Case study

Audience emotion as a
leading indicator of engagement

Most analytics tells you what an audience did. We measure what they felt — and show how that feeling predicts the business outcome you care about. Below, the live model across a Formula 1 season as a worked example.

The model, live

Which emotions move engagement?

Ranked by standardized effect on engagement rate, controlled for channel. The same engine runs on any audience corpus — F1 is simply a vivid, public demo.

What Drives Engagementemotion → engagement rate
Loading model…
What Drives Engagement · Race by Race
Loading…
Why you can trust it

The caveats are the credibility.

Association, not naïve causation
Comments are a response to a video, so we report a leading-indicator association measured per viewer — never a causal promise.
Controlled for channel
A handful of large channels dominate reach. Channel is a fixed effect in the model, so we are not just rediscovering "the big channel is big."
Rates, not raw counts
We model likes- and comments-per-view (activation per viewer), because absolute counts mostly measure audience size.
Honest about confidence
Every effect ships with a 95% confidence interval and significance; non-significant effects are visibly faded, and the season-trend is the conservative, N-limited part.
See it across the season