Step by step
How to run a background check on an influencer. Properly, before you sign.
Six steps that cover what a brand actually needs to know, what each one involves if you do it by hand, and where an automated check earns its place.
Updated September 2026 · About a six-minute read
The short answer
To background-check an influencer, collect every account they run — not only the one in the pitch — then read the full history of each in every format it was posted in: captions, spoken audio, on-screen text and images, back to the first post. Recover deleted posts from web archives. Sort what you find into categories — hate speech, NSFW, violence, profanity, political, legal, public feuds, brand criticism, substance, deceptive — and attach the exact post and a stored copy of the media to every finding. Finish by checking the public record, carrying every allegation to its outcome, and decide against your own brand’s tolerances rather than a generic verdict.
The six steps
The work, in the order it has to happen.
Each step sets the scope for the next. Skip the first and the rest is a careful read of the wrong accounts.
- 1
Collect every handle, not just the one in the pitch
Start from the media kit, then keep going: the link-in-bio page, the accounts tagged in their own posts, the X or Threads account they treat as a group chat, the old YouTube channel, the gaming or streaming handle. Risk concentrates on the platform the creator thinks nobody from a brand is reading. Write down each platform and handle before you read anything — the scope you set here is the scope of the whole check.
A check that reads three polished accounts and skips the fourth is a check of the creator's marketing, not the creator.
- 2
Decide whether the creator takes part
There are two honest ways to run a check. A public-content review needs no participation and no notification — appropriate when you are still deciding whether to approach someone. A creator-authorised review has the creator connect their own accounts and consent to the analysis — appropriate once they are in conversation with you and you want the deeper, consented read. Pick one deliberately; mixing them without telling the creator is how a vetting process becomes a trust problem.
CreatorScore sells both as Quick and Verified, each $29.99. Verified is consented through each platform's OAuth; it never scrapes audience data.
- 3
Read the full history, in every format it was posted in
Go back to the first post, not the last hundred. For each account that means captions and text posts, but also the spoken audio in every video, the text burned into frames — memes, stitched replies, screenshots — and the images themselves. Captions are written for the algorithm; the sentence that ends a campaign is usually spoken forty seconds into a vlog or sitting in a frame with a clean caption above it.
By hand, this is the step that does not scale. Transcribing one ten-minute video takes longer than ten minutes.
- 4
Recover what was deleted
Deleting a post removes it from the feed, not from the internet. Check public web archives for captures of each profile over the years and compare them against the live timeline; anything in the archive that is no longer on the profile is a recovered deletion. Note the capture date and the removal — the fact that something was taken down is itself information a brand should have.
- 5
Categorise every finding and attach the evidence
Do not write a verdict. Sort what you found into categories — hate speech, NSFW, violence, profanity, political, legal, public feuds, brand criticism, substance, deceptive — and for each finding record the exact post, the frame or timestamp, the original-language quote and a stored copy of the media. Platform image links expire within days; a finding you cannot show six months later, when legal asks, is a rumour.
Judge severity with the speaker, audience and region in frame. A term reclaimed inside a community is an insult from outside it.
- 6
Check the public record, then decide in context
Search for the creator's name alongside lawsuit, arrest, settlement, fine, ban and apology — and for every allegation you find, keep reading until you find the outcome: charges declined, case dismissed, settled, served, or genuinely unresolved as of today. An allegation reported without its resolution is not a risk assessment. Then make the call against your own brand's tolerances, because the same stance that disqualifies a creator for one brand is exactly why another would hire them.
What to look for
Ten categories, and where each one hides.
The categories are the easy part. The reason manual checks miss things is format: the signal is in the voice track, the frame or the deleted post, not the caption.
- 01
Hate speech & discrimination
Content targeting or demeaning people by race, ethnicity, gender, religion, sexual orientation or disability.
Where it hides: Old replies and quote-posts, and slurs spoken in video rather than typed in a caption.
- 02
NSFW & explicit content
Sexually explicit or suggestive imagery and language a family or retail brand could not sit next to.
Where it hides: Frames inside a video, linked accounts on other platforms, and posts later deleted.
- 03
Violence & weapons
Depicting, promoting or glorifying violence or weapons.
Where it hides: Background objects in a frame, stitched clips, and gaming or prank content read out of context.
- 04
Profanity & vulgar language
Habitual strong language — reported so you can judge it against your own tolerance, not as a verdict.
Where it hides: Spoken audio. Captions are written for the algorithm; the swearing is in the voice track.
- 05
Political & divisive content
Partisan statements, endorsements and divisive commentary.
Where it hides: Election-period posts from years ago, and X threads that were never cross-posted.
- 06
Legal & criminal matters
Arrests, lawsuits, regulatory action and references to criminal proceedings — each carried to its outcome.
Where it hides: Press coverage rather than the creator's own feed, and the resolution that came years after the headline.
- 07
Public feuds
Calling out, attacking or feuding with other creators, brands or public figures.
Where it hides: Reply chains and response videos, which rarely surface in a profile scroll.
- 08
Brand criticism
Attacks on specific brands or companies — including, possibly, yours or a competitor you cannot be seen beside.
Where it hides: Review-style videos and boycott calls spoken rather than captioned.
- 09
Substance promotion
Promotion of alcohol, drugs, vaping or related products, and depictions of intoxication.
