How we made the JustBite customer story

How we made the JustBite customer story — HappyClient playbook

happyclient.

TEAM PLAYBOOK

How we made the
JustBite customer story.

A repeatable workflow — from 5 raw phone clips to a brand-ready testimonial in six formats.

THE RESULT

Five raw clips in.
A polished story out.

One 85-second customer testimonial — captioned, colour-graded, scored, and exported in 6 ready-to-post cuts.

JustBite-branded motion graphics

Word-synced captions

B-roll matched to the script

Original lo-fi soundtrack

Final — landscape, framed captions

HappyClient · JustBite customer-story playbook

THE STARTING POINT

What the shoot gave us.
Five questions, one take each.

clip_1
Name, job title & company

clip_2
Which JustBite service & why

clip_3
What they use it for

clip_4
What the team says

clip_5
Why they'd recommend it

The filenames already map to the script — that becomes our edit map.

HappyClient · JustBite customer-story playbook

THE WORKFLOW

Eight steps,
every single time.

Transcribe

Word-level timestamps

Select & trim

Best take, cut filler

Map the edit

edit.json: clip → time

Stitch & grade

ffmpeg, colour

Motion graphics

Remotion + brand

B-roll

Match the words

Music

Lo-fi bed

Export

6 formats

HappyClient · JustBite customer-story playbook

THE TOOLKIT

Four tools.
All scriptable, all repeatable.

Whisper

Transcription with word-level timestamps (ElevenLabs in the recipe; local Whisper here).

ffmpeg

Cut, trim, colour-grade, normalise audio, encode every format.

Remotion

React-based motion graphics — captions, lower-thirds, cards, b-roll, rendered to video.

Audio

Original royalty-free lo-fi beat, mixed under the voice with steady ducking.

HappyClient · JustBite customer-story playbook

STEP 1 — TRANSCRIBE

The transcript is the map.
Everything keys off the words.

0.75s  My name is Rebeca and I'm the
          Workplace Experience Manager
          in our Barcelona Hub at Preply.

0.69s  We've been working with JustBite
          for over a year now…

Why it matters

• Word timestamps drive the caption timing.

• They tell us exactly where each sentence starts and ends — so we trim cleanly.

• They let us drop b-roll on the right word (e.g. “wholesale portal”).

• Proper nouns get fixed once, here (Preply, JustBite).

HappyClient · JustBite customer-story playbook

STEP 2 — SELECT & TRIM

Keep the best words.
Lose the dead air.

Latest take usually wins

People warm up — the last take is normally the cleanest.

Fewest “ums”

Score takes by filler words; pick the smoothest.

Cut the warm-up “Hey [name]”

Drop any greeting before the real sentence starts.

Trim the silence

Top & tail each clip to ~0.2s before the first word.

This shoot: one take per scene, no warm-ups — so we just trimmed silence.

HappyClient · JustBite customer-story playbook

STEP 3 — MAP THE EDIT

How we link the footage:
one JSON edit map.

{
  "scene": 2,  "label": "The service",
  "source_clip": "clip_2…mp4",
  "in": 0.49,  "out": 15.75,
  "transcript": "We’ve been working
     with JustBite for over a year…",
  "broll": [
    "video: wholesale_portal @4.2-9.2s",
    "image: office @9.7-13.8s" ]
}

One row per scene

• Which clip, and the exact in/out time range to use.

• The transcript, so captions and b-roll stay in sync.

• Which b-roll covers which seconds of the voice-over.

• Edit this file → re-render. No timeline scrubbing.

HappyClient · JustBite customer-story playbook

STEP 4 — COLOUR GRADE

Un-mute the footage.
Pick a look, bake it in.

Original → Natural → Warm Pop → Pink Punch → Cinematic

We chose Warm Pop — +contrast, +saturation, a touch of warmth — then upscaled 720p → 1080p and loudness-normalised the audio so every clip sits at the same level.

HappyClient · JustBite customer-story playbook

STEP 5 — MOTION GRAPHICS

Brand it in Remotion.
Captions, names, question cards.

Lower-third + captions

Question cards

Branded outro

Captions reveal word-by-word (active word in JustBite orange). All graphics read from the same edit map — change the data, re-render.

