To identify fake followers, you do not need to inspect thousands of accounts. You need to know whether an audience behaves like an audience of its size: whether people watch, react and comment in proportion to the follower count. This guide gives you a 10-minute manual check for TikTok, Instagram and YouTube, the size-adjusted lines that separate normal from suspicious, two worked examples, and what Kolect's database of 716,539 creator accounts shows about how often each platform trips the main signal. It is written for brands and agencies vetting creators, and for creators who want to check their own account.
Which kind of fake-follower check do you need?
There are three ways to check an account, and they answer different questions. A public-signal estimate reads the numbers anyone can see on recent posts; a follower-sample audit classifies a sample of the follower accounts themselves; the creator's own analytics show the platform's first-party data. Screen a long list with the first, then confirm your shortlist with the second or third.
| Public-signal estimate | Follower-sample audit | Creator's own analytics | |
|---|---|---|---|
| What it reads | Followers, views, likes, comments, shares and following on recent posts | A sample of follower accounts, each sorted into real people, mass followers, suspicious accounts or bots7 | Reach, audience countries, age and follower activity from TikTok Studio, Instagram Insights or YouTube Studio |
| Platforms | TikTok, Instagram, YouTube (Kolect's checker1) | Modash's fake-follower data is Instagram only6; HypeAuditor describes its Audience Quality Score for Instagram influencers9 | All three |
| Cost and time | Free, a few minutes; Kolect's browser checker needs no sign-up1 and the full check is free 20 times a day5 | Modash's free Instagram check needs no sign-up7; full reports come with paid plans, e.g. Modash from $199/mo billed annually8 | Free, but the creator has to cooperate |
| Best for | Screening 50–500 candidates | An Instagram shortlist before a larger fee | The final yes before you sign |
| Blind spot | Does not classify individual followers; a low score means "look closer" | Vendor-defined models that cannot be compared with each other; strongest on Instagram | Screenshots can be cropped or edited, so ask for a live screen recording |
Our HypeAuditor alternatives page compares the audit tools in more depth. The rest of this guide is the public-signal method, which you can run on any platform today.
How to identify fake followers in 10 minutes
Run these seven steps in order. The first two take under four minutes and eliminate most problem accounts, so you only spend time on comments and follower lists for accounts that pass.
- Collect the numbers (3 minutes). Note followers and following. Open the last 12 posts, skipping pinned ones, and write down views, likes and comments (and shares on TikTok). Use the middle value, not the average: one viral post inflates an average and hides a dead audience.
- Divide median views by followers (30 seconds). This is views per follower, the most telling single number. Compare it with the line for the platform and account size in the next section.
- Divide engagement by views (1 minute). Add likes, comments and shares and divide by views. Far below the line means viewers do not react. Also check likes alone: in Kolect's check, likes above 25% of views is suspicious and above 40% is a red flag1. At that level, many likes are coming from accounts that never watched.
- Read 20 comments across three posts (2 minutes). Real audiences ask questions and react to specifics. Warning signs: emoji-only replies, the same praise under every post, languages the creator never uses, and the same handful of accounts commenting first on every post (an engagement pod).
- Open 10 followers (2 minutes). Accounts with no photo, no posts and random handles are likely bots. Accounts following thousands of others rarely see a creator's posts: HypeAuditor calls real accounts with over 1,500 followings "mass followers"9.
- Check growth history (1 minute). On a public tracker such as Social Blade, organic growth comes in steps after popular posts. A jump of thousands of followers overnight with no viral post behind it suggests purchased followers, and a sharp drop soon after often means the platform removed them.
- Do the platform-specific check (30 seconds). TikTok: total profile likes divided by followers; Kolect's check treats under 0.3× as a red flag1. Instagram: open the profile menu and "About this account" for the date joined and the number of username changes. YouTube: compare subscribers with long-form views, not Shorts.
Then enter the numbers in Kolect's free fake follower checker for a 0–100 score, and compare engagement with creators of the same size in the engagement rate calculator.
Which numbers are red flags?
A number is a red flag only relative to the account's size, because large accounts reach a smaller share of their followers. Kolect's check scales its views-per-follower lines by follower count: 1.5× the base line at 5K–50K followers, 0.75× at 200K–1M and 0.5× at 1M+1.
Views per follower: suspicious line / red-flag line by account size (Kolect fake-follower check1)
| Followers | TikTok | YouTube | |
|---|---|---|---|
| 5K–50K | 4.5% / 1.5%1 | 3% / 0.75%1 | 3% / 0.75%1 |
| 50K–200K | 3% / 1%1 | 2% / 0.5%1 | 2% / 0.5%1 |
| 200K–1M | 2.25% / 0.75%1 | 1.5% / 0.38%1 | 1.5% / 0.38%1 |
| 1M+ | 1.5% / 0.5%1 | 1% / 0.25%1 | 1% / 0.25%1 |
These lines sit far below a typical creator, so they catch outliers, not average performers. The median TikTok creator with 10K–100K followers gets views equal to 17.47% of followers, and the median with 100K–500K gets 10.89%3. An account under the red-flag line is reaching a fraction of what its peers reach.
