How to vet a creator: audience authenticity, engagement, and the fake-follower problem
On a typical account, roughly a fifth of followers are fake or inactive. Here is how to read the signals that separate a real audience from an inflated one, before you pay, with the four checks that catch most of it.
The market for fake influence is large and cheap. A few dollars buys tens of thousands of followers, and the supply chain has kept pace with detection: bought likes, comment pods that trade engagement in a group chat, view farms in warehouses. The result is that the single number a brand looks at first, the follower count, is the one number that tells you almost nothing about whether real people will see the post and buy.
Vetting is the fix, and the good news is that it is mostly not guesswork. The signals that separate a real audience from an inflated one are measurable, and most of them are visible from the public profile before you ever sign a contract. Here is how to read them.
Engagement rate, but against the right benchmark
Engagement rate is the workhorse metric: likes plus comments, divided by followers. The mistake brands make is judging it against a flat number they read somewhere. A 2% engagement rate is healthy for a mega account with millions of followers and thin for a nano account with five thousand. Engagement dilutes as audiences grow, so the only honest read is against the floor for the creator's tier.
Healthy Instagram engagement rate by tier
- Nano (1k–10k)~4.0%
- Micro (10k–50k)~2.5%
- Mid (50k–500k)~1.6%
- Macro (500k–1M)~1.2%
- Mega (1M+)~0.9%
Use the median of the last six to twelve posts, not the single best one. And watch the other end of the scale too: an engagement rate far above the healthy line for the tier is its own flag. It usually means an engagement pod or a cherry-picked post rather than a genuinely extraordinary audience.
The comment ratio gives the bots away
Likes are the cheapest thing to fake, so the smart move is to look past them. Bought engagement is overwhelmingly likes: a like-bot taps a heart, it does not write a sentence. So when comments are a near-zero share of total engagement, well under a third of a percent, the likes above them are doing a lot of suspicious work.
The opposite pattern is also a tell. Comment pods trade engagement in bulk, so a comment ratio that is oddly high, paired with a wall of generic praise ("🔥", "nice!", "love it"), points to coordinated activity rather than a community. Real comments are specific: questions about sizing, references to a previous post, the small messy texture of actual people. Read fifty of them and you will know.
“Likes are cheap to fake and comments are expensive. The ratio between them is the cheapest lie detector you have.”
Followers arrive in batches when they are bought
Organic growth is gradual and a little lumpy: a viral post here, a steady climb there. Purchased followers arrive all at once. Pull the follower history on a tool like Social Blade and the bought accounts give themselves away with vertical jumps, tens of thousands of new followers in a few days, often followed by a flat line or a slow bleed as the fakes get purged.
One spike around a genuine viral moment is fine, and you can usually corroborate it against a post that actually took off. A staircase of unexplained spikes is not.
Real but irrelevant is still a miss
Authenticity and relevance are different questions, and a creator can pass one while failing the other. An audience can be entirely real and entirely useless to you if it sits in a country you do not ship to or an age bracket you do not sell to. This is where the public numbers run out and you have to ask. Any creator worth paying can show you a screenshot of their own analytics: audience country split, age and gender, reach on recent posts.
If they cannot or will not, that is an answer in itself.
A workflow that holds up
Put the four signals in order and the process is quick. Screen the public profile first, engagement against tier, comment ratio, growth pattern, a skim of the comments. Anyone who clears that gets a request for first-party analytics. Anyone who clears that gets a small paid test before a bigger commitment, because real conversions beat any audience report ever written.
From shortlist to signed
- 01
Screen the profile
Engagement vs. tier, comment ratio, growth graph, a read of the comments.
2 min
- 02
Request first-party proof
Audience country, age and gender, recent reach. Screenshots from their own dashboard.
before paying
- 03
Run a paid test
One brief before a bigger deal. Measure conversions, not impressions.
1 brief
- 04
Scale or skip
Repeat with the winners, or brief from a pre-vetted network and skip the audit.
ongoing
Creator vetting scorecard
Free: the four-signal rubric and tier benchmarks on one printable sheet.
We built a free tool that runs the first gate for you: enter a creator's public numbers, answer three judgement checks, and get a live authenticity score with the flags called out. And for brands that would rather skip the audit entirely, every creator on the Idukki network has already cleared a screening pass and grants content rights up front.
Related reading + tools
More from Rohin Aggarwal
- Conversational commerce
Why we built the Conversational PDP
Most product-page exits are a single unanswered question. Here is the case for answering it on the page, from your own evidence, and the story of why we built a Q&A that is curated-first and AI-second.
- Strategy
PDP before and after UGC: what actually changes on the page
Strip a product page back to brand-only content, then layer verified customer photos, video and reviews into the middle scroll, and watch what moves. A scroll-by-scroll look at the before and after, the numbers the public studies actually support, and where "just add UGC" gets oversold.
- Industry playbook
How to get brand deals as a UGC creator (without a big following)
You don’t need 100k followers to get paid for content. You need a believable audience, a clear offer, and proof you can deliver. Here’s the playbook, including the one thing brands check before they pay.
