Three kinds of fake account, the check that works for each, what a fake-follower score measures, and what none of it can prove.
By Jules CarmauxMarketing tips
Quick answer
"Fake" covers three different problems, each needing a different check. A bot is the easiest: no posts, a machine-made username, generic comments. An impersonator shows in the account's history, not its content: when it was created, and how often the username has changed.
The hard one is a real creator with bought followers. The account is genuine and only the audience is not, so no visual check settles it. You need the engagement, the growth curve and the audience data together.
Three different things people mean by a fake account
Before checking anything, work out which problem you have. They look similar and behave nothing alike.
Engagement against audience size, then the audience data
The rest of this guide takes them in that order, because that is the order of difficulty.
The sixty-second check, if that is all you have
If you only have a minute, do these three, in this order.
Count the posts on the profile, open About This Account for the join date and the username-change count, then read five comments under the most recent post. That is the whole check, and it takes about as long as reading this paragraph.
An empty grid, a very recent join date and five interchangeable comments is a strong enough combination to treat the account as suspicious and move on. Anything short of that deserves the longer version below, and if money is about to change hands, it deserves the audience check further down as well.
Eleven signs of a bot account, in the order worth checking
Eleven signals, in the order they are worth your attention. A single one proves nothing; three or four together on the same account are enough to walk away.
1. No posts, or a handful. A bot exists to follow, comment or message. Posting costs effort and attracts attention from spam detection, so most do not bother. An empty grid on an account that follows thousands is the strongest single signal on this list.
2. A username assembled by a machine. `user34948`, `j0hn_d0e_123`, or a real-looking name with four digits stuck on the end, all of which come from a generator working through a list. Real people pick a name they can say out loud to someone in a bar.
3. Following thousands, followed by dozens. Bots follow in bulk hoping for follow-backs. Be careful with this one in reverse: a normal person following 2,000 accounts is not suspicious, and a bought-follower account often has a perfectly normal ratio.
4. Comments that fit any post. "Nice pic!", "Great content!", a fire emoji, on a funeral photo and a product launch alike. Bot comments are written once and sprayed everywhere.
5. Replies that arrive in seconds, at any hour. A person answers when they wake up, from one time zone, with gaps for work and sleep. Automated accounts can reply immediately at any hour, repeatedly, without the gaps a human account shows.
6. A profile with nothing in it. No bio, or a bio that is a wall of hashtags and a shortened link. No profile photo, or a stock portrait that appears on a hundred other accounts.
7. A recent join date with heavy activity. New accounts that already follow thousands and comment constantly were built for the purpose.
8. Posts that do not belong together. A fashion account that suddenly posts about a crypto platform, or a travel grid with three unrelated product promotions in a row. Resold accounts change subject overnight because the buyer wanted the followers, not the theme.
9. A profile photo that belongs to someone else. Stock portraits and stolen photos are standard. A reverse image search settles it in seconds.
10. Promotion in every post. Real accounts post about their life and occasionally sell something. An account where every third post is an offer, a link or a giveaway is either a shop or a farm.
11. Direct messages you did not ask for. An unsolicited DM with a link, an investment offer or a job that pays too well is the point of most bot accounts. The message is the product; the profile is the packaging.
None of these is conclusive alone. The account of a real person who joined last month and has not posted yet ticks two of them, and a small business that promotes constantly ticks another.
The fake account checklist
Copy this and work down it. The first five settle most accounts; the last five are for when money is involved.
Post history: how many, and whether they belong together
Join date and username changes, from About this account
Comment quality: read twenty, do not count them
Profile photo, reverse image searched
Follower-to-following ratio
Follower growth pattern: steps, or a vertical line
Engagement rate against audience size
Audience location against your target market
Branded post performance against normal posts
Audience Authenticity Score, dated, on every creator on the shortlist
Why there are so many fake accounts in the first place
Understanding the motive tells you which signals to trust, because each kind of fake is built differently.
Accounts built to be sold. This is the largest category and the reason the username-change count matters at all: an account is grown or bought, left to age so it stops looking new, then resold to someone who wants a head start, which is why it arrives with a genuine posting history attached to an audience that has nothing to do with its current owner.
Accounts built to inflate someone else. Created in batches to follow, like or comment on a target, they are cheap, empty and never meant to be looked at directly, which is exactly why the manual checks above catch them in seconds.
Accounts built to scam. Impersonation, fake shops, crypto and romance approaches all belong here, and they are the ones worth real caution, because the operator invests in looking real: stolen photos, a plausible bio, sometimes months of ordinary posting before the first message arrives.
Accounts built for anonymity. A real person who wants to watch without being seen. Not fraud, and they will fail several of the checks below, which is worth remembering before you accuse anyone.
Only the first two categories are what a brand is paying for by mistake. The last one is the reason a "fake account" verdict should never rest on an empty profile alone.
