A customer quote may arrive as a 40-minute interview, but the finished campaign needs a 20-second clip for LinkedIn, a silent loop for a landing page, a square version for paid social, and a still image for channels that cannot use video. Treating those as one export with different crops usually produces tiny captions, cut-off faces, and files nobody can confidently publish.
The useful unit of work is a governed testimonial package: one approved claim, several deliberately composed outputs, and enough context that a viewer can understand the clip without sound. TinyImage can export MP4 and WebM video from Figma and help keep the production work close to the layouts. Consent, claim approval, transcription, and channel upload checks still need human owners.
Lock the evidence before designing the clip
Start with the source recording and an approval record. Capture the speaker’s name, role, company, recording date, approved excerpt, permitted channels, territory, and expiry or withdrawal terms. Do not improve a hesitant quote into a stronger promise. If three sentences were recorded several minutes apart, do not edit them to imply one continuous statement without review.
Create a verbatim transcript with timecodes, then mark the smallest complete excerpt that still preserves meaning. A useful selection answers three questions:
- What problem was the customer facing?
- What changed after using the product or service?
- Is the result presented as their experience rather than a universal guarantee?
Keep the approval text beside the source link, not buried in a chat thread. Regulated or performance-based claims may need legal review even when the customer genuinely said them.
Design for the place where the clip will live
Define an output matrix before opening the timeline. For example:
| Placement | Composition | Audio assumption | Practical constraint |
|---|---|---|---|
| landing-page proof | 16:9 or container ratio | optional | fast first frame, controlled page weight |
| LinkedIn feed | 1:1 or 4:5 | often muted | large burned-in captions |
| story placement | 9:16 | often muted | keep identity and captions inside safe areas |
| sales follow-up | 16:9 | likely on | clear speaker identity and useful file name |
Do not simply center-crop the widescreen master. Recompose each ratio in Figma. Check that the speaker’s eyes remain comfortably placed, captions do not cover their mouth, and the attribution survives UI overlays used by the destination platform.
If the recording is low resolution, avoid enlarging it until facial detail becomes visibly soft. A framed treatment with brand color, quote text, and generous spacing is often more credible than pretending a small webcam capture is studio footage.
Make captions part of the layout
Burned-in captions are valuable when a channel starts video silently, but they should not be the only accessibility artifact when the platform supports a caption file. Use the verified transcript for both.
Keep each caption to a short, readable phrase. Use strong contrast, avoid placing text at the extreme bottom edge, and let captions remain on screen long enough to read. Preserve meaningful pauses; rapid word-by-word animation can make a sincere customer sound like an advertisement.
Names and roles also require care. Confirm current spelling and titles. If the customer only approved a first name or anonymous attribution, follow that agreement exactly. Never expose an email address, meeting title, participant list, or other recording-interface detail in the crop.
Build a clean export package
Use predictable names such as:
acme-testimonial-jordan-landing-16x9-v03.mp4acme-testimonial-jordan-social-4x5-v03.mp4acme-testimonial-jordan-story-9x16-v03.mp4acme-testimonial-jordan-poster-16x9-v03.webp
With TinyImage, export the selected Figma frames to the format required by the destination. MP4 is a practical delivery format for many channels; WebM may be useful for controlled web playback. Export a poster image separately so the web team is not forced to accept an arbitrary video frame.
This workflow differs from the release-note GIF and MP4 workflow, which demonstrates product changes, and the customer case-study asset workflow, which governs a broader editorial package. Testimonial video adds consent, spoken-claim, caption, and performance concerns.
Review the real files, not only the Figma frame
Play every exported file from beginning to end. Check the first frame, last frame, audio start, lip sync, caption timing, spelling, attribution, crop, and any abrupt cut. Inspect it at actual feed size on a phone, not only enlarged on a desktop.
For web delivery, measure the exported file and test it on a representative page and connection. Video can easily become the heaviest asset on a landing page. Decide whether it should load on demand, whether a poster should appear first, and whether motion should start automatically. Respect reduced-motion choices and do not rely on autoplay with sound.
Finally, upload one pilot to each destination. Platforms may recompress video, crop previews, obscure safe areas, or ignore the chosen poster. A local file that looks perfect is not proof of the published result.
Testimonial package sign-off
Before release, confirm:
- the excerpt is accurate, contextual, and approved for each channel;
- usage rights, attribution, and expiry are recorded;
- every aspect ratio was recomposed rather than blindly cropped;
- captions match the recording and remain readable on a phone;
- no private recording metadata appears;
- the output format, dimensions, duration, and file size fit the destination;
- poster images and accessible caption files are included where needed;
- exported files have been watched in full;
- pilot uploads have been checked after platform processing;
- there is a clear owner for takedown or replacement requests.
A testimonial earns trust because it feels specific and true. The production workflow should protect that truth all the way from the recording to the final compressed file.
Read as Markdown Related resource index Site index for AI agents
Pro Bundle