An engagement index is a single continuous score, usually normalised to 0–1, that summarises how much attention a learner is giving to content at a point in time. It is derived by combining several weaker signals rather than measured directly.
What it is
No single observable equals engagement, so an index fuses several: focus-adjusted dwell time, tab-switching, scroll behaviour, interaction latency and — where lawful and enabled — affective signals. Each contributes a weighted term, and the result is smoothed over a short window so that a momentary glance away does not register as disengagement.
The weighting is the whole design decision. An index that over-weights interaction will score a click-heavy quiz as more engaging than a genuinely absorbing video. One that over-weights dwell will reward slow readers and confused ones equally.
Why it matters
An index is what makes engagement comparable — across sections, cohorts and content versions. Without one you have a pile of incommensurable signals; with one you can say module A held attention better than module B and mean something precise.
A common confusion
An engagement index is not a measure of learning. Someone can be fully engaged with a badly wrong explanation. Pair it with assessment data before drawing conclusions about effectiveness.
Related
See also content effectiveness, time on task and engagement heatmap. See the full glossary for the rest, or the learner engagement guide for the wider context.
See where your content loses people
Book a walkthrough and we will show you the engagement data on your own content.