Accuracy Benchmark · v1.1

The brand-safety score you can check

Anyone can claim “most accurate.” We're the only creator brand-safety score that lets you verify it — built for the creators brands actually partner with: the 10K–1M middle class, not just the celebrities everyone has already heard about. Same creator, same inputs, same score, with the evidence shown and the date it was scored.

Same production pipeline, same rules — run on real mid-tier creators:

79–95
six clean creators'
Brand Safety agent
vs
55
a course creator
no search would flag

At a glance (v1.1, July 2026): run through the same v2.2 seven-agent pipeline, six creators carrying zero risk flags scored 78–87 with Brand Safety 79–95; a course creator with no news footprint was capped at 55; a creator with one lawsuit reference and no flagged post was surfaced at 62 rather than condemned; and a creator with a legally documented felony landed at 36 under a critical-controversy cap. Every figure is the creator's unified cross-platform footprint.

Why this page can exist

A score you can't reproduce can't be benchmarked

Reproducible by design

The same creator and the same inputs always produce the same number. That's what makes a benchmark possible at all.

Catches what you can't Google

The risk that matters most for mid-tier creators never makes the news. The score finds it in the content itself — and shows the posts.

Why others can't

A self-learning model that adapts to each brand's risk tolerance has no fixed answer to measure — so there is nothing to benchmark.

Open methodology

How the benchmark works

Anyone with the same inputs gets the same answer — which is the whole point.

1 — Labeled panel

Real mid-tier creators (10K–1M followers) — the people brands actually partner with. Some carry documented risk; a control group is clean. Every label is backed by evidence before scoring.

2 — Production pipeline, no special-casing

Each creator runs through the exact same v2.2 seven-agent pipeline a paying brand uses — same weights, same knockouts. Nothing is hand-tuned to make the benchmark look good.

3 — Did it separate risk from clean?

We check whether risk-carrying creators are graded down or capped, whether clean creators land in safe tiers, and whether the specific evidence is surfaced rather than just counted.

4 — Show the receipts, dated

Every case cites the creator's unified footprint score, the Brand Safety agent from that same row, and the date it was scored. Flagged creators are anonymized — the proof is the pattern, not a name.

Verified results

Real production scores — not a demo

An accurate score has to do two hard things at once: catch the risk a brand could never find on their own — and not blacklist a good creator for the topics they cover. Here's both, with the date each score was produced.

1 · The risk that never makes the news — caught anyway

55Poor
Risk caught — anonymized

Course / “make money online” creator

~800K on the largest account · 4-platform footprint · Brand Safety 70 · scored 23 April 2026

What we found: A paid business course sold with claims the content can't support, pushed in 39% of analyzed posts — plus a no-refund policy and student complaints. The pattern was detected in the posts themselves; a web record corroborates it independently.

What the score did: The deceptive-content knockout requires both conditions — scam/get-rich-quick patterns in at least 20% of posts AND external corroboration — so it fired and capped the footprint at 55/100. A brand looking at a polished 800K profile sees Poor, with the offending posts cited individually.

2 · Visibility, not verdict — surfaced, but not condemned

The flip side of accuracy. An external reference is not the same as something the creator posted, and the fastest way to lose a creator's trust is to score them as if it were. We show the brand the signal, say plainly where it came from, and let them decide.

62Fair
Surfaced, not condemned

Baking & lifestyle creator

~1.26M total reach · 3-platform footprint · Brand Safety 77 · scored 7 July 2026

What we found: One lawsuit reference and one published article naming the creator. Nothing in her own content was flagged — not a post, not a comment, not a frame.

What the score did: The flag states that in its own words: “no individual in-app post was flagged, so this alert rests on the external reference(s).” The brand sees the external record and decides. We surface it; we don't pretend her content did something it didn't.

3 · Clean creators — correctly scored safe

Six creators with zero risk flags and zero knockouts. Not catching risk is easy; not inventing it is the harder half.

87Excellent
Clean control

Nava Rose

435K followers · Instagram · Brand Safety 93 · scored 14 April 2026

Public record: Fashion and DIY creator. Zero risk flags in the system.

What the score did: Correct non-flag. No invented risk, no penalty for topics she covers.

86Excellent
Clean control

Lauren Giraldo

~1.59M total reach · 2-platform footprint · Brand Safety 79 · scored 14 April 2026

Public record: Lifestyle creator. Zero risk flags in the system.

What the score did: Lands in Excellent on the footprint, not on her best single account.

85Excellent
Clean control

Katie Sturino

~821K total reach · 2-platform footprint · Brand Safety 92 · scored 14 April 2026

Public record: Beauty and body-positivity creator. Zero risk flags in the system.

What the score did: Correct non-flag across both platforms.

80Excellent
Clean control

Rambo The Puppy

~376K total reach · 4-platform footprint · Brand Safety 95 · scored 2 July 2026

Public record: Pet creator across four platforms. Zero risk flags in the system.

What the score did: Highest Brand Safety agent score in the panel — 95. Four platforms resolved to one creator, one score.

80Excellent
Clean control

Dr Rupy Aujla

634K followers · Instagram · Brand Safety 86 · scored 14 April 2026

Public record: Doctor and food creator. Zero risk flags in the system.

What the score did: Health content scored without tripping the unqualified-medical-claims cap — because he has the credentials, which the cap checks for.

78Good
Clean control

Hyram

~6.23M total reach · 2-platform footprint · Brand Safety 80 · scored 15 April 2026

Public record: Skincare creator. Zero risk flags in the system.

What the score did: Safe tier, no flags — a creator who reviews products for a living isn't penalized for it.

4 · And yes — the critical, public-record case too

36Poor
Critical — capped

Taylor Frankie Paul

~8.9M total reach · 2-platform footprint · Brand Safety 77 · scored 13 July 2026

What we found: Pleaded guilty to felony aggravated assault (2023); ABC shelved her already-cast Bachelorette season in 2026 after the incident video resurfaced. The live flag names the corroborated controversy in full.

What the score did: The critical web-controversy knockout capped the footprint at 40 — and the seven agents landed it at 36, below the cap. Note her Brand Safety agent still reads 77: her posted content is unremarkable. The risk is in the public record, which is exactly why we search it.

Source: NPR

The honesty note

This is a living benchmark

v1.1 rebuilt the panel against production. Every case here was re-read from the database on 30 July 2026, and anything that didn't match was corrected or removed — including two cases where v1.0 had quoted a creator's single-platform number instead of their cross-platform footprint, and one creator whose published figure we could not reproduce at all. That case is off the page until we can explain it.

We anonymize flagged creators because we exist to serve the creator middle class, not to publicly shame it, and we never name a creator who sits on a paying brand's roster — that would leak the brand's shortlist. Some internally-flagged creators stay off this page entirely when the public evidence isn't strong enough to stand behind.

We'd rather publish six honest cases than a dozen we can't defend. Inflated accuracy claims are exactly what this benchmark exists to make impossible — including our own.

Methodology v1.1 · scoring pipeline v2.2 (seven agents + knockouts) · every figure is a unified cross-platform footprint score, dated, traceable to evidence in-product. See the full rubric.

See the score on a creator you care about

Run any creator through the same pipeline used in this benchmark — full brand-safety breakdown, with the evidence behind every signal.

Accuracy Benchmark — The Brand-Safety Score You Can Check | CreatorScore