Why is follower count a bad way to pick influencers?

Because it answers "how many people" instead of "which people". A creator with 500k followers whose audience is hobbyist learners will not sell a $500/month B2B tool, while a 20k-follower creator whose audience is agency owners might. Follower count also says nothing about whether the audience holds budget authority — and it is the easiest number on the internet to inflate.

What does an audience persona actually contain?

The useful version has four parts: (1) role — what job the audience does; (2) seniority — are they decision makers who can approve spend, hands-on practitioners, or learners with no budget; (3) pains — the specific problems they keep coming back to; (4) recurring topics — what the creator returns to again and again. Together these tell you whether a pitch will land, which follower count never does.

How many posts do you need to read to judge an audience?

More than the two or three most people skim. A handful of posts shows what a creator talks about; it takes dozens to see who they consistently serve, because creators post off-topic and chase trends. We read up to 100 recent posts per creator and scale a confidence score with the volume — around 60+ posts gives high confidence, 15 or fewer gives a weak read, and under 5 we refuse to output a persona at all rather than guess.

Can’t I just look at the creator’s topic or niche tag?

Topic tells you the subject; persona tells you the buyer. "AI tools" covers a creator serving video editors and another serving enterprise IT buyers — same tag, completely different customers. This is why matching on niche keywords produces shortlists that look right and convert badly.

Stop Picking Influencers by Follower Count — Match on Audience Persona Instead

Updated 2026-07-16

Most influencer shortlists are built on two numbers — follower count and a topic tag — and both answer the wrong question. Follower count tells you how many people. A topic tag tells you what the creator posts about. Neither tells you who is listening, which is the only thing that determines whether your pitch lands. Here is how to read the audience instead, and a real case where doing so reversed the decision.

The mistake: on-topic but wrong audience

Take "AI tools" as a niche. Under that single tag you will find a creator whose viewers are freelance video editors, and another whose viewers are enterprise IT buyers. Same tag, same nominal niche — completely different customers. Pitch a B2B workflow platform to the first and you get polite silence; pitch a consumer video app to the second and you get the same.

The failure is not that the creator was irrelevant. It is that relevance was measured on the creator instead of on their audience.

What an audience persona should contain

Role — what job does the audience actually do? Media buying, SEO, support, e-commerce ops, engineering?

Seniority — and this is the part people skip: are they decision makers who can approve spend, practitioners who use tools but do not buy them, or learners with curiosity and no budget? A large audience of learners is a content-marketing asset and a terrible sales channel.

Pains — the specific problems they keep returning to. If your product does not touch one of them, no amount of reach helps.

Recurring topics — what the creator comes back to week after week, as opposed to one viral outlier.

Note what is missing from that list: follower count. It belongs in your reach planning, not in your fit decision.

How to read it yourself

Pull their recent posts — dozens, not three. A few posts show topics; you need volume to see the pattern of who is consistently being served. Creators post off-topic and chase trends, and a small sample lets one viral outlier define your whole impression.

For each post ask "who is this written for?" — not "what is this about". A tutorial on account structure and ROI is written for media buyers. A tutorial on turning a product photo into a video is written for content creators.

Read the comments for seniority signals. "What plan should I buy for my team" is a decision maker. "How do I do this for free" is a learner.

Write the persona down in one sentence, then hold it against your own target customer. If you cannot say both in the same breath, they do not match.

Only then look at reach — among creators whose audience is your customer, pick by size and cost.

A real case: 100 posts reversed the verdict

We ran this on a TikTok creator in the AI category with several hundred thousand followers — an account that scored well on every conventional filter: right niche, healthy grade, real engagement. On follower count and topic tag alone, an obvious "yes".

Reading 100 of their recent posts produced this persona: mid-sized content creators who want AI to generate video fast — hands-on practitioners, whose recurring pains were converting static images to video, quality loss on upload, and juggling too many tools.

Matched against a hypothetical AI ad-optimization product (target customer: media buyers at e-commerce and gaming companies), the verdict came back weak, 0.2 — the audience makes content, it does not buy media. Same creator, same numbers, opposite decision. The reach was real; it was just pointed at the wrong people.

What this does not fix

Persona matching tells you whether the kind of person is right. It does not tell you how many of them there are, and it cannot tell you whether two creators share the same followers — that requires full follower lists, which no platform exposes to third parties (we tested this and wrote up why precise overlap numbers are not measurable). Persona is the fit question; a small tracked pilot is still the conversion question.

mg.land automates the reading part: it pulls up to 100 recent posts per creator, extracts the audience persona (role, seniority, pains, recurring topics) with a confidence score that scales with how much content was available — and refuses to output a persona at all when there is too little to judge. Add your own product description and it scores the match both ways, telling you in plain language whether the two groups are the same people, and saying "don’t bother" when they are not. Free, no login.

Find creators in your industry in a minute

Free · sign up with email

Open the free tool →