Is it an impersonator, or the real person?
An impersonator copies someone real, so the content looks right. The history does not.
Open About This Account, and know exactly what it gives you. Instagram's own documentation is precise about this, and most guides get it wrong. Logged in, on any account, you see two things: the date joined, and the number of times the username has been changed. Not the former usernames themselves, only the count. On *some* accounts you additionally see the country the account is based in, the public accounts it shares the most followers with, its active ads, and whether it is verified.
That is enough. An account claiming to be a well-known person, created three weeks ago, with four username changes behind it, has answered the question. A username changed four times is the single most useful field here: real people rename occasionally, resold and repurposed accounts do it repeatedly.
Check the ads, if the account runs any. Where About This Account exposes active ads, they open in the Meta Ad Library, which shows what an account is actually promoting. A profile posing as a personal account while running commercial ads has told you what it is.
Check for the real account. Search the name and look at what else comes back, because a copied account is almost always a copy of one that still exists, usually with more followers, older posts and a history the copy cannot fake.
Reverse image search the profile photo. Copies either lift photos from the original account or buy stock portraits, so a search that returns the same face on a dozen unrelated accounts settles the question without any further work.
Treat a verification badge as one signal, not proof. Since Meta Verified became a paid subscription, a blue tick means someone submitted a government ID and pays a monthly fee, which is a real identity check and still not the same claim as the older badge, which said the account was notable and confirmed. Read it as evidence a person exists, not that this account is the business it says it is.
Does the account have bought followers?
This is the one that costs money to get wrong. The account is real, the person is real, and the audience is padded. Nothing on the profile page looks wrong.
Run these four in order. Each says what it can, and each has a case where it is wrong.
1. Engagement rate. Divide likes and comments on recent posts by follower count. On Instagram, a healthy rate for a mid-sized account usually sits in low single digits. An account with 100,000 followers and 200 likes a post is the classic shape of bought reach.
*When it is wrong:* large accounts naturally have lower rates than small ones, and a creator posting daily will show lower per-post engagement than one posting weekly.
2. Comment quality. Read twenty of them. Real audiences argue, ask where something is from, tag a friend, misspell things. Bought engagement is short, generic and interchangeable.
*When it is wrong:* some genuine audiences comment in emoji and little else, especially on visual niches.
3. Follower growth pattern. Real accounts grow in steps: a video lands, a feature happens, growth resumes. Bought followers arrive as a vertical line on a chart.
*When it is wrong:* a genuine viral post also produces a vertical line. Look at whether engagement grew with the followers or stayed flat.
4. Audience location. A creator in France whose audience is 70 percent in one distant country is worth a question.
*When it is wrong:* this is the signal most often misread. Diaspora audiences, English-language creators, and anyone whose niche travels will show a foreign audience for entirely real reasons. A mismatch is a reason to ask, not a verdict. It may also simply mean the audience is real and wrong for your campaign, which is a different problem with the same answer: do not pay for it.
You can run all four by hand on one account in about ten minutes. On a shortlist of thirty, you will not, which is the entire reason scores exist.
Click Analytic, September 2026.
TikTok and YouTube need different reads from Instagram
The four signals hold, the thresholds do not.
TikTok distributes by video rather than by follower, so follower count predicts reach far more weakly than on Instagram. A creator with 40,000 followers and a video at two million views is normal, not suspicious. Judge the last ten videos rather than the follower number, and treat comment quality as the strongest available signal.
Click Analytic data, September 2026.
YouTube hides the audience almost entirely: no public follower list, and view counts that mix subscribers with recommendation traffic. The useful reads are the ratio of views to subscribers over the last ten uploads, and whether comments discuss the actual content. Bought subscribers show up as a subscriber count that the views never justify.
If you want the audience read rather than the manual one, the free checker covers all three platforms.
What a fake-follower score actually measures
A checker does not detect fakes the way a metal detector finds metal. It estimates, from signals, how much of an audience behaves like real people. Knowing how the estimate is built is what lets you argue with it.
Audience credibility is the base. Every profile carries a credibility read from 0 to 1. The fake-follower share is that read inverted: fake followers percent = (1 minus credibility) times 100. A credibility of 0.92 means roughly 8 percent of that audience looks fake.
Three more signals sit on top. How real the followers look, whether that audience actually reacts, and whether it is safe to attach a brand to. Those four combine into one Audience Authenticity Score from 0 to 100, so a beauty creator on Instagram and a gamer on YouTube can be compared on the same footing.
The scale is published, not implied. 85 and up is excellent: real, engaged, safe to back. 75 to 84 is strong, 65 to 74 is fair and worth reading the split before committing, and below 65 means a larger share of the audience looks fake or inactive. That is a flag to look closer before you spend, not an accusation.
