I would echo an answer often given for many types of questions: just do what seems to work for you.
I’ve developed a rhythm this summer with uploading and identifying (my own observations).
I make a nature outing where I take lots of photos.
When I get home, I start to sort them out. I’m a bit methodical with this process and it can take me from a few hours to a few days to do this. If I’m sure on the id, I make it during the upload. If I’m confident in some higher taxon id, I make that.
But if I don’t know at all or I’m not sure, I first check iNat’s computer vision suggestion. If it looks pretty good, I mostly go with the highest taxon suggestion. (family or genus rather than a species level). If I’m not sure, I will go pretty high.
I don’t do more than a few minutes of research at this point. I have too many photos to upload and I don’t want to be distracted from getting that done.
Once done with uploading, I drag the folder of photos marked by date and location into a pending file. Then I let the observations sit awhile. Usually, within a day or two, I have another batch I’m uploading. (and I do other things that naturalist stuff!)
After a week or so (longer in the middle of the summer), I find a moment to start reviewing all the observations in the oldest pending folder. If I trust the identifiers who got one to research grade, I’ll just consider it correct. For observations not at RG or with suggestions from people I don’t know (or may want to double check) I will start to research (and learn) how to determine down to species level (if possible). I might spend close to an hour on a particular observation using various internet sources* and field guides. If I think I can identify it to species, (with explanations of why), I do that. If not, I move on to the next. I learn a lot during this process.
For some reason, I like delaying my research. I think one uses different parts of one’s brain for different sorts of tasks. Being in and observing nature is one type of mindset. Sorting through dozens of photos trying to decide which is best and matching photos of the same organism from different cameras utilizes another type of mindset. And researching and picking out sometimes minute details from the photos to determine X vs Y species is yet another mindset. I can’t really switch from one to another quickly and seamlessly.
But my nearby friend who uses iNat has a completely different approach to identifying her observations. And you might find a different process works best for you.
*I have seen a few comments in various threads about mis-identifications on iNat potentially confusing people using iNat photos to help in identification. Although I do consult the computer vision to give me a direction to research, I don’t rely on it. And I really don’t do more than take a cursory look at the photos submitted by other iNat users for a species. I use them to either discard a species suggestion or consider it. But I don’t use the photos for help in id-ing. I have found too many mistakes. I go almost immediately to other sources I find more reliable and helpful.