HappyClient · JustBite customer-story playbook

STEP 6 — B-ROLL

Show what she says.
B-roll, timed to the word.

“wholesale portal”

→ desktop portal screen-recording

“easy to order / keep track”

→ mobile portal in a phone frame

“best-sellers / drinks / variety”

→ product hero shots

“experience in our hubs”

→ office / team lifestyle

The in/out seconds live in the same edit map as the captions.

HappyClient · JustBite customer-story playbook

STEP 7 — MUSIC

An original lo-fi bed.
Low under the voice, up on the cards.

Synthesised, royalty-free

82-BPM lo-fi hip-hop — no licensing, made from scratch.

Steady, not pumpy

A level envelope (not heavy ducking) keeps it stable.

Knows its place

~−25 LUFS under speech, lifts to ~−18 on the question cards.

HappyClient · JustBite customer-story playbook

STEP 8 — EXPORT

One edit.
Six ready-to-post cuts.

Landscape 16:9

Square 1:1

Vertical 9:16

× two caption styles · framed (captions beside) and overlay (captions on the video) = 6 deliverables

HappyClient · JustBite customer-story playbook

TWO CAPTION STYLES

Same story.
Framed, or straight on the video.

Framed — captions in their own zone

Overlay — captions on the full-bleed video

Both come from the same Remotion project — just a different layout flag.

HappyClient · JustBite customer-story playbook

WATCH IT

Press play.

▶

Framed — landscape

▶

Overlay — landscape

Six finished cuts ship alongside this deck (landscape / square / vertical × framed / overlay).

HappyClient · JustBite customer-story playbook

THE PROMPTS · 1 OF 2

The one that starts it all.
Your copy-paste kickoff.

“I'm processing the recordings in @<folder>; the script is @<folder>/<script>.md. Run transcription on them, then stitch the best shots into one clip. Notes: multiple takes per scene — the best are usually the last with the fewest ‘ums’; I re-shot scene 1 at the end; cut any opening ‘Hey [name]’. Create a JSON showing, per scene, the clip and time ranges used. Then assemble with ffmpeg. Orchestrate it. /goal don't stop until there's a final video.”

Why it works: it states the rules (take selection, trims), the artifact (the JSON edit map), the tool (ffmpeg) and a clear finish line.

HappyClient · JustBite customer-story playbook

THE PROMPTS · 2 OF 2

Then you just talk.
Each tweak was one sentence.

“Use the JustBite branding, not ours — and add b-roll from this folder.”

“Keep the motion graphics on screen a little longer.”

“Add slow lo-fi hip-hop in the background — softer, and more stable.”

“Remove the question badge and the doodles; make the question bigger.”

“Remove the intro.”

“Cinematic subtitles beside the image; also make square + vertical (stories).”

“New version with the subtitles on top of the video.”

Small, plain-language asks. The pipeline re-runs itself each time.

HappyClient · JustBite customer-story playbook

YOUR TURN — THE RECIPE

Five moves to your own cut.
Same recipe, any customer.

Drop the clips in a folder

Name files by question. Add a /b-roll folder if you have it.

Run the kickoff prompt

Use the Personal Video Editor skill + the prompt on the previous slide.

Approve the grade

Claude shows options — pick one.

Point to the brand

Share the brand sheet + logo; say which brand to use.

Ask for the formats

“Framed + overlay, in landscape, square and vertical.”

HappyClient · JustBite customer-story playbook

WHAT MADE IT GOOD

Small things,
big difference.

Fix names once

Correct proper nouns in the transcript before anything renders.

Let cards breathe

A beat longer on screen reads as premium, not slow.

B-roll = the words

Match what's shown to what's said, second by second.

Music serves the voice

Background means background. Duck it; lift it only in the gaps.

One source of truth

The edit map drives captions, b-roll and every format.

Render small first

Check stills before the full render — catch issues cheaply.

HappyClient · JustBite customer-story playbook

happyclient.

Your turn.
Same recipe. Any customer story.

Folder of clips in. Six ready-to-post cuts out.

1 / 20

↑ ↓ / scroll to navigate