Where it hides: Party footage in the background of unrelated posts; sponsored vape content on a secondary account.
- 10
Deceptive & scam content
MLM promotion, fake testimonials, misleading income or health claims, and undisclosed paid promotion.
Where it hides: Affiliate codes in pinned comments, and sponsorships that were never labelled as ads.
Manual vs automated
Where a human review and an automated one part ways.
A person is better at judgement; software is better at reading everything. A good process uses the software to put the full record in front of the person.
| By hand | Automated (CreatorScore) | |
|---|---|---|
| Coverage | The last few hundred posts a person has time to scroll, usually on the platform you were pitched. | The full available public history on up to eleven platforms, with a per-account statement of how many posts were read. |
| Video and audio | Watched at 1× speed, or skipped. Transcribing by hand is slower than watching. | Every video transcribed and analysed frame by frame; on-screen text read in six scripts. |
| Deleted content | Rarely checked; requires knowing the archive tooling and comparing captures by hand. | Web-archive captures compared against the live profile automatically, with capture dates. |
| Non-English content | Machine-translated first, which flattens sarcasm and in-group register, or not read at all. | Detection runs in the original language; only the evidence shown to you is translated. |
| Evidence | Screenshots in a folder, with links that expire. | Each finding traced to the post, frame or line, with media copied to storage so it does not expire. |
| Time | Days per creator for a thorough read; hours for a shallow one. | Typically within the hour; a decade-deep catalogue with thousands of videos takes longer. |
| Price | Billed by the hour, or by the report — rarely published. | $29.99 per creator, Quick or Verified, no subscription. |
The honest case for doing it by hand is a single creator, a small catalogue and a reviewer who already knows the niche. The honest case against is everything else: video, other languages, deleted posts, and the second account. Most brand crises trace back to one of those four, and none of them is a reading problem — they are a coverage problem.
If you run checks regularly, the vetting checklist turns these steps into a form your team can sign off, and the cost breakdown puts the hours above against a flat price. Both tiers of the CreatorScore check return the same depth of read; see what the check includes.
Social Background Check
Run the check instead of the spreadsheet.
Every post, transcript, frame and recovered deletion on the accounts you submit, grouped into the ten categories above with the evidence attached to each finding. Results typically within the hour.
- Quick or Verified — $29.99 per check, no subscription
- Findings with evidence, in the original language
- Deleted content recovered from web archives
- No score. A categorised report, not a number
Common questions
How long does it take to run a background check on an influencer by hand?
+
Longer than most teams budget. Reading alone scales with the catalogue: at roughly ten seconds per post, a mid-sized creator with 4,000 posts across their accounts is about eleven hours of reading before a single video has been watched or a single deleted post looked for. Watching video adds its own runtime at 1× speed. An automated check does the same read, including transcription and frame analysis, typically within the hour.
Do I need the influencer's permission to run a background check?
+
Not to review their public content for a commercial partnership decision — that is publicly available material and the Quick tier reads only that. You do need their participation for a creator-authorised check: the Verified tier has the creator connect their own accounts through OAuth, so it is consented by definition. Neither is an employment screen; see the compliance note below.
What should I look for in an influencer's history?
+
Ten categories cover almost everything that has ever ended a campaign: hate speech and discrimination, NSFW content, violence and weapons, profanity, political and divisive content, legal and criminal matters, public feuds, brand criticism, substance promotion, and deceptive or scam content including undisclosed ads. Look for each in every format — caption, spoken audio, on-screen text, image — and on every account, not only the one you were pitched.
Can a background check find posts an influencer deleted?
+
Often, yes. Public web archives capture profiles over time; comparing a capture against the live timeline shows what was removed and when. CreatorScore does this as part of every check and reports each recovered post with its capture date. It cannot recover what was never captured, and it says so rather than implying a complete record.
Is an influencer background check the same as a brand-safety score?
+
No. A background check is a one-time, full-history inspection that returns categorised findings with evidence and deliberately produces no number. A CreatorScore is a separate, continuously updated trust score across a creator's whole footprint, used to manage a roster over time. They answer different questions — what is in this person's past, versus how is this creator trending — and buying one does not give you the other.
Can I use this process to screen a job candidate?
+
Not with CreatorScore, and you should be careful doing it by hand. Social media screening for employment, housing, credit or insurance decisions is regulated in the United States by the Fair Credit Reporting Act and requires a consumer reporting agency with specific consent, dispute and adverse-action processes. CreatorScore is not a consumer reporting agency; its reports are for brand-partnership decisions, which are commercial, not consumer reports.
Keep reading
More on background checks for creators
Influencer background check
The product page: what a $29.99 full-history check reads, what it returns, and how to order one.
ReadInfluencer vetting checklist
A printable checklist across identity, audience, content history, public record, FTC history and brand conflicts.
ReadInfluencer background check cost
What drives the price of a check, what manual vetting costs in hours, and what $29.99 covers.
ReadSocial media screening for brand partnerships
For compliance, procurement and HR readers: scope, consent, data handling and the questions to ask a vendor.
ReadSocial media background check
What a social media background check is, what it covers, and why it is not an employment screen.
ReadBackground check vs monitoring
One reads the whole past once; the other watches what happens next. Which you need, and when you want both.
Read