Click Analytic, September 2026.
That snapshot is the whole point. Audience quality is not a fact about a creator, it is a reading taken on a date, and a figure quoted from a 2019 industry study tells you about a platform that no longer exists. Ours comes from 23.6 million accounts read at the end of December 2025, and it is the number we hold ourselves to when we say most audiences are less clean than their follower count suggests.
Read it before you judge a roster. Roughly one Instagram account in seven clears both bars, so a shortlist of twenty where every name looks perfect was filtered by someone, and you want to know how.
Reading audiences is what we do: Click Analytic indexes more than 400 million public creator profiles and scores every one of them on the same scale, which is where the numbers in this article come from.
You can run a single account through the free fake follower checker for Instagram, TikTok or YouTube without signing up, or put a whole shortlist through creator vetting when there are more than a handful of names to get through.
Comparing creators, which is what the check is actually for
Knowing an account is not fake is a low bar. The useful question comes next: of the eight creators on your shortlist, which one is worth the money?
Every account carries some fake followers, including the honest ones. Bot accounts do not only follow the people who bought them. They follow popular accounts at random to look plausible, which means the bigger and better a creator gets, the more junk arrives uninvited. Two creators can both read at 4 percent fake on the Audience Authenticity Score, one because bots found them and one because they bought early and stopped, and a single reading cannot tell them apart. This is why an absolute threshold is a poor tool and a comparison is a good one.
So compare, on the same reading, at the same time. Put the whole shortlist through the same audience read on the same day, and the differences between them tell you far more than any one number. That is what our creator database and the rankings we publish are built on: the same score, computed the same way, on every profile, so a creator in one niche can be held next to a creator in another without arguing about methodology.
Read the indirect signals too. A fake-follower share is one input, not the answer.
Reels views against follower count. Whether saves and shares move with likes. How the last ten posts did against the previous ten. Whether the audience sits in the country you ship to.
Those separate an audience that is real from an audience that is real *and yours*. Reading six of them at once is most of what experience means in this job.
Then open the account and watch the content. Metrics tell you the audience exists. Only the content tells you whether your product belongs in front of it.
Be fair to the creator here. You are one person; they are talking to a hundred thousand. Not liking a video says nothing about whether it works, because you are almost certainly not the audience it was made for. What matters is whether the people who *are* that audience keep watching, and the numbers above answer that better than your taste will.
Past campaigns are the best evidence there is. If a creator has run brand work before, look at how those posts performed against their normal ones. Genuine audiences turn up for sponsored content at a lower but recognisable rate. An account whose brand posts collapse to nothing has an audience that is either bought or bored, and both cost you the same.
What a padded audience costs the brand that pays for it
The money does not disappear all at once, which is why this gets missed.
A creator with a padded audience still delivers the posts, so the content gets made, the invoice gets paid and the report comes back full of impressions, and only the second half is missing: the clicks, the code redemptions and the sales that were supposed to follow. On a single gifted post that is a nuisance, but on a retainer it compounds quietly for months until the whole channel gets written off as "influencer marketing does not work for us", when the accurate version is that you bought reach which was never there.
There is a second cost that is easier to miss, because it does not look like a loss at all: a padded audience skews every benchmark you build afterwards, so if your first ten creators carried heavily inflated followings, the engagement rate you now think of as normal is wrong, and you will keep hiring against a number your own bad data invented.
The fix is unglamorous but it compounds in your favour: read the same audience number on every creator before the first payment and keep the readings somewhere you will find them again. A shortlist checked once is a decision, while a shortlist checked every time turns into a benchmark that belongs to you rather than to the industry averages everyone else quotes.
What none of these checks can prove
Every honest version of this guide has this section. Most do not.
Inactive is not bought. A follower who signed up in 2019 and never opened the app again is dead weight rather than fraud, and although both drag an audience score down in exactly the same way, only one of them is the creator's doing.
A foreign audience is not a fake audience. This is the most common false positive in the entire subject, and the reason a French creator with a largely Indian audience gets accused of buying followers when the honest explanation is usually that they post in English, or that their niche simply travels further than they do.
A spike is not proof. A giveaway, a feature in a much larger account and a post that travelled all draw the same vertical line on a growth chart as a bulk purchase does, and the way to tell them apart is to check whether engagement climbed with the follower count or stayed exactly where it was.
A score is an estimate with a date on it. It reflects the audience at the moment it was read, so an account audited in January and hired in June has not been audited.
Nobody outside the platform sees the ground truth. Instagram knows which accounts it has actioned under its inauthentic behaviour policy, and everyone else, us included, is inferring from public signals. Any tool that claims it can certify an account as clean is overselling what that data allows.
What to do about a fake account once you have found one
Be realistic about what each action achieves. Reporting works at the scale of the platform, not at the speed of your afternoon.
