A demonstration teaches a decision path, not hand movement
A learner needs to know what the task is for, what starting conditions matter, which cue drives each choice, and how to verify the result. Recording a cursor through a perfect sequence may show location while hiding judgment. Begin with the user situation and desired outcome. At every consequential step, state what is observed, which options exist, why one is selected, and what would make another option appropriate.
CMU’s active-learning guidance includes worked examples and opportunities for learners to explain or practice. A demonstration functions as a worked example when it exposes reasoning, not merely the final sequence. Watching remains insufficient evidence of performance. The design must create pauses where learners predict, attempt, or diagnose before the presenter reveals the next move.
Evidence: Carnegie Mellon University Eberly Center
Context determines whether the procedure can transfer
State the version, permissions, sample data, prior setup, audience, and constraints before beginning. Explain what is simplified for teaching and what would differ in live work. A tutorial made with an administrator account may fail for ordinary users; a clean dataset may hide validation; a fast connection may hide waiting states. Context is part of the instruction, not production clutter.
How People Learn II emphasizes prior knowledge and context. That supports surfacing assumptions and varying later practice so learners can recognize when the demonstrated procedure applies. It does not mean every edge case belongs in one lesson. Select the conditions that change the decision and link to bounded prerequisites or variants.
Evidence: Carnegie Mellon University Eberly Center; National Academies of Sciences, Engineering, and Medicine
Failure and recovery make the mental model visible
Include one likely error, interruption, or ambiguous state. Pause before correcting it and ask what evidence distinguishes the causes. Then show the safest recovery, what data or work might be lost, and when to stop or escalate. A flawless demonstration can create false confidence because learners encounter their first failure alone, without a model for diagnosis.
Recovery should not normalize unsafe experimentation. Do not stage actions that risk real accounts, confidential data, money, health, or irreversible changes without appropriate safeguards. Use a sandbox, duplicate, mock data, or still image where possible. Label constructed failures and do not claim they were customer incidents or TenMultigure tests.
- Likely symptom shown
- Competing causes named
- Safe observation selected
- Reversible recovery demonstrated
- Stop or escalation boundary stated
Evidence: Carnegie Mellon University Eberly Center; National Academies of Sciences, Engineering, and Medicine
Accessibility must survive the delivery format
W3C’s media accessibility guidance covers captions, transcripts, audio description, and accessible media players. Captions should represent meaningful speech and sounds; transcripts provide an alternative and searchable reference; essential visual information needs verbal or text description. Avoid relying on color, cursor motion, or tiny interface text as the only cue. Give learners control over playback and pacing.
CAST’s UDL Guidelines support multiple means of engagement, representation, and action or expression. Use a text procedure or annotated steps alongside video when appropriate, but keep alternatives synchronized with the same version. Accessibility is not an optional polish stage. If the medium cannot convey the target without an irrelevant barrier, choose another medium or redesign the task.
Evidence: CAST; W3C Web Accessibility Initiative
Artifact: the context–decision–recovery script
Write one row per scene with: learner goal, starting state, visible cue, decision, action, narration, on-screen emphasis, likely failure, recovery, verification, and accessibility asset. Add the time at which learners pause for an attempt and the prerequisite link if they cannot proceed. This script separates instructional necessity from decorative production.
During review, remove any scene that neither changes a decision nor enables verification. Check that the final demonstration verifies the outcome rather than stopping after the last click. Then ask a reviewer unfamiliar with the production to reconstruct the starting conditions and a recovery branch. Missing answers reveal instructional gaps before recording.
Starting state reproducible.
Reasoning appears before action.
Likely failure has a safe recovery.
Captions, transcript, and visual descriptions planned.
Final outcome can be independently verified.
Evidence: Carnegie Mellon University Eberly Center; CAST; W3C Web Accessibility Initiative
A polished demonstration still needs learner evidence
Completion, watch time, and positive comments do not establish independent use. Ask learners to perform a parallel task with changed surface details, record assistance, and verify the result. Preserve failures and access barriers. A small observation can expose design problems but cannot prove general effectiveness or durable transfer.
Start with a three-to-five-minute segment containing one decision and one recovery, then test the script and accessibility assets before producing a long tutorial. This duration is illustrative, not a universal evidence-based rule. Recheck source, platform, version, and instruction by 2027-02-10 or sooner when the interface or task changes.
Evidence: Carnegie Mellon University Eberly Center; W3C Web Accessibility Initiative; National Academies of Sciences, Engineering, and Medicine
Sources and further reading
These references informed this article. A source supports a claim; it does not imply endorsement of TenMultigure or any future product reference.
- Concrete Strategies for Active LearningCarnegie Mellon University Eberly Center · Accessed August 10, 2026
Provides official active-learning and worked-example techniques used to turn passive viewing into prediction, explanation, attempt, and feedback around a real task.
- CAST Universal Design for Learning GuidelinesCAST · Accessed August 10, 2026
Supports varied and accessible ways to engage with and demonstrate the task, informing pacing controls and alternatives without changing the substantive outcome.
- Making Audio and Video Media AccessibleW3C Web Accessibility Initiative · Accessed August 10, 2026
Provides official media accessibility requirements and practices for captions, transcripts, visual descriptions, and accessible playback in demonstration assets.
- How People Learn II: Learners, Contexts, and CulturesNational Academies of Sciences, Engineering, and Medicine · Accessed August 10, 2026
Adds independent context on prior knowledge, feedback, and transfer, supporting explicit starting conditions and changed-context learner verification.
Reviewed by TenMultigure Editorial Team. See an error or a source that has changed? Tell the editorial team.
Review method: AI-assisted desk research with editorial checks. Reviewed ; next scheduled review . Rebuilt TM-231 around visible decision cues, starting context, safe failure recovery, synchronized accessibility assets, a scene-level script, and independent learner verification.