Complete Influencer Vetting Checklist

What matters most

Core areas are identity and representation, content and audience relevance, performance context, engagement quality, suspicious signals, sponsorship history and saturation, brand safety, legal and platform eligibility, conflicts, production capability, communication reliability, and data confidence.

Review a defined representative sample: recent content, normal and high-performing posts, sponsored work, platform mix, and older content only where the campaign risk justifies it. Record exact evidence instead of labels such as safe or authentic.

Use proportionate outcomes: approve, approve with conditions, request clarification, escalate, hold, or reject with a reason. Missing information should become a question or confidence note, not a disguised accusation.

A practical workflow

1. 01

   Set the review window, platforms, evidence standard, and campaign-specific risk areas.

2. 02

   Verify identity, accounts, representative, relevance, audience, performance, and commercial history.

3. 03

   Review content, comments, suspicious patterns, conflicts, disclosure, safety, and professionalism.

4. 04

   Separate observed evidence, inference, contradiction, and not-enough-data states.

5. 05

   Record outcome, conditions, escalation, reviewer, date, and next recheck trigger.

Scale the review to the risk

Reviewing every creator to the same depth is both expensive and wrong: it over-scrutinises a gifting campaign and under-scrutinises a regulated one. Set two or three tiers and write down what triggers each, so the decision is made by the campaign rather than by whoever happens to be doing the review.

A light review suits low-commitment activity such as seeding with no obligation: confirm the account is real and active, confirm relevance, scan recent content, and check nothing obvious conflicts. A standard review suits a paid partnership: add audience composition, engagement quality, sponsorship history and saturation, a defined content window, and a check on representation and eligibility.

A deep review is for campaigns where a mistake is expensive or public: regulated categories, financial or health claims, anything reaching a young audience, high-value ambassadorships, and anything the brand's own executives will be associated with. That tier justifies a longer content window, a second reviewer, and an explicit sign-off. It should be rare, and the trigger should be written down, because a deep review applied by instinct is indistinguishable from surveilling people you happen to feel uneasy about.

What engagement rate can and cannot tell you

Engagement rate is at least three different metrics wearing one name. Engagements divided by followers, engagements divided by views, and engagements divided by reach produce different numbers with different meanings, and they are not comparable to each other. Before comparing two creators, confirm the same denominator was used for both, because most published benchmarks do not say which one they used.

The rate also falls predictably as an account grows, which is a property of large audiences rather than a defect in the creator. Comparing a creator with fifteen thousand followers to one with two million on the same threshold will reject the second one every time. Compare within a size band and within a platform, and treat the number as one input rather than a gate.

The signals that actually suggest inauthentic engagement are structural rather than numerical: comments that respond to nothing specific in the content, a follower count that steps rather than curves, engagement that does not move with view count, an audience whose stated geography does not match the content's language or references, and a gap between public engagement and the platform's own reported reach when the creator shares it. Any one of these has innocent explanations. Several together, on a creator whose growth has no visible cause, is worth a question rather than an accusation.

Read the comment section

The comment section is the cheapest high-signal check available and the one most often skipped, because it does not produce a number. It tells you what the audience is actually there for, whether they trust the creator's recommendations, how they reacted to previous sponsorships, and whether the tone of the community is one the brand wants to stand next to.

Look at a sponsored post specifically, and read past the top comments. Audience reaction to a previous partnership is the closest available proxy for how they will react to yours: whether the disclosure was accepted, whether people engaged with the product or only with the creator, and whether the sponsorship was treated as normal or as a departure.

Also read what the creator does in their own comments. A creator who answers questions, corrects mistakes and handles criticism in public is describing how they will behave when something goes wrong in your campaign, which is information no metric captures.

Record confidence, not verdicts

A vetting record that says a creator is safe has thrown away everything that made the decision reviewable. Record the finding, the evidence that supports it, the date, and which of three states it belongs to: observed directly, inferred from indirect evidence, or not enough data. The third state is a real answer and should be recorded rather than resolved by assumption in either direction.

Keep the distinction between no adverse findings and clean. The first is a statement about the review; the second is a statement about the creator, and no review can support it. This matters most in the situation that eventually arrives, where something surfaces after launch and the question becomes what was known and when.

Write down the recheck trigger with the decision. Creators change, and a review is only current for as long as the conditions it was made under hold. A new controversy in the category, a change of representation, a platform enforcement action, a significant shift in content direction, or simply the passage of a defined period should all send a creator back through review before the next campaign rather than after it.

Common mistakes

• Searching for isolated controversy without a defined relevance or time standard.
• Equating unusual growth with fraud before investigating context.
• Using sensitive inferred traits as brand-safety shortcuts.
• Failing to retain the evidence behind a decision.

Working checklist

• Review scope and evidence standard match campaign risk.
• Fit, audience, performance, sponsorships, safety, conflicts, and operations are covered.
• Findings link to source evidence and dates.
• Missing and negative evidence are distinct.
• Outcome, conditions, owner, and recheck trigger are recorded.

Questions and answers

What is a normal engagement rate?

There is no single figure, and any benchmark quoted without a denominator, a platform and a size band is not usable. Build your own reference instead: take fifteen to twenty relevant creators in the same size band on the same platform, calculate the rate the same way for all of them, and use the distribution as the comparison. An outlier against that set is meaningful; an outlier against a generic benchmark usually is not.

How far back should you review a creator's content?

Set the window by campaign risk before you start reviewing, so the standard is not chosen after you find something. A standard partnership is usually well served by six to twelve months of content plus every sponsored piece in that period. A deep-tier review extends further. An open-ended search through someone's entire posting history is rarely proportionate and produces findings you then have no consistent standard for judging.

Should an old post disqualify a creator?

Decide the standard first, in writing, and apply it to everyone. A useful test asks how old the content is, whether the view was central or incidental, whether the creator has since addressed it, whether it touches the campaign's own subject or audience, and what the realistic consequence is for the brand. Applying different standards to different creators is both unfair and, in practice, the thing that becomes the story.

Do you need permission to review a creator's public content?

Reviewing content someone has published publicly is ordinary diligence and does not require permission. The line worth respecting is between reviewing what a creator published and assembling a profile of inferred characteristics about them, particularly sensitive ones. Using inferred traits as a selection or rejection shortcut is a discrimination and data-protection problem regardless of jurisdiction, and it is also poor vetting: it substitutes a guess for the evidence you were meant to collect.