The dashboard says “sample data.” The voiceover says the product has tripled customer output. Those two messages do not explain the same thing. Replacing customer names with fictional ones does not establish the performance claim beside them.
For overseas teams commissioning SaaS or connected-product app videos with Chinese developers and production partners, demonstration data deserves its own brief. It must support a real product workflow, remain consistent across screens and stay separate from evidence of customer outcomes.
Easier recording leaves the data decision intact
Adobe’s “Record the screen and webcam footage in timeline,” updated October 28, 2026, describes recording screen, webcam and voiceover separately or together in Firefly Timeline. The page identifies Timeline as a beta feature.
This establishes a documented capability and product status, not industry adoption, lower project costs or better outcomes. ONCE’s practical inference is that capturing a demonstration can become more integrated while responsibility for what its numbers mean remains with the people preparing it.
The following is an evergreen production method supported by current material. It is not a feature summary or a report of an ONCE client test.
Distinguish three kinds of number
Fictional demonstration data helps explain a workflow. An observation from a real environment records what happened under particular conditions. A performance claim, such as reduced costs or improved productivity, needs its own supporting material and approved scope.
The same percentage can appear in all three contexts. A fictional dashboard showing 30% growth does not become evidence of actual growth because the software displayed it during a real screen recording. Conversely, an authorized actual result still needs its relevant date, scope and conditions when it is turned into motion graphics.
A sample-data label can explain that a company or its records are invented. It should not be used to excuse unsupported customer-success narration. If the task is teaching viewers to read a report, explain the report. If it is communicating an outcome, prepare the outcome evidence separately.
ONCE’s data-story check
For each important demonstration, write a short account of its starting state, the change performed and the resulting state. This is ONCE’s planning method, not software certification or a claim that the sample represents a market.
Prepare records for a specific question
Choose the task before populating the screen. A demonstration of filtering open support tickets needs meaningful status, owner or date differences. Filling a page with random rows may look busy without helping the viewer understand the operation.
Define the relationships the camera will show: what the total includes, which filter is active and the period behind a chart. You do not need a vast invented database. You do need the visible relationships to make sense.
Follow the change across screens
Consider a wholly invented example: 20 tickets are open, and an action marks five as completed. If the expected remainder is 15, check the list, counter and active filter against that result. If the displayed number instead counts all historical tickets, an unchanged total may be correct. The narration must describe the same measure.
Do not splice unrelated before-and-after states together just to make a graph more dramatic. If a sequence uses a prepared result, prototype animation or manually staged screen, have the product owner confirm what capability it represents and communicate its nature appropriately. Do not present it as an unassisted live output.
Read the conclusion at its actual scale
“You can now find these five completed records” describes a visible action. “Every team becomes more efficient” makes a much larger assertion. Link each concluding line to what the demonstration shows. A broader promise requires additional support and approval.
Analytics, marketing and financial software need particular care with amounts, percentages and comparisons that can resemble business results. Where genuine evidence is unavailable, demonstrate states, sorting or process rather than inventing larger returns.
Review more than the names
Check whether fictional identities have been mixed with real business material. Order details, avatars, notes, document previews and notifications may remain after a name has been changed. The team authorized to handle the information should prepare an approved demonstration environment and review the recording scope. We have not inspected a customer account.
Then review the picture without narration: do the starting records, action and result agree? Listen separately: does an explanation of a feature turn into a claim that customers have achieved a return? These checks address different questions; a sample-data corner label cannot answer both.
Where a visible explanation is needed, keep it inside the actual delivered frame. For localized versions, review units, currency, dates and what a percentage refers to. Translated buttons alone do not establish unchanged meaning.
Include data preparation in the production scope
The product team confirms paths and numerical logic. The brand approves external statements and genuine outcome evidence. Production turns those inputs into readable images and accurate narration. If the supplier creates the sample dataset, specify fields, scenarios, reset arrangements and review scope rather than improvising rows during capture.
Separate preparation, recording, interface treatment and later updates in the quote. A changed dataset can affect several screens and spoken lines; a changed feature can invalidate an old scenario. An integrated recorder does not remove that work.
For a repeat content program, retain approved samples, product versions, field explanations, scripts and permitted uses together. That lets the next team distinguish illustrative numbers from actual supported results before making a localized cut or sales walkthrough.
When briefing ONCE’s product demonstration team, bring one operation to explain, approved demonstration data and a decision about whether actual customer outcomes belong in the film. Resolve the data story before adding visual polish.
Sources and scope
Adobe Help Center, “Record the screen and webcam footage in timeline,” updated October 28, 2026, and “What’s new in Adobe Firefly,” updated October 25, 2026, October entry. Checked October 29, 2026. The platform facts establish recording capability and beta status only. The data-story method and ticket example are ONCE’s analysis and hypothetical illustration, not customer results. The original AI cover is a conceptual commercial scene.
