What shows from outside with no insights at all
The first giveaway is the ratio of reactions to follower count. A living account with ten thousand followers usually collects between two hundred and eight hundred likes per post. An inflated one at the same ten thousand often sits at thirty.
The second gap sits between likes and comments. Bots supply likes cheaply, while a comment needs text. So an inflated profile can pair a hundred likes with a single comment, and that one from the author.
The third sign lives in the growth curve. A living audience grows unevenly: a good post spikes, then plateaus. A purchased one arrives in steps of several thousand a day and then sits still for weeks.
Why engagement rate alone settles little
Engagement rate, the ratio of reactions to audience, gets quoted as a verdict. It helps, but it is easy to fake: buy likes in the same proportion and the number returns to normal.
Looking at who reacts is sturdier. Fifty likes from accounts with no avatar, names made of letters and digits, and zero posts say more than any percentage.
What a bot account itself looks like
Mass bots get made in batches, and a batch shows through its sameness. Accounts from one batch share a registration year, a name pattern and a following count.
Single signs prove nothing on their own. A living person can also lack an avatar and posts. The verdict comes from several signs landing together in the same group of followers.
- A name made of letters with digits at the end
- An empty feed or two or three random pictures
- Hundreds of follows against a dozen followers
- A private profile with no avatar
- A whole group registered in the same month
Abandoned real accounts form their own category. Somebody made a profile, followed fifty creators and stopped opening the app. That is not a bot, and it delivers exactly as much reach as one.
What bought followers do to an account afterwards
Instagram shows a post to a small slice of the audience first and watches the response. If the people who opened it scrolled past, distribution stops there. Bots do not open and do not watch through, so every new post starts on a weak signal.
The effect runs backwards: the more you buy, the smaller the share that reacts, and the worse the platform rates the material. An account with fifty thousand dead followers gets fewer impressions than one with three thousand living ones.
A purge helps, though not instantly. Bots leave in batches, the counter drops, and for a while the profile looks worse than before. Returns come back once the reacting share climbs again.
When an advertiser checks your audience
An advertiser looks at the same gaps described above, and needs no access to your insights for it. Everything listed here shows from outside.
An account that never bought anything still benefits from knowing its own numbers in advance. Dead weight accumulates on its own: people leave the app, profiles get deleted, old followers stop visiting.
A sensible habit is checking composition once a quarter and keeping the reacting share at hand. It answers the advertiser’s question before they ask it.
Quarterly also matches how fast the picture shifts. Over a week almost nothing changes; over three months a giveaway, a viral post or a long silence each leave a visible mark on who still opens your posts.
Keep the previous quarter’s numbers beside the new ones. A single reading says little on its own, while two readings show direction, and direction is what an advertiser actually pays for.
Where dead audience comes from on its own
An account that never bought anything still collects ballast. Some followers stop opening the app, some delete their profile, some move to other platforms and return twice a year.
Ten to fifteen percent inactive over a year counts as normal. Anything above that usually means the audience was once gathered through mass follow-for-follow or giveaways.
Giveaways produce their own kind of ballast. Somebody follows for a prize, the prize gets drawn, and they stay on the list without opening a single post. Alive on paper, identical to a bot in what they deliver.
Follow-for-follow and how it ends
The follow-me-and-I-follow-you scheme builds a counter fast. Within a month half the participants unfollow, and the rest never read the feed, because they followed out of politeness.
The outcome matches a purchased audience: new posts start on a weak signal, because nobody is there to open them.
What to ask a creator before paying for an ad
An insights screenshot is easy to fake, so asking for one alone achieves little. Asking for things that are hard to fake on the spot works better.
- Reach for the last five posts in a row, not a hand-picked set
- Viewer counts on the first story across several ordinary days
- Audience geography, if the product sells in one city
- A screenshot with the date visible, not cropped out
Any gap between what shows from outside and what the screenshot claims is itself an answer. An account collecting thirty likes per post does not deliver twenty thousand reach, whatever screenshot arrives.
Check before paying rather than after. The platform offers no refund mechanism for creator ads, and talking it out afterwards rarely works.
How to clear an audience and what follows
Removing a follower by hand works: open the list, tap the three dots beside a name, choose remove. The person gets no notification, and your own following list stays untouched.
Bulk cleaners promise to do this in batches. Paying for that access means handing a password to an outside service, and the platform treats automation as direct grounds for blocking an account.
Cleaning by hand in portions works better. A hundred removals a day passes without trouble; a thousand in an hour reads as software and brings temporary limits.
What the numbers do in the first weeks
The follower counter drops immediately while reach lags behind. The platform needs several posts to rebuild its read on an account with a new audience composition.
In the first days after a purge the engagement rate looks suspiciously good: the audience shrank while reactions held. Drawing conclusions from that number is premature.
A steady picture appears after roughly three weeks and five or six posts. By then either reach grew on a smaller audience, or it turned out the bots were never the problem.
When cleaning will not help
If few people react and no bots are present, the cause sits in the posts rather than the follower list. Removing a hundred dead accounts changes nothing there.
Checking is easy: work out the reacting share among followers who look alive. If that number is low too, there is nothing to clean.
What counts as normal
No single norm exists: a cooking account and a legal one deliver differently at the same size. Keeping the order of magnitude in mind still helps separate a weak account from an inflated one.
A profile under a thousand followers usually shows the highest reacting share: acquaintances follow, and so do people who genuinely need the subject. Ten percent turns up often here.
After that the share falls almost always. At ten thousand, three to five percent counts as workable; at a hundred thousand, one to two. The fall alone signals nothing about bought followers, it follows from a broader audience caring less.
Where the fall turns suspicious
A low share is not what alarms; incompatibility with the other numbers is. An account at fifty thousand with thirty likes and two comments shows hundredths of a percent. That barely happens naturally.
The second warning sign is evenness. A living account collects unevenly: one post takes a hundred likes, another four hundred. If every post lands on thirty give or take two, those likes do not come from people.
The third is a mismatch of language and geography. A Russian-language account where half the reacting followers subscribe to content in other languages and write in them gathered an audience off-topic.