If a bot is messaging you. Report it from the account's own menu and pick the reason that matches, because impersonation runs on a different track from spam, then block it. Blocking does not remove the account, it removes it from your life, which is usually what you actually want. Do not reply first: a reply confirms a live human is reading, and live humans are worth targeting again.
If someone is impersonating you or your brand. This is the one case where reporting genuinely moves faster, because Instagram handles a report from the impersonated party differently from a third-party report. File it from your own logged-in account rather than asking your followers to mass-report, since a pile of identical third-party reports is weaker evidence than one first-party one. Screenshot the profile and its posts before you file, because a removed account takes its evidence with it, and tell your own audience it exists so the next person who gets a message knows.
Where to draw the line before you pay a creator
There is no universal threshold, so set yours against what the campaign is worth. This is our reading, not a benchmark.
What is at stake
A reasonable bar
What to do at the edge
A gifted product, one post
Glance at the audience read
Proceed. The downside is a parcel
A few hundred dollars, one creator
Upper band, comments read like people
Ask for their own audience report and compare
A few thousand, several creators
The same bar on every name, read the same day
Drop the outliers rather than negotiating them
A retainer or ambassador deal
Audience read plus six months of growth
Do not sign on one reading. Ask again in a month
A regulated category
The bar above, plus brand-safety signals
If the audience data is unavailable, that is your answer
The second column barely moves down the list, and that is the point: the bar does not rise with the budget. What rises is how much verification the spend justifies before you accept the answer.
If you are being asked to pay a creator. Nothing needs reporting. Ask for their audience report from whatever tool they use, dated, and read it next to your own. A creator with a clean audience sends it without a fuss, and one who will not send it has told you something without saying it. Then set your threshold before you look at the next name, so the number decides rather than the follower count.
If the account is simply not for you. A real creator with a real audience in the wrong country, or the wrong niche, is not a fraud and does not need reporting or a lecture. It needs a no.
How common fake accounts actually are
Two numbers worth carrying, both current and both sourced.
Meta acts on fake accounts at a scale that dwarfs the visible problem. Its Community Standards Enforcement Report publishes the volume actioned on Facebook every quarter, running into the hundreds of millions, alongside Meta's own estimate of what share of monthly active accounts are fake, a figure it has put in the low single digits of its user base. Enforcement is constant, which is why the accounts you actually meet are the ones that got through this week rather than the ones that exist in total.
On the audience side, our own cohort of 23.6 million creator accounts above 10,000 followers, read at the end of December 2025, puts 14.6 percent of Instagram accounts in the group that combines good audience quality with at least 2 percent engagement. Read that the right way round: most accounts are not fraudulent, they are simply less clean than their follower count suggests, and the gap between those two statements is where most wasted budget lives.
Work out which kind of fake you are dealing with first. A bot shows itself through an empty grid, a machine-made username, generic comments and instant replies. An impersonator shows itself in About This Account: join date, country and former usernames. A real account with bought followers shows nothing on the profile at all, and needs engagement, growth and audience data together.
It is a decent signal for bots, which follow in bulk hoping for follow-backs, and a poor one for bought followers, where the ratio usually looks normal. Never treat it as proof on its own.
No. Only the platform sees who actioned what. Everything outside it, including our own score, estimates from public signals: audience credibility, engagement, growth and audience make-up. A good tool tells you how confident it is and when it may be wrong.
There is no universal line, because inactive accounts count against a score alongside bought ones. What matters is the comparison: read the same score across every creator on your shortlist, at the same time, and treat the gap between them as the signal rather than any single number.
Not any more. Meta Verified is a paid subscription with an identity check attached, so a badge means someone paid and submitted ID. Useful, but not the same as the account being who it claims to be in a commercial context.
The profile will look completely normal, because it is. Compare recent engagement against the follower count, read twenty comments for whether real people wrote them, look at whether the follower curve has a vertical step in it, and check where the audience actually is. One signal means nothing; three pointing the same way is your answer.
Bought followers can be blocked or removed manually, and the account owner can request removal in bulk through third-party tools, but the damage to the ratios stays visible for a while: the follower count drops while historical engagement does not change. From a brand's side this is not your problem to solve. Your job is to read the audience as it is today and price accordingly.
Report it as impersonation from your own logged-in account rather than asking others to report it, keep screenshots of the account and its posts, and tell your audience on your real account. Reports from the impersonated party are handled on a different track from third-party reports.
Try our Influencer Marketing Platform for free
Find Influencer email addresses and Contact Influencers with Click Analytic Influencer marketing software.
How To Find Authentic Influencers Among Your Followers
The benefits of including authentic influencers in brand campaigns are truly seismic! There are so many influencers, celebrities and loyal customers that can hide within your own audience who are truly passionate about your products. Unfortunately these authentic…
