콘텐츠로 이동

title: In-product video (onboarding, help, explainers) applies_to: [video, in-product, onboarding, help] version: 1.0.0 last_updated: 2026-05 stability: stable


In-product video

Video inside the product itself: onboarding tutorials, help videos, feature explainers, embedded demos. Different rules from marketing video — these viewers already chose your product; they want help, not seduction.

Read video-fundamentals.md first.

Three categories

1. Onboarding video (first-run)

Shown to new users during signup or first session.

Spec Target
Length 30-90 seconds
Format 16:9 desktop / 9:16 mobile
Click Skippable; optional
Audio Optional (often muted with captions)
Goal Get user to first value moment

Structure:

0:00-0:05  Welcome (warm; 2-3 word headline)
0:05-0:30  What this product does (1-2 sentences max in caption)
0:30-0:60  How to do the first thing (specific action)
0:60-0:90  What's next (CTA: try it)

Don't over-explain. Onboarding video is supplemental; the actual product walkthrough does the teaching.

2. Help video (in-context)

Shown when user is stuck or asks for help. Embedded in help articles, settings pages, error states.

Spec Target
Length 30-120 seconds (focused on one task)
Format 16:9 mostly; 9:16 for mobile-only flows
Click Always playable; controls visible
Audio Voiceover preferred; captions mandatory
Goal User completes the task they're stuck on

Structure:

0:00-0:05  What this video shows
0:05-end   Task steps in order, each with brief voiceover
0:end      "Try it now" or close

Pace: brisk for power users; deliberate for beginners. Can produce 2 versions.

3. Feature explainer / changelog video

Promotes a new feature or update. Lives in product changelog, release notes, in-app announcements.

Spec Target
Length 15-60 seconds
Format 16:9
Click Click-to-play; auto-play loop variant for hero
Audio Optional
Goal User understands new capability + tries it

Structure:

0:00-0:05  What's new (feature name + benefit)
0:05-0:30  How it works (short demo)
0:30-end   CTA (try it, learn more)

Production approach

Screen recording vs filmed

Most in-product video is screen-recorded (capturing actual UI): - Tools: Loom, ScreenFlow, OBS, Riverside, CleanShot X. - Pros: cheap, fast, accurate. - Cons: Less polished; cursor / UI quirks visible.

Some is filmed (actor + product on phone / laptop): - Tools: full production setup. - Pros: more emotional, brand-feeling. - Cons: expensive, dates fast (UI changes → reshoot).

Hybrid: filmed cutaways + screen-recorded core. Best of both.

Screen recording tips

  • Resolution: record at 2× display resolution; deliver at 1×.
  • Cursor: highlight (yellow circle, color trail) — visible at small sizes.
  • Click feedback: animate cursor click (Loom does this automatically).
  • Zoom on detail: zoom into the relevant UI region; don't show whole screen at small video size.
  • Hide irrelevant UI: minimize taskbar, hide dock, clean desktop.
  • Realistic data: use representative names, not "Test User #1".
  • Korean data: use Korean names / addresses for KR audiences (don't show "John Doe" in a Korean app).

Voiceover for in-product video

  • Korean primary for KR audience.
  • Pace: slightly slower than marketing — viewers are following along.
  • Tone: helpful, neutral; not salesy.
  • Female / male voice: pick one and consistent across the product video library.

For AI voice: ElevenLabs / 클로바보이스 produce decent Korean output. Always review for pronunciation of brand / product names.

No-voice variant

Some in-product video has no voice — just visual + on-screen text.

Pros Cons
Faster production (no recording) Limited information density
Localizable easily (text only) Can feel sterile
Plays in muted contexts Requires clear visual direction

For multi-region products: text-only might be better than recording 5 language voiceovers.

Embedded video player UX

When video is embedded in product UI:

┌──────────────────────────────────────┐
│  [Video thumbnail or first frame]    │
│                                      │
│        ▶ Play                        │
│                                      │
│  Duration: 0:45    [CC] [Speed]      │
└──────────────────────────────────────┘

Controls users expect: - Play / pause. - Progress bar (seekable). - Volume (or muted-by-default toggle). - Captions toggle (default ON for product video). - Speed control (1x, 1.25x, 1.5x, 2x — power users skim). - Fullscreen. - Close / dismiss (if modal-embedded).

