---
name: "course_video_design"
description: "Design and plan an instructional video course: module structure, lesson runtimes, talking-head and screencast format choices, and retention techniques. Use when someone asks to create, plan, or fix an online video course or training series."
---

# Course Video Design

## Purpose

Turn a subject into a video course people actually finish. The design is grounded in published research on video engagement (Guo, Kim & Rubin 2014; PNAS 2023), curriculum chunking practice, and Bjork's desirable-difficulties techniques. The skill produces a module blueprint: modules, lessons with target runtimes, format per lesson, objectives, and checks for understanding.

## Workflow

1. **Define the outcome first.** Write 3-5 course-level goals in plain language: what the learner can do after finishing that they could not do before. Then break each goal into learning objectives using Bloom's revised taxonomy verbs: Remember (list, define), Understand (explain, describe), Apply (demonstrate, implement), Analyze (compare, examine), Evaluate (assess, justify), Create (design, build). Every objective must be assessable; if you cannot write a quiz or task that proves mastery, rewrite it.
2. **Group into modules.** Cluster objectives into 3-10 modules, each completable in one sitting. Keep the module pattern consistent: goal statement, 3-10 lessons, recap, one check or quiz. Sequence simple to complex, known to unknown, theory before practice. Cut any content that serves no objective.
3. **Size the lessons.** One lesson = one objective (or one small cluster). Research benchmark: videos under 6 minutes hold the highest engagement; by the 9-12 minute range engagement is roughly half. TechSmith's practical starting point is 6-12 minutes per lesson, adjusted to complexity. Never ship a single lesson over 12 minutes; split it instead. Shorter lessons are also cheaper to revise.
4. **Choose the format per lesson.** Use the decision rules:
   - Talking head (instructor on camera): introductions, motivation, opinions, stories, anything where the person matters. Guo et al.: videos with instructors beat slides-only, and alternating talking head with slides/code held the highest engagement. PNAS 2023: onscreen instructor presence improved post-test performance, though some viewers find it distracting, so keep the instructor at a fixed screen location.
   - Screencast: software walkthroughs, demos, code sessions. Narrate what you do as you do it; zoom to the area of action; keep the cursor visible and deliberate.
   - Drawing board (Khan-style tablet capture): derivations, diagrams, step-by-step reasoning. Guo et al. found this more engaging than PowerPoint or code screencasts.
   - Never read a classroom recording straight to video; that format tested poorly online.
   - One practical note from Guo et al.: an informal desk setting outperformed a professional studio, so do not stall production chasing polish.
5. **Build retention in, not on.** After every lesson, add a retrieval check: one question the learner must answer from memory before continuing. Space checks across modules so earlier material returns later, and interleave problem types instead of blocking them. Expect learners to find spaced, interleaved practice harder; it raises retention anyway.
6. **Script each lesson before recording.** Write scripts as short speakable lines, not essays. Lesson structure: hook (why this matters, 15-30 seconds), one core idea with one visual that carries it, one worked example, one-sentence summary, one check question. Cut sentences that need rereading.
7. **Record for the medium.** Camera at eye level, medium close-up with eyes near the top third and a little headroom, soft light on the face, uncluttered background, a close mic (lav or USB). Record in short sections so flubs cost seconds, not takes. Captions always, for mobile viewers and accessibility.

## Output Contract

When asked to design a course, produce:

- 3-5 course goals, then 2-5 measurable objectives per module.
- A module blueprint: module titles, lesson titles, target runtime per lesson (flag any over 12 minutes), format per lesson, and one check question per lesson. `bin/course_blueprint.py` generates and validates this from a lesson list.
- Per-lesson script outlines following the hook / one idea / worked example / summary / check structure.
- A retention map: which objectives get revisited in which later modules.

## Operating Rules

1. One lesson, one idea. If a lesson has two ideas, it is two lessons.
2. Runtime is a ceiling, not a target. Under 6 minutes when the idea allows it; 12 minutes is the hard cap, never a goal.
3. Format follows function. Talking head for the human parts, screencast for the screen parts, drawing board for reasoning; never default to whatever is easiest to record.
4. Every objective is testable or it is cut. Every lesson ends with a check the learner must answer from memory.
5. No confident unverified claims in any course content this skill helps produce: numbers, studies, quotes, and events are verified against a primary source or cut.