How to Vet Influencers: The Checks That Predict Performance

Vetting an influencer means answering three questions before you send money: do their posts actually get seen, is the audience seeing them able to buy from you, and will this person deliver on time. Follower count answers none of the three, which is why programs that screen on follower count keep getting surprised.

Budget ten minutes per creator. That is enough to kill the obviously bad ones and price the good ones properly, and short enough that you do it on all forty candidates rather than the first six.

Check median views before anything else

Open the creator's last twelve posts on the platform you are hiring them for. Ignore the viral one and the flop, and take the median view count of the rest. That number is your best single predictor of what your post will do, and the number your rate should be built on.

Two things it tells you that followers cannot. First, whether the account is currently being distributed — a creator with 200,000 followers whose recent posts sit at 4,000 views is not being shown to their own audience, and buying that account buys the follower number, not the reach. Second, how consistent they are. A creator whose last twelve posts range from 20,000 to 45,000 views is a safer buy than one ranging from 2,000 to 900,000 even at a higher median, because you are buying one post and the spread is your risk.

As rough orientation — these are working ranges to argue against, not measurements of any market:

Platform	Median views as a share of followers	What is worth a second look
TikTok	Highly variable; 10–100% is all normal	Under ~10% consistently, on an account that isn't new
Instagram Reels	Often exceeds follower count on distributed accounts	Reels views far below follower count for months
Instagram feed posts	Typically a fraction of followers	Feed engagement wildly out of step with Reels reach
YouTube Shorts	Almost unrelated to subscriber count	Judge on the last 20 Shorts, nothing else

The ranges are wide on purpose. Anyone quoting a single benchmark for "good engagement" across platforms and niches is selling something. What matters is the creator's own recent trend line, not a cross-industry average.

The fake-follower signals worth your time

Most audience fraud is not a bot farm. It is an account that bought followers once in 2023, or a creator who spent a year in engagement pods, or a genuinely popular account whose audience is in a country you do not ship to. Three checks catch nearly all of it:

Comment quality. Read thirty comments on two recent posts. Real audiences ask specific questions, reference the creator by name, and argue with each other. Fabricated engagement is generic and interchangeable — emoji strings, "nice", one-word compliments with no reference to the content. This is the most reliable check and it takes two minutes.

Follower growth shape. Look for step changes. Organic growth is jagged but continuous; a flat line with a vertical jump of 30,000 in a week, followed by flat again, is a purchase. Most creators will share a screenshot of their growth chart if you ask, and the request itself is a useful filter.

The ratio that does not add up. An account with 300,000 followers, 800 likes and 6 comments per post has a problem. Which problem matters less than the fact that you are paying for reach that is not there.

What to do with a red flag: ask. "I noticed your Reels views have dropped over the last couple of months — is something going on with the account?" gets an honest answer surprisingly often. Creators know when they have been throttled, when they changed niche, when a post got them a strike. A creator who explains it clearly is usually a good hire at a lower rate.

Audience fit is where most vetting stops too early

A real audience that cannot buy from you is worth nothing. Ask every shortlisted creator for a screenshot of their audience insights — top countries, age split, gender split. Every platform gives creators this, it takes thirty seconds, and a creator who will not share it is telling you something.

The checks that actually change decisions:

• Geography. If you ship to the UK and 62% of their audience is in the US and India, your cost per reachable viewer is three times what the rate implies. This is the most common and most expensive vetting miss, and on TikTok especially, distribution crosses borders more readily than creators expect.
• Age. A skincare brand for 35+ buying a creator whose audience is 68% 18–24 will get views and no orders.
• Purchase context. Does this audience buy what the creator recommends, or watch for entertainment? The tell is whether previous sponsored posts have comments asking where to get it.

Then find their last three paid posts. Did those perform in line with their organic median, or well below it? A creator whose sponsored content underperforms their organic by 60% has an audience that tunes out ads, and you are the next advertiser to find out. Check who they have worked with recently too — not to be impressed, but to catch a direct competitor in the last 90 days, which is a real risk and a reason to ask for exclusivity.

Brand safety and reliability, in four questions

Scroll back six months of their content, not two weeks. You are looking for what becomes a screenshot: content you would not want your logo beside, a public falling-out, a pattern of undisclosed ads. That last one is both an ethical and a legal problem — a creator who does not label paid partnerships will not label yours either, and the FTC disclosure obligation sits with the brand as well as the creator.

Then, in the first email exchange, establish four things:

1. Do they have a media kit and a rate card? Not because you will pay the rate card, but because a creator who has one has done this before.
2. Have they hit a deadline for a brand recently? Ask for one reference or example campaign.
3. How do they handle revisions? The answer reveals whether they treat this as a service or a favour.
4. Are they invoicing as a business, and where? This sets your payment path and tax paperwork, and is better known now than after the content is live.

The part that gets lost is the record. You vet forty creators, hire eight, and three months later nobody remembers why you passed on the other thirty-two — so the next campaign re-vets the same people from scratch. This is where CreatorCast earns its place: the discovery and outreach flow keeps the median views, audience notes and pass reason on the creator record, so a rejected creator stays rejected with a reason attached and a strong one resurfaces next time you brief in their niche. The second year of a program should be cheaper than the first, and that only happens if the vetting is written down.

One last discipline: vet again before you rebook. A creator who was your best performer in March may have changed niche, lost distribution, or picked up a competitor since. The re-check is the cheapest three minutes in the program.

Frequently asked questions

What engagement rate is good for an influencer? There is no single number that survives contact with a real niche. Compare a creator against their own recent history and against the other creators on your shortlist in the same category, and buy on median views rather than on a percentage.

Should I pay for an influencer analytics tool? Only once you are screening more creators than you can read by hand — roughly 50+ a month. Below that, thirty comments and a screenshot of audience insights tell you more than a third-party quality score, which is itself an estimate.

Can I ask a creator for proof of their audience? Yes, and you should. Requesting screenshots of recent post insights and audience demographics is standard practice, not an insult. Ask for the raw screenshot rather than a figure typed into an email.

How do I vet a creator with no brand-work history? Weight consistency and comment quality more heavily, keep the first deal small and single-deliverable, and be explicit in the brief about deadlines and revisions. First-time creators are often excellent value; they need the process spelled out.