The science

What is influence and can it be measured?

Every brand-creator match in the creator economy rests on a word no one has defined. We've spent a decade trying to. This is what we've found.

Influence isn't magic. It is measurable.

Conversion only happens when a brand, a creator, and an audience all align.

This is how we measure that alignment.

§3.2 · The framework

Every collaboration sits on a triangle.

We predict the expected conversion rate for every brand–creator pair. Score, ranking, and pricing all fall out of that one number.

Brand, creator, audience. Conversion only emerges when all three edges are sufficiently strong. Our research formalises this as the Brand–Creator–Audience Triangle, and the Trudy Score is built edge by edge.

Expected conversion rate
E[CVRb,i] = f( brand–creator fit, audience–brand fit, creator–audience influence )

For brand b and creator i. Predicted CVR feeds predicted ROAS; accept the collaboration when predicted ROAS clears the brand's margin-implied threshold.

"Conversion emerges only when all edges are sufficiently strong; weakness in any single edge limits performance."

Trudy research programme, 2025–

BRAND AUDIENCE CREATOR AUDIENCE–BRAND FIT BRAND–CREATOR FIT CREATOR–AUDIENCE INFLUENCE

Figure 1 · Brand–Creator–Audience triangle

Edge by edge

Each edge is a separate measurement problem, surfaced in the product as one of the three Trudy sub-scores.

Brand–Creator fit
edge 1
Is the endorsement credible and congruent?

Topical overlap between the creator's content and the brand's category, stylistic and narrative compatibility, value and persona alignment, plus brand-safety and creator-safety signals. Does this pairing feel authentic, or bolted on?

In the product Content Fit
Audience–Brand fit
edge 2
Is the audience the brand's target market?

Demand-side fit. Audience composition against the brand's target market: demographics, geography, language, observable proxies for purchasing power and willingness to pay, plus preference and lifestyle alignment where inferable.

In the product Audience Fit
Creator–Audience influence
edge 3
Conditional on exposure, can the creator move them?

Attention and persuasion. Retention around sponsored segments, engagement quality, and the psychometric profiles of creator and audience that proxy for trust and susceptibility. Holding product and audience constant, how much incremental conversion does the creator generate?

In the product Psychometric Fit
Training data

The world's largest dataset linking creator psychology to performance.

Every other tool in this category runs on one customer's data; one brand's history, one platform's view, locked inside a walled garden.

Trudy is built differently. We're the data layer underneath the creator economy: the only place where campaign outcomes from 800+ leading brands are pooled, harmonised, and labelled by the same scientific framework.

dataset · trudy_global_v4 last_refresh · 2026-05-04
brands
800+
contributing outcomes
campaigns
230k+
labelled outcomes
creators
10M+
profiled across markets
history
10yr+
operator-collected
signals
160+
per creator analysed
retrain
weekly
rolling across pool

Your data alone is biased and siloed. Industry data alone is generic. Trudy solves both by having the only standardised data pool of its kind, with your brand's signal on top.

Research lineage

Our methodology lives in a research programme, not a deck.

Trudy's scoring sits on a body of research, not a single paper. Three streams feed the model: a dedicated research programme via GCDIR , the world's largest dataset linking creator psychology to campaign performance (10+ years, 230,000+ tracked campaigns, 10M+ creators profiled), and applied work that draws on existing literature in psychometrics, psycholinguistics, and brand archetype.

What the programme covers
  • · Psychometric and psycholinguistic signals in creator content
  • · Big-five trait research applied to creator–audience congruence
  • · Brand-archetype literature mapped to creator profiles
  • · Held-out validation against a decade of agency campaign outcomes
  • · Causal inference on the contribution of each triangle edge to CVR
Selected outputs
Birds of a Feather or Digital Echo Chambers: Congruency in Influencer Marketing · live working paper, 2025– · GCDIR research programme
Ongoing programme studies · psychometric signals, big-five traits, brand archetype, audience congruence
Visit gcdir.org →
Honest limits

What we don't claim.

The Trudy Score is a prediction, not a guarantee. A creator with a 90 can still underperform on a single campaign, and a 60 can break out. What the score does is shift selection from judgement to evidence, and make that evidence legible to the people who have to defend the pick.

The model does best on creators with a meaningful content history (we need roughly 50 hours of their content to build a reliable psychometric profile). It does worst on brand-new creators and on categories where our training data is thin. We flag both cases in the product.

The model is retrained, not static. Every campaign outcome that lands in the system sharpens the prediction for the next one. We publish methodology updates and retrain cadence through GCDIR.

Get involved

Join the research programme.

We work with academic researchers and practitioners across the creator economy. Whether you're studying digital influence, building creator tools, or running campaigns at scale, there's a seat at the table.