For minimal UX: at least play/pause + caption toggle.

See examples/component-video-player.md for spec.

Lazy-loading video in-product

In-product videos shouldn't slow down the app:

// Show poster only; load video on click or on visibility
<HelpVideoEmbed
  poster="/help-videos/onboarding-poster.jpg"
  src="/help-videos/onboarding.mp4"
  loadOn="click"   // or "visible" via IntersectionObserver
/>

For modal-embedded videos: lazy-load when modal opens, not when product mounts.

Captions for in-product video

Captions mandatory for any in-product video with speech. Reasons:

  • Accessibility (deaf / hard-of-hearing users).
  • Sound-off viewing (open offices, public, late night).
  • Non-native speakers (Korean → English content).
  • Indexing (caption text searchable).

Default: captions ON for product video. User can toggle off if preferred (remember preference).

Variable speed for power users

Allow 1x, 1.25x, 1.5x, 2x. Power users skim help videos at 1.5-2x. Don't force them to wait.

Localization

For multi-region products:

  • English original + Korean / Japanese / etc. caption tracks.
  • OR separate videos per language (different voiceover, captions).
  • OR no-voice + localized text overlay (cheapest scaling).

Choose based on: - Audience size per region. - Voice / brand consistency requirements. - Budget.

Korean in-product video

  • 자막 ON by default.
  • Korean voiceover for KR audience.
  • Reach pace: slightly slower than marketing video.
  • Brand voice: 해요체 (~해요) for friendly products; 합쇼체 for B2B / financial.
  • Korean data in screen recordings (Korean names, KRW currency).

For Korean fintech in-product video: 해요체, female voice (Toss-style), captioned, brand-color CTA at end.

Updating videos when product changes

In-product videos date fast. UI changes → video shows old UI → confused user.

Strategies: - Stylized animation (After Effects mockup) instead of screen recording — doesn't date with UI changes. - Periodic refresh schedule (every 6 months: review + reshoot). - Versioned URLs so old videos can be archived without breaking links. - Track which UI flows are video-documented so you know what to update.

For SaaS with rapid UI iteration: animation-based explainer videos > screen recordings.

Performance considerations

  • Compress aggressively — in-product video doesn't need 4K.
  • Provide poster — first paint shows poster, video loads behind.
  • HLS / adaptive bitrate for longer videos (5+ minutes) — players adjust quality based on connection.
  • CDN delivery — Cloudflare, Fastly, AWS CloudFront, or platform native (YouTube embed for some cases).

Accessibility

  • Captions: required.
  • Transcripts: post the full transcript for help articles (text-searchable, screen-reader-friendly).
  • Audio descriptions: for video where critical info is visual-only.
  • Don't auto-play with sound: users with sensitive ears, in shared spaces, abandon.
  • Keyboard controls: spacebar to play/pause; M to mute; arrows to seek.
  • Pause control: for any auto-play.

Common in-product video mistakes

  • Too long — users skip.
  • Outdated UI in screen recording — confused users.
  • No captions — accessibility fail + 80% miss the message.
  • Auto-play with sound — disorienting; abandoned.
  • Video where text would do — reading is faster for many tasks.
  • Salesy tone — these are existing users, not prospects.
  • Generic stock footage — feels off-brand for a help video.

When NOT to use video

Situation Better alternative
Linear text task ("how to delete") Text + screenshot
Search / scan-friendly content Text article (skimmable)
Multi-step process where users go at their own pace Annotated screenshots
Reference (looked up later) Text + animated GIF
Common error explanation Inline help in UI

Use video when: - Spatial / motion is hard to describe in text (drag-and-drop, gesture). - Emotional moment (welcome, celebration). - Walkthrough where seeing > reading.

Don't

  • Don't ship a 5-minute onboarding video. Cut to 60 seconds.
  • Don't skip captions. Mandatory.
  • Don't auto-play with sound.
  • Don't show outdated UI in screen recording. Update when product updates.
  • Don't use video for everything. Text + screenshots are often better.
  • Don't put video in time-critical flows (checkout, payment) — slows users.
  • Don't lock features behind "watch the video first" — let users skip.

Cross-reference