How the fake follower check works
Bought followers rarely watch, like or comment. So the check looks for numbers that do not add up for an account of that size on that platform: too few views for the follower count, viewers who barely react, likes that outrun views, an account that follows everyone back. It is the rule set Kolect runs on app.kolect.ai: eight checks, lines that scale with follower count, and a score out of 100 where 100 means no warning signs.
| Result of a check | Points taken off |
|---|---|
| Normal | 0 |
| Suspicious | 10 |
| Red flag | 25 |
| Views per follower under a fifth of the red-flag line | 50 |
| Audience location mismatch (an AI estimate, so it weighs little) | 5 |
A score of 80–100 is low risk, 55–79 medium risk and below 55 high risk. There is no verdict for accounts under 5,000 followers, where buying followers makes little sense and the ratios swing too much, or without at least five posts with views.
The eight checks
| Check | What it catches | Normal line at 50K–200K followers | Platforms |
|---|---|---|---|
| Views per follower | Followers who never watch: the most common trace of bought followers | TikTok 3%, Instagram and YouTube 2% (red flag under 1% and 0.5%) | All three |
| Engagement per view | People who watched but barely reacted: bot views, unreal audiences | TikTok 1%, Instagram 0.8%, YouTube 0.5% | All three |
| Total likes per follower | A follower count that outgrew every like the account ever got | At least 1× (red flag under 0.3×) | TikTok |
| Following per follower | Follow-for-follow growth | At most 80% (red flag: 7,000+ following and over 50%) | TikTok, Instagram |
| Followers per post | Bought or recycled accounts | Suspicious under 30 posts with over 20,000 followers per post; red flag under 10 posts and over 50,000 | All three |
| Likes per view | Bought likes | At most 25% (red flag over 40%) | All three |
| View consistency | Single posts with bought views | At least 0.1 on a 0–1 scale | All three, full check |
| Audience location | An audience from countries the creator does not make content for | At least 10% of the audience in the creator's market (AI estimate) | All three, full check |
Engagement here is likes, comments and shares divided by views, taken per post. The free checker above runs the first six; view consistency and audience location need the creator's latest posts and audience data, so they run in the full check.
Thresholds by follower count
Bigger accounts reach a smaller share of their followers, so the two core lines move with size: ×1.5 for 5K–50K followers, ×1 for 50K–200K, ×0.75 for 200K–1M and ×0.5 above 1M. Below the first number a creator is suspicious; below the number in brackets it is a red flag.
| Check, by followers | 5K–50K | 50K–200K | 200K–1M | 1M+ |
|---|---|---|---|---|
| TikTok views per follower | 4.5% (1.5%) | 3% (1%) | 2.25% (0.75%) | 1.5% (0.5%) |
| TikTok engagement per view | 1.5% (0.45%) | 1% (0.3%) | 0.75% (0.23%) | 0.5% (0.15%) |
| Instagram views per follower | 3% (0.75%) | 2% (0.5%) | 1.5% (0.38%) | 1% (0.25%) |
| Instagram engagement per view | 1.2% (0.3%) | 0.8% (0.2%) | 0.6% (0.15%) | 0.4% (0.1%) |
| YouTube views per follower | 3% (0.75%) | 2% (0.5%) | 1.5% (0.38%) | 1% (0.25%) |
| YouTube engagement per view | 0.75% (0.23%) | 0.5% (0.15%) | 0.38% (0.11%) | 0.25% (0.07%) |
The lines are deliberately low, set to miss some fakes rather than accuse real creators. For what a normal creator looks like, compare with the medians below.
Typical views per follower (median and middle half of creators in Kolect's database)
| Platform | Nano<10K | Micro10K–100K | Mid-tier100K–500K | Macro500K–1M | Mega1M+ |
|---|---|---|---|---|---|
| TikTok | 31.6%13.9%–80.6%n=27,816 | 17.5%6.85%–51.3%n=119,186 | 10.9%4.02%–29.3%n=67,013 | 7.89%2.76%–21.5%n=17,043 | 5.14%1.78%–14.0%n=20,040 |
| 74.7%38.4%–154.8%n=33,932 | 29.4%14.7%–62.1%n=195,046 | 20.8%9.50%–46.9%n=92,975 | 19.9%8.98%–41.6%n=12,961 | 15.4%6.82%–34.5%n=10,454 | |
| YouTube | 34.1%11.7%–83.2%n=45,539 | 19.2%7.74%–43.9%n=77,718 | 9.73%3.61%–25.5%n=39,375 | 6.59%2.14%–19.7%n=9,385 | 4.59%1.30%–14.4%n=11,861 |
What 716,539 creator accounts show
Kolect ran the views-per-follower check on every account in its creator database with 5,000+ followers and median views above zero (data refreshed September–October 2026):
| Platform | Accounts | Below suspicious line | Below red-flag line | 1M+ below suspicious line |
|---|---|---|---|---|
| TikTok | 235,232 | 15.8% | 3.9% | 21.7% |
| 327,893 | 1.1% | 0.1% | 1.4% | |
| YouTube | 153,414 | 11.7% | 3.1% | 20.8% |
- About one TikTok account in six with 5,000+ followers gets far fewer views than its size suggests, rising to one in five above 500K followers. On TikTok and YouTube, check views before anything else.
- Instagram rarely trips this check: typical Instagram creators get views far above the line (a median of 29.44% of followers for 10K–100K accounts), so on Instagram engagement per view and likes per view do more of the work.
- One signal is not a fake follower rate. Low views also follow a pause in posting or a creator changing topic. The database holds creators indexed for brand campaigns, not a random sample of all accounts.
How to check for fake followers by hand
You can run the same screen in about ten minutes without any tool:
- Open the creator's last 10–20 posts and note views, likes and comments on each. Use the middle value, not the best post.
- Divide typical views by followers and compare with the line for their platform and size in the table above.
- Add likes, comments and shares and divide by views. Under the engagement line, people who watched did not care.
- Read twenty comments. Emoji-only replies, "nice post" on every video and languages the creator never uses point to bought engagement.
- Open ten followers. Accounts with no photo, no posts and random handles are bots.
- Check growth on a public tracker such as Social Blade: real accounts grow in steps after hits, bought followers arrive as a cliff overnight.
- Put the numbers into the checker above, and compare engagement with creators of the same size in the engagement rate calculator.
Run the full check free on Kolect
The manual screen works for one creator. For a shortlist, sign up free and let Kolect do it:
- Free every day. The first 20 fake follower checks each day cost nothing; after that a check is 5 credits, and every new account gets 90 credits. A check that reaches no verdict is not charged, and a creator checked in the last 7 days shows the earlier result for free.
- Any creator, any list. Paste profile links or @handles, up to 20 at once (10 on the free plan), check a saved collection in one go, or press "Check fake followers" on any search result.
- All eight checks. The full check reads the creator's recent posts and estimated audience, so it adds view consistency and audience location to the six checks above.
- Inside your AI assistant. Kolect's MCP server has a check_fake_followers tool, so Claude, Codex and other assistants can screen creators for you (setup guide).
Plans and credit packs are on the pricing page.
What this check cannot tell you
- It is an estimate, not an audit. It reads public numbers; it does not classify individual followers. Tools such as HypeAuditor and Modash analyse a sample of a creator's followers on Instagram, a deeper but slower and costlier check (see HypeAuditor alternatives compared).
- A low score means look closer, not proof of fraud. Read the comments and the follower list before you decide.
- Audience location is estimated by AI, which is why it costs only 5 points.
- Accounts under 5,000 followers get no verdict.
Once a creator passes, the influencer rate calculator shows what creators of that size are typically paid.