The scoring Kolect uses is simple enough to apply by hand: start at 100, subtract 10 for each suspicious result and 25 for each red flag, and subtract 50 when views per follower fall below a fifth of the red-flag line; 80–100 is low risk, 55–79 medium and under 55 high1. The checker's page lists the remaining lines (engagement per view, following, followers per post) for every size band. It gives no verdict for accounts under 5,000 followers or with fewer than five posts that show views1.
Why "engagement under 1%" is the wrong test
The most repeated rule in fake-follower checklists, that engagement below 1% of followers signals fake followers, goes back years (one 2018 guide gives it as "sub 1%"15). Applied today it flags normal creators. In Kolect's creator database the median Instagram account engages 0.82% of its followers per post, falling to 0.63% for accounts with 100K–500K followers and 0.49% at 1M+; the median TikTok account engages 1.09%3. A flat 1% cutoff would fail more than half of the Instagram accounts in that sample.
Two fixes make engagement useful. Divide by views instead of followers, which measures whether the people who actually saw the post reacted: the medians are 7.83% on TikTok and 2.83% on Instagram3. And compare a creator only with creators on the same platform and of the same size.
Two worked examples
These two accounts are hypothetical, built to show how the lines and the score work together.
Example 1: a mid-sized TikTok creator. The inputs are the medians of the last 12 videos (views, likes, comments, shares) plus the profile's followers, following and total likes, shown in the calculation column.
| Check | Calculation | Line (50K–200K followers) | Result |
|---|---|---|---|
| Views per follower | 2,700 views ÷ 180,000 followers = 1.5% | Suspicious under 3%, red flag under 1%1 | Suspicious (−10) |
| Engagement per view | (290 likes + 14 comments + 6 shares) ÷ 2,700 = 11.5% | Suspicious under 1%1 | Pass |
| Likes per view | 290 ÷ 2,700 = 10.7% | Suspicious over 25%1 | Pass |
| Total likes per follower | 2.1M total likes ÷ 180,000 = 11.7× | Red flag under 0.3×1 | Pass |
| Following | 210 following ÷ 180,000 = 0.1% | Red flag at 7,000+ following and over 50%1 | Pass |
| Score | 100 − 10 | 80+ = low risk1 | 90, low risk |
Reading it: the people who watch react normally, so this is probably not a bot audience. The problem is reach: most followers no longer see the videos, which happens when growth came from one viral moment or the creator changed topic. Price the deal on the views the account actually gets, not on its follower count; the influencer rates guide shows what creators charge per 1,000 views.
Example 2: a mid-sized Instagram creator. The inputs are the medians of the last 12 Reels plus the profile's followers and following.
| Check | Calculation | Line (50K–200K followers) | Result |
|---|---|---|---|
| Views per follower | 900 views ÷ 85,000 followers = 1.06% | Suspicious under 2%, red flag under 0.5%1 | Suspicious (−10) |
| Likes per view | 520 likes ÷ 900 = 57.8% | Red flag over 40%1 | Red flag (−25) |
| Engagement per view | (520 likes + 9 comments) ÷ 900 = 58.8% | Suspicious under 0.8%1 | Pass (high, not low) |
| Following | 6,400 following ÷ 85,000 = 7.5% | Red flag at 7,000+ following and over 50%1 | Pass |
| Score | 100 − 10 − 25 | 55–79 = medium risk1 | 65, medium risk |
Reading it: more than half of the people who "viewed" also liked, which real audiences rarely do; likes are arriving from accounts that did not watch. With generic comments on top, this account fails the 10-minute check. Run a follower-sample audit or ask for a screen recording of Instagram Insights before going further.
How common are low-view accounts? Data from 716,539 creators
Low views per follower are common on TikTok and YouTube and rare on Instagram. Of the TikTok accounts with 5K+ followers in Kolect's database, 15.8% had median views below the suspicious line for their size, against 1.1% on Instagram and 11.7% on YouTube2.
Share of accounts below the suspicious views-per-follower line, by follower tier (Kolect creator database, refreshed Sep–Oct 20262)
| Followers | TikTok | YouTube | |
|---|---|---|---|
| 5K–10K | 8.4%2 | 0.5%2 | 9.4%2 |
| 10K–100K | 14.1%2 | 0.9%2 | 9.5%2 |
| 100K–500K | 17.4%2 | 1.6%2 | 12.4%2 |
| 500K–1M | 20.8%2 | 1.9%2 | 18.6%2 |
| 1M+ | 21.7%2 | 1.4%2 | 20.8%2 |
| All with 5K+ (n) | 15.8% (235,232)2 | 1.1% (327,893)2 | 11.7% (153,414)2 |
Below the stricter red-flag line the shares were 3.9% on TikTok, 0.1% on Instagram and 3.1% on YouTube2. Three things follow for vetting:
- On TikTok and YouTube, check views first, and check them hardest on big accounts. About one in five accounts with 500K+ followers falls below the line on both platforms2; a large following there is a weak signal on its own.
- On Instagram, views rarely catch a problem, so likes per view, comments and the follower list carry more weight.
