Skip to main content

Audience insights

The Audience report turns anonymous plays into people. It scores each identified viewer on attention, predicts their lifetime value, and flags who's about to churn — so you stop treating a first-time bouncer and a repeat buyer as the same "view".

Where to find it​

Open the Analytics workspace and select Audience. Viewers are identified by an email · cookie · UTM combination, so the depth of this report depends on how much identity your traffic carries.

Headline KPIs​

  • Identified viewers — People matched to an identity this period.
  • Avg attention — Mean attention score on a 0–100 scale.
  • Predicted LTV — Average forecast value per identified viewer.
  • Churn risk — Share of viewers flagged at risk, with the count.

The attention score​

Every viewer gets an attention score from 0 to 100 that blends how much they watch, how they interact (unmutes, scrubs, rewatches), and how consistently they return. On the distribution histogram, the top 10% sit at 92+, the median is around 71, and the bottom 25% fall below 42. Attention correlates with value: high-attention viewers are far more likely to convert and repeat.

Attention × LTV scatter​

The scatter plots each viewer by attention (x) and predicted LTV (y). Dot size reflects session count and color reflects segment. Two quadrants matter most: High attention · High LTV (your best people to clone in acquisition) and the Churn quadrant (low attention, low value). Filter to All, Converters, Returners, or At risk to isolate a group.

Where audience data comes from​

The report profiles viewers at the session level. For segment comparisons by device, country, or traffic source, use the Cohorts report and Traffic sources & filters to scope the workspace before you open Audience.

Top viewers table​

The Top viewers table lists identified people with geo, attention, predicted LTV, sessions, last seen, and a color-coded churn risk badge. The search field is Find viewer... and filters the table. The in-card Export button is still presentational; use the workspace toolbar's Export CSV for the report rows.

Churn risk and re-engagement​

A callout can surface viewers at risk of churning with a suggested action. Its action button is currently presentational and does not launch a campaign. Open Ask AI directly if you want help drafting a follow-up.

Best practices​

  • Acquire lookalikes of your High/High quadrant, not just more traffic. Attention-vs-LTV tells you who's actually worth cloning.
  • Act on churn early. The flag fires while there's still a relationship to save — wait too long and the recovery rate drops.
  • Read attention next to retention. A high average retention with low attention often means autoplay padding the numbers.

Frequently asked questions​

What is the attention score based on?

A 0–100 blend of watch depth, interactions (unmutes, scrubs, rewatches), and return frequency. The top 10% score 92+, the median is ~71, and the bottom 25% fall below 42.

How are viewers identified?

By an email · cookie · UTM combination. Anonymous sessions still count in aggregate reports, but per-viewer LTV and churn are strongest when identity is captured (for example via a lead gate).

Can I act on the churn-risk list?

The churn action and in-card Export controls are presentational. Find viewer... filters the table. Use the workspace toolbar's Export CSV for available Audience rows, or open Ask AI directly for help drafting a follow-up.