Methodology — Veltrace

The score

One number. Four signals.

Every creator on Veltrace receives a Momentum Score between 0 and 100. It is not a vanity metric — it does not measure follower count or total views. It measures the rate and quality of growth: are things accelerating or decelerating right now?

The score is a weighted composite of four components, each capturing a different dimension of creator performance.

Score components

What goes into the score.

Each component captures a different dimension of performance. Combined, they produce a score that reflects the full picture of a creator's momentum.

Growth Acceleration

Measures the rate of change in subscriber or follower growth — not the raw count, but whether that count is accelerating week-on-week. A creator doubling their growth rate scores higher than one with more followers but slowing growth.

Engagement Stability

Tracks average engagement rate over a rolling 30-day window and penalises volatility. Consistent, high engagement across multiple uploads scores better than one viral video surrounded by silence.

Posting Consistency

Scores how regularly a creator publishes content. Algorithms reward consistency, and so does this component. A creator who posts on a predictable cadence scores higher than one who posts in bursts.

Cross-platform Activity

Measures whether the creator is active across multiple platforms — primarily YouTube and Twitch. Creators building simultaneous audiences on more than one platform tend to be more resilient and signal greater long-term ambition.

Conviction rating

How confident is the signal?

Every score comes with a conviction rating — HIGH, MEDIUM, or LOW — that reflects how much data backs it up and how consistently strong the underlying signals are. A creator can have a score of 75 with LOW conviction if that score is based on sparse or erratic data.

HIGH

Consistent across all four components

All four signals are elevated. The score is data-dense and unlikely to be a flash in the pan.

MEDIUM

Strong in most, not all areas

Two or three components are strong but one or more is lagging. Worth watching — could resolve upward or downward within the next scoring cycle.

LOW

Inconsistent or limited data

The score is based on limited history, high variance, or a single strong component. Treat with caution. More data is needed before acting on this signal.

Drift detection

When a creator is migrating platforms.

↑ Drift Detected Cross-platform signal

Drift detection identifies creators who appear to be shifting their primary audience from one platform to another — most commonly from YouTube to Twitch, or vice versa. This is a significant early signal for talent teams, because platform migration often precedes a major spike in total reach.

Drift is flagged when a creator shows sustained growth on a secondary platform at a rate that outpaces their primary platform, combined with rising cross-platform activity scores. It does not require a drop on the primary platform — growth on both with acceleration on the secondary is enough to trigger the flag.

Creators with drift detected are worth prioritising. They are typically in an expansion phase and building leverage across multiple audiences simultaneously.

Breakout probability

The chance of breaking through.

Breakout Probability 0 – 100%

Breakout probability is a score representing the likelihood that a creator is approaching a step-change in audience size. It is a leading indicator, not a lagging one — designed to surface signal before it becomes obvious.

Note on timing: Breakout probability is a leading indicator, not a guarantee. It identifies creators who are showing the conditions that historically precede breakouts — not creators who have already broken out. By the time a breakout is obvious to the public, the opportunity for early positioning has typically passed.

Data & cadence

When scores are updated.

All scores are recalculated once every 24 hours, at midnight UTC. Data is pulled directly from YouTube and Twitch APIs. Each scoring run processes every tracked creator independently — there is no sampling or extrapolation.

Score history is retained for 30 days and is available on paid plans. The full trend line, component breakdown, and historical conviction ratings are all visible on individual creator profile pages.