Developing an Instagram tool - need some advice

Hellooo guys and gals,

As the title says i am currently developing an Instagram tool to help with managing accounts and i need some help.

This tool will mostly be deep and detailed analytics of you account/post/followers etc. and your competitions. With some features for scheduling posts and stories and a small follow/unfollow fearture(basicaly it will suggest how you should follow that has the best change to follow you back and interact with your account.

My question is how do you guys distinguish fake users from real users? Now this needs to be done automaticly, the app will get all the info for the account and it needs to determine if the acc is fake or real by lets say ammount of likes/follows/comments per hour in a certain time frame (if for example a acc is liking 20 post per hour, every hour for the entire day its probably fake i mean the person does needs at list 6-8 hours of sleep ).

If the account is public or you are following it i can get all that info, the exact time someone is following/liking posts and then sort that info however i want (since this is already available with the following option in insta).

How do you target people to have the best enlargement, people to like or follow you, visit you website etc? Since the tool will analyze your followers to see what hastags they follow, what posts they like with what hastags used.. for your account as well as for your competition. So that you will know how to make your posts/stories. What time to publish them that will have the most effect on likes/follows/visits etc.

I already have some ideas on how to make the algoritam but i would love to hear how you do this. My plan(this is a rough analogy) was to get the account activity for the last week or two and then look for patterns. Likes per hour per hastag for the liked post, example: the person is liking 6-7 posts in the fashion hastag every 2-3 hours posted by a kardashian or some similar account - that is a pattern.
All this data to be compared to the previus week/s, how many photos or stories do they post every day, what hashtags to they use if any. There are people that post a story or a photo with the hashtag GoodMorning every day, etc etc. I do plan on adding machine learning to help with everything but that is far away as i still only know the basics for this.

Once i have a working version i will share it here with the source code of the algoritam so that it can be tested and improved.

NOTE: I understand that some of you would not like to share all this, as it will give away your methods, but any general ideas that dont compromise your strategies are welcomed. Also this as i said it will be an analytics tool, pure white hat (as per instagrams terms), i will be using the official api from instagram developers, so i will not have access to sensitive info nor can i build something that comprimises their terms (They allow this kind of tools i checked). I will look over what you guys say with what i can acually use to make the algoritam.

I am building this is my free time as i do have work and collage stuff so develepment will be slow, please have some patience with me .

Regards,
Lazareski

TO THE MODS: If this is posted in the wrong selection please move it.

I only follow users who have posted at least once to their stories. I also have a long dictionary list of first and last names common in Eastern Europe and Asia that I filter out. This makes sure that I am only following people relevant to my niche.

first you need to definte what is a fake. A real user running a really offline business may be using automation, does that count as fake?
A spammy child account, but just be a funnel to a parent account, but acts normal, is that a fake account?

regensteiner said:

I only follow users who have posted at least once to their stories. I also have a long dictionary list of first and last names common in Eastern Europe and Asia that I filter out. This makes sure that I am only following people relevant to my niche.

Click to expand...

Thanks your your input, interesting idea using common names for regions.

rootjazz said:

first you need to definte what is a fake. A real user running a really offline business may be using automation, does that count as fake?
A spammy child account, but just be a funnel to a parent account, but acts normal, is that a fake account?

Click to expand...

Yes that also need to taken in account, part of the reason for this post is to determine what would be considerd a fake account i may phrased my queston wrong, english in not my native lang. But i plan on making this for brands to grow so business account dont really qualify as you need normal people interacting with you.

Accounts that have 50% or more of the likes coming from non-followers are 99.9% fake. Yes, a few here and there will have that much engagement because they went viral or hit explore page, but that is not the norm. So, 3 our of 100 are good, and you'll follow the other 97 fakes? No way.
I work with influencers and brands, so there are many tricks that I know that are being used and how brands are trying to catch on.