How to make a data-driven editorial reel

Last updated 2026-08-16 · 6 min read

What is the short version?

A data reel is a documentary in under a minute: a claim, a number, a chart, a consequence. The reason so few exist is that each of those parts needs a different craft, and the numbers have to survive being spoken aloud as well as drawn.

1. Go to Vizard Agent and give it the figures and where they came from.
2. Say the angle — what the data actually shows.
3. Say the language, the length and the feed.

What do you need before you start?

The numbers and their source. Vizard Agent will structure, narrate and animate them, and it cannot verify a statistic you invented, so the credibility of the whole video rests on figures you can point at — which is also what makes this format worth making at all.

• The figures. Written out, with units, and with the period they cover.
• The source. A named institution or dataset. It belongs on screen.
• The angle. "This is more than people think" is a different reel from "this has doubled since 2019".
• The language. Vizard Agent narrates and subtitles in the language you name, including ones with accented characters.
• The length. Fifty seconds is about six segments, which is a natural size for this format.

What do you type into Vizard Agent?

Give the numbers, the source and the angle in one message. Vizard Agent breaks that into segments and decides which figure deserves a chart and which deserves a single large number on screen, so you do not need to storyboard anything yourself.

Prompt

Variants worth knowing:

• Name a house style. Referencing a publication whose look you admire is a legitimate and useful instruction.
• One number, one reel. A single striking figure can carry thirty seconds better than six can carry fifty.
• Chart types. Say if a figure must be a map or a share breakdown rather than a bar.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real data reel. Two of the checks are the sort you only think of after being burned once: whether the narrator says the numbers correctly, and whether the font can actually draw the language's characters.

1. Checks the voice, music and motion-graphics tools available for the job.
2. Searches for narrator voices in the target language and generates test reads.
3. Compares the two candidates for pronunciation, accent and tone before choosing.
4. Narrates all six segments, then checks specifically how the numbers were pronounced.
5. Measures the exact duration of every segment so the visuals can be built to fit.
6. Generates the sound design — a receipt tearing, a tick, a digital click.
7. Locates the installed font files and renders a test image of the language's accented characters.
8. Looks at that test image to confirm the diacritics draw correctly, and finds a monospace face for the figures.
9. Renders the scene visuals and reviews them, chart by chart and card by card.

Step 7 is the one worth stealing. A missing glyph does not throw an error — it draws a blank box, and Vizard Agent checked one rendered image rather than discovering it in the finished reel.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 52.5 seconds, AAC audio. Vertical, just under a minute, six narrated segments with animated charts, subtitles throughout, and music and sound effects mixed under the voice.

Ask for a wide version for another platform and Vizard Agent re-lays the graphics rather than cropping them, since a chart that gets its edges cut off is worse than no chart.

Fifty seconds is about six segments, and the pacing is deliberately uneven: the opening claim gets room, the middle figures move quickly, and the closing number is held long enough to be read and remembered.

Timing was not measured separately for this kind of job. The closest measured work runs a median of 28 to 38 minutes end to end, with the middle half spread considerably wider. Across all projects the median cost by tier is Flash 47, Pro 55, Max 242, Ultra 263 credits.

When does this not work well?

The format's authority comes entirely from the figures, and Vizard Agent presents whatever you give it with equal confidence. Everything it does — the charts, the narration, the pacing — makes a number more persuasive, which is a problem when the number is wrong.

• It does not fact-check your data. Vizard Agent builds what you supply. The sourcing is yours.
• A chart can mislead honestly. A truncated axis exaggerates. Say if you want the scale to start at zero.
• Too many figures in fifty seconds. Six segments is a lot already. Cutting to four usually strengthens the reel.
• Numbers read aloud are easy to garble. Vizard Agent checks the pronunciation, and long or unusual figures still deserve your ear on the result.
• Attribution belongs on screen. A source named only in the narration disappears for the majority who watch without sound.

How do you fix a result that came back wrong?

Name the segment or the number that is wrong. Vizard Agent keeps the narration, the measured segment durations, the rendered scenes and the font work it verified, so correcting a single figure re-renders one chart and re-reads one line rather than rebuilding the whole reel from the beginning.

• "The third figure is wrong." That segment's narration and chart, redone.
• "The voice mispronounces the number." Re-read with the figure written out.
• "Start the axis at zero." A re-render of that chart alone.

How does Vizard Agent compare to doing it yourself?

By hand this is a spreadsheet, a motion-graphics tool, a voice recording, and a sound mix — four programs and a day. It is why data reels are made by publications with staff for it, and almost never by the analyst who actually has the numbers.

Common questions

Do I need to provide the data? Yes. Vizard Agent builds the reel around your figures and their source; it does not go and find statistics for you to publish.

Can it work in my language? Yes. Vizard Agent searches for a narrator in that language and verifies the fonts can render its characters.

Can it match a publication's style? Name the look you want. Vizard Agent builds to that register rather than a template.

How many figures fit in a minute? Around six segments in fifty seconds. Fewer usually lands harder.

Will the source appear on screen? Yes, if you ask. It should — most viewers never hear the narration.

Can I get a wide version as well? Yes. Vizard Agent re-lays the graphics for the new shape rather than cropping the vertical ones.

How long does one take? Most of the work is the narration and the charts. Vizard Agent builds them in one pass and later corrections come back much faster.