- Low views are not the same as fake followers. A posting break, a topic change or a dead viral spike also lowers views, and the database holds creators indexed for brand work rather than a random sample. Treat these shares as how often the signal fires, not as a fake-follower rate.
How to check fake followers on Instagram, TikTok and YouTube
The method is the same everywhere; what differs is which numbers each platform shows and how it cleans up fakes.
- Use Reels for views: each Reel shows its view count, and Instagram rarely trips the views line, so give likes per view and comments extra weight.
- "About this account" in the profile menu shows the date joined and how many times the username changed; a brand-new account with a large following, or many name changes, deserves a closer look.
- Instagram has said since 2018 that it removes inauthentic likes, follows and comments created through third-party apps that boost popularity12, so a sudden follower drop can mean a past purchase was cleaned up.
- For a sample audit, Modash's free check classifies a sample of followers on any Instagram account without sign-up7.
TikTok
- The profile grid shows views on every video, so the median of the last 12 takes a minute.
- Compare total profile likes with followers, a number TikTok shows on every profile. Kolect's check expects total likes to at least equal the follower count (1×) and calls under 0.3× a red flag1. A large following that has left few likes on anything is hard to explain with real viewers.
- TikTok is where views per follower fails most often: 15.8% of accounts with 5K+ followers in Kolect's database fall below the suspicious line2. Ask TikTok creators for a screen recording of the Followers tab in TikTok Studio, which shows where followers are based.
YouTube
- Compare subscribers with the views of the last ten long-form videos, and judge Shorts separately.
- Mind the counting change: since 24 August 2026, YouTube counts a view the moment a video starts to play, across Shorts, long-form and live streams14. Compare a channel's recent videos with each other rather than with uploads from before that date.
- YouTube's fake engagement policy bans promoting services that inflate views, likes and subscribers, and says traffic found to be artificial is not counted13. A large subscriber count with a small, steady view count is the pattern to question.
What to do when an account fails
Decide by how many signals fail, not by any single one. One weak result usually has an innocent explanation; several together rarely do.
| Result | What it usually means | What to do |
|---|---|---|
| No flags, score 80+ | Audience behaves like its size | Proceed; still read comments for brand fit |
| One suspicious result | Stale followers, a topic change or one viral spike | Price on views, not followers; ask for analytics |
| One red flag, or two suspicious | Inflated likes, bought followers or a pod | Ask for a live screen recording of analytics; run a follower-sample audit on Instagram |
| Two or more red flags, or a severe views result | Most of the audience is not real or not watching | Drop the creator |
Protect the deal in writing as well as in vetting. In the US, the FTC's rule on reviews and testimonials prohibits selling or buying fake indicators of social media influence, such as followers or views generated by a bot or hijacked account, where the buyer knew or should have known they were fake and they misrepresent the buyer's influence for a commercial purpose10; it has been in effect since 21 October 202411. The FTC's Q&A also says a business is not liable merely for hiring influencers who happen to have fake followers11. The rule will not win back a budget spent on a fake audience, so vetting and the contract have to: ask creators to confirm in writing that they have not bought followers or engagement, and tie part of the payment to the post's delivered views.
After the post: check that the views arrived
The last check comes after publishing: ask for the post's insights after a week and compare the views with what similar posts deliver. Across 1,049 Kolect-run partnerships with published sponsored posts, the median partnership's views in roughly its first month equalled 28.2% of the creator's followers on Instagram, 26.1% on TikTok and 7.6% on YouTube4. A sponsored post far below that for its platform, from a creator who passed vetting, is a reason to ask what happened before the next payment.
How to find fake followers on your own account
Creators collect fake followers without buying any, because bot accounts follow popular profiles on their own. Modash's help center describes 10–20% fake followers as typical for creators under 50K and 20–30% for larger ones, by its own Instagram model6. To clean up and to answer brands' questions:
- Run your numbers through the fake follower checker to see what a brand will see.
- Scroll your followers list and remove obvious bots (no photo, no posts, random handles); Instagram and TikTok both let you remove a follower.
- Never pay for followers, views or likes: the platforms remove them, and they lower the ratios brands check first.
- Keep a short screen recording of your analytics (audience countries, reach of recent posts) ready to send with your media kit.
Check a whole list at once
The manual method suits a handful of creators; for a long list, a tool does the arithmetic. Kolect's full fake-follower check reads recent posts and runs all eight checks, free for 20 creators a day and 5 credits each after that, with 90 credits on sign-up5. Every creator in Kolect's search results also carries a 0–100 authenticity estimate. To build the list in the first place, see how to find influencers, browse ranked top TikTok influencers by country, or read more guides on finding and vetting creators.
Methodology
Thresholds and scoring are those of Kolect's fake-follower check as published on its tool page in October 2026. Kolect's figures come from three first-party datasets: views-per-follower shares from creator accounts with 5,000+ followers in Kolect's operating creator database, refreshed September–October 2026 (rule-based estimates from public metrics, not platform data, and one signal of eight); engagement medians from the same database; and sponsored-post delivery from 1,049 published partnerships in Kolect-run campaigns between August 2025 and October 2026. No brand or creator is identified. Full definitions, windows and sample sizes are in the sources below. The two worked examples are hypothetical.