind Google has offered for years, matches a file. You give it a picture and it looks for that picture, or something close enough to be a crop or a re-encode of it. If you never repost the same JPEG, it has very little to work with.
Reverse face search matches a face. You give it any photo of a person and it returns other photos of the same person that it has never seen alongside the first one: different camera, different year, different haircut, different clothes, taken by someone else entirely. That is a different capability, and it is the reason not reposting a file is not a defence.
How an engine builds the index
A face-search engine crawls public web pages the way any search engine does. On every image it finds, it runs a face detector. Each detected face is converted into an embedding — a list of numbers describing the geometry of that face in a way that stays roughly stable across lighting, angle and age. The embedding goes into the index along with the page it came from.
A search runs the same pipeline on the photo you upload, then looks for the nearest stored embeddings. Your name is not involved anywhere in this process. The index is geometric, which is why it can connect an anonymous account to a corporate bio page without either one naming you.
Why a professional headshot is the ideal input
Face detectors and embedding models work best on exactly the image a good headshot is: frontal, evenly lit, unobstructed, in focus, high resolution, one face, neutral expression. Those are the conditions the models were trained on and the conditions under which the resulting embedding is cleanest.
The consequence is uncomfortable. The photo you chose because it makes you look competent and approachable is, from an indexing perspective, the single most useful photo of you in existence. A poorly lit party photo at three-quarter angle is far weaker evidence.
The two things that change your exposure, and they are not the same thing
If a photo of you is already indexed, the only thing that removes it from results is a removal request to that engine. Opting out de-lists you from that engine; it does not delete the photo from the website hosting it.
If a photo is not yet published, cloaking it before you upload changes what the indexer can extract from that file. It has no effect on any copy already out there. Treat these as two separate jobs on two separate timelines, because substituting one for the other is the most common mistake people make here.