How to check for fake followers

What counts as a fake follower

A fake follower is an account that inflates a creator's follower count without ever being a real viewer. That covers purchased bulk followers, bot accounts, and engagement-pod activity traded rather than earned.

TikTok's Integrity and Authenticity guidelines are explicit about this. TikTok states that if it detects accounts or content with inauthentic metrics, it will remove fake likes, followers, or other inflated signals. So the platform treats bought followers as a policy violation, not a grey area.

Worth separating from fakes: inactive followers. Real people who followed once and stopped opening the app. They drag engagement rate down the same way fakes do, and from the outside they look similar. The practical consequence is identical, which is why the checks below measure the audience that shows up rather than trying to label individual accounts.

How do you calculate engagement rate to spot fake followers?

Engagement rate is the fastest signal because buying followers is cheap and buying proportional engagement is not. An account with 200,000 bought followers and organic-only likes shows a rate far below its tier.

The formula most brands use:

• Instagram: (likes + comments) divided by followers, times 100.
• TikTok: (likes + comments + shares) divided by views, times 100.

Those denominators differ, which is why the two platforms are not comparable. TikTok is measured on views, so TikTok rates read higher than Instagram rates by design. Comparing a TikTok number to an Instagram benchmark is the most common mistake here.

Against our 2026 benchmarks:

• Instagram: 1% to 3% is average for most accounts. Nano and micro creators reach 3% to 8%.
• TikTok: 3% to 9% for mid-sized accounts. Nano creators (1,000 to 10,000 followers) often hit 8% to 15%.

Under 1% engagement on a large Instagram account is the threshold worth acting on. That gap between audience size and audience response is the signature of a purchased or long-dead following. You can run the math with our free engagement rate calculator, which compares the result against tier benchmarks.

One caveat before you treat a low number as proof: engagement rate falls naturally as accounts grow, because reach outpaces the share of people who interact. A 500,000-follower account at 1.2% is normal. A 20,000-follower account at 1.2% is not.

Which signals actually indicate fake followers?

Engagement rate gets you a flag, not a verdict. These four checks are what separate a real anomaly from a bought audience.

Do the comments respond to the video?

Open three or four recent posts and read the comments as content, not as a count. Real comments reference something specific: the product, a claim, a moment in the video. Purchased comments are interchangeable ("Nice!", "love this", stacked emojis) and would fit under literally any post.

Be careful with this one. Giveaways, trending sounds, and a young audience all produce short low-effort comments. What matters is whether any thread shows genuine engagement with the topic, not whether every comment is thoughtful.

Does follower growth line up with a real event?

Look at growth over time. Organic growth is uneven but explainable: a video takes off, a press mention lands, a collaboration posts. Purchased growth appears as a vertical jump with no corresponding content spike.

A spike alone is not evidence. The question is whether you can point at the thing that caused it. If a creator gained 40,000 followers in a week and their best video that week did 12,000 views, the two numbers do not reconcile.

Are view counts consistent?

Real audiences produce a band. A creator with 50,000 engaged followers reliably lands somewhere in a recognizable range, with occasional breakouts above it.

An account with an inflated follower count often shows views that sit far below what the follower number implies, because the fake portion never watches anything. Views running at a small fraction of followers, post after post, is a strong signal.

Does the audience geography match your market?

Ask for the creator's audience breakdown. If you sell to US buyers and the audience is majority outside your shipping range, the partnership will not convert regardless of whether the followers are real. Bulk follower services also tend to deliver from a narrow set of regions, so a geography that has nothing to do with the creator's language or content is worth a question.

When a low number is not fraud

Several honest situations produce numbers that look bad:

• A viral video inflates the denominator. One breakout post can add followers who never engage again, depressing the rate for months.
• Large accounts engage at lower rates. This is arithmetic, not decline.
• A niche audience watches without interacting. Some categories read quietly and buy anyway.
• Bought followers can be involuntary. Creators get targeted with unsolicited bot follows, and some inherit a bad audience from an agency that ran growth services without telling them.

That last one is why the useful framing is commercial rather than accusatory. You are deciding whether a partnership is likely to work, not building a case. Ask the creator about anything odd. A real creator usually has a straightforward explanation and often volunteers it before you ask.

A five-minute vetting process

Run this before any sample or payment:

1. Calculate engagement rate and compare it to the creator's tier on the correct platform benchmark.
2. Read comments on three recent posts, including an average one rather than only the best performer.
3. Check growth against content. Look for a spike with no video to explain it.
4. Compare views to followers across several posts to see whether the band is consistent.
5. Request audience geography and age for anything above a sample-level commitment.

Anything that fails two or more checks is a conversation, not an automatic no. Anything that fails all five is a pass.

For a paid partnership, the strongest protection is structural: start with a commission-based or sample-based deal before a flat fee. If an audience is fake, affiliate deals expose it immediately, because clicks and orders do not appear regardless of how the follower count looks. Performance-based terms make audience quality self-reporting.

Why this matters for TikTok Shop brands and agencies

On TikTok Shop, fake followers hit differently than on a pure awareness campaign. You are not buying impressions, you are buying the chance that a video converts to an order. A creator with 100,000 fake followers produces zero GMV, and you find out after the sample has shipped and the commission structure is live.

The scale makes it worse. A brand seeding 300 creators a month cannot run a manual audit on each one. That is how bad audiences enter a program: not because nobody checked, but because checking 300 profiles by hand is not a job anyone does consistently.

This is the practical argument for commission-first structures at the top of the funnel. When creators earn on orders rather than upfront fees, an inflated audience costs you a sample instead of a flat fee, and real performance surfaces within a couple of weeks. Reserve flat fees for creators who have already delivered.

Agencies feel this most, because a client sees the GMV number, not the vetting work behind it. Being able to show why each creator was approved, and what their engagement looked like going in, turns a judgment call into something defensible on a monthly report.

Hubfluence surfaces engagement rate, posting cadence, and actual TikTok Shop GMV per creator, so audience quality is a filter you apply to a list rather than an audit you run one profile at a time. If you want to see how that screening works against your current shortlist, book a demo and we will run it on real creators in your category.