What is worth taking off before you publish.
Not everything embedded in a photo is a risk, and not everything that is a risk can be removed by a metadata tool. Here is the honest split.
Location, first and by a distance
Coordinates are stored to five or six decimal places, which is a few metres — a doorway rather than a district. Because a phone writes them silently whenever location is enabled for the camera, the photos most likely to carry them are the ones taken where you live. If you do one thing before posting, do this one.
The thumbnail, if the photo was cropped
EXIF holds a small copy of the picture, generated at the moment of capture. Crop or retouch the photo afterwards and that copy can still show the original frame. It has happened often enough in public that it is worth treating as a rule: if you cropped something out deliberately, remove the metadata rather than trusting the crop.
Names, devices and software
Author, copyright and software fields are written by editors and often carry a real name or an account handle. Some vendor blocks include a device or lens serial number, which links separate photos to the same camera even when nothing else does. These matter less than location for most people, and more than location if you are trying not to be linked across posts.
What a metadata tool cannot do for you
Removing the EXIF segment from a JPEG does not remove XMP, IPTC or the colour profile — they sit in different segments and survive. Any claim that one operation strips everything from a JPEG is wrong. Beyond the file, the filename may be more revealing than any tag, and nothing in a metadata tool addresses what the picture actually shows. Clean the file, then look at it again as a stranger would.
Start with the location
Open the photo you are about to post. You will see the coordinates if they are there, and can download a copy without them — a JPEG keeps everything else, and other formats are cleaned by re-encoding. Your original is untouched either way.