Pretty much all been said above, but here’s my version.
For something like a flying wedge-tailed eagle, Location Accuracy (LA) can be 5km. I try to guess where it is, not where my camera is. For the nest, easy to find again, it could be 50m. For specific small plants, maybe a 20cm high orchid buried in long grass, a GPS won’t provide enough LA, so say 5 or 10m, but mark with a flag, or carefully describe landmarks, eg 2.3m @ 295 degrees from a big rock, 4.7m @ 170 deg from a broken tree. And describe it - ‘dog vomit slime mold on south side of dead tree’
Liberal but careful use of notebook, compass, tape measure. Even a cheap laser range finder.
Other than that, I have a camera without GPS, so I use a phone app (or a real GPS) to record a track, then produce a .gpx file.
Then I use GPSprune, GPScorelate or similar. Google: ‘Geotagging Software’ They work by lining up timestamps between the GPS track and the photo to work out the location of the photo, and then add geolocation data hidden inside my .jpg’s. So you do have to carefully align the time of your camera with the GPS, and a forgotten Daylight Saving change is annoying (but can be dealt with)
You’ll have to research and adapt from the above outline, for your own software, image format, etc. eg lots of image formats won’t hold exif data, but might work some other way.
Useful for indexing, GPSprune (and others?) can produce a .kmz with thumbnails, so you can stick that on Google Earth to visually keep track of where you were.
Like everything, including using iNat, there’s a learning curve, but after the 3rd time it seems easy.