Chunking in learning theory derives from George Miller's foundational 1956 paper observing that human working memory holds roughly seven plus or minus two items at once — but that items can be grouped into higher-order "chunks" that each count as a single unit, effectively expanding usable capacity. In instructional design for video, chunking means grouping related concepts into discrete, named segments rather than presenting them as an undifferentiated stream. A ten-minute video on database indexing, for example, might chunk into: "what an index is," "how B-tree indexes work," and "when not to index" — each segment a self-contained unit with a clear label that signals its scope. This structure serves cognitive load management by giving the learner's mind a schema to attach incoming information to, reducing the effort of integration. In video production, chunking maps directly to chapter markers and segment titles, which are also the navigational anchors that let a returning learner jump to the section they need rather than seeking through an undivided file. The connection to microlearning is direct: microlearning takes chunking to its logical endpoint by making each chunk an independent deliverable rather than a subdivision of a longer video. From a platform perspective, good chunking also improves analytics: per-segment drop-off data tells a content team which chunks lose attention, giving them specific targets for re-editing rather than a vague instruction to "make it shorter." The practical trap is over-chunking: splitting content into so many tiny fragments that the learner loses the thread of the overall argument and has to expend effort reconstructing the narrative structure.
