In-video search is a feature that indexes the transcript of a video so learners can type a keyword or phrase and receive a list of timestamps where that term appears, with each result linking directly to that position in the player. It transforms a passive video into a navigable reference document — particularly valuable in long-form technical or compliance training where a learner returning to review a specific concept does not want to scrub through the whole video. The search implementation typically uses a full-text search index built from the word-level timestamped transcript; Elasticsearch, Algolia, or a simple inverted index on a relational database are common backend choices depending on the scale of the catalog. Results are ranked by relevance and can be grouped by chapter so the learner sees not just the raw timestamp but the section context. In-video search depends directly on transcript quality: if ASR has misrecognised a technical term, that term is unsearchable until the transcript is corrected, which makes human review of ASR output a prerequisite rather than a nice-to-have. At the xAPI level, search queries and the resulting seeks can be tracked as custom statements, giving learning analytics data on which terms learners look up most — a useful signal for identifying content gaps or confusing segments. In-video search is closely related to chaptering: chapters provide the section labels shown in search results, and chapter boundaries help the indexer cluster nearby hits so the learner lands in a coherent segment rather than mid-sentence.
