How to make a before-and-after Reel for a cleaning or repair business

Last updated 2026-08-16 · 6 min read

What is the short version?

A cleaning or repair business already films every job, and the footage sits on a phone because assembling it takes an evening. The result is the most persuasive marketing this kind of business has: proof, in twenty-five seconds, that the work is real.

1. Go to Vizard Agent and upload the clips from one job.
2. Say the order — before, the work, the result.
3. Say the length, the shape and what the on-screen text should say.

What do you need before you start?

The footage and the sequence. Vizard Agent works out what is in each clip by looking at frames, so you do not have to label them, and the thing worth spelling out is the order you want them to run in and the words you want on screen.

• The clips. Before, during, after. Phone footage is exactly the expected input.
• The order. Number them in your brief if the filenames do not make it obvious.
• The length. Twenty to twenty-five seconds. This format works fast.
• The on-screen text. The service, the result, and a line telling people how to book.
• Whether the original sound matters. A vacuum or a drill under the music sells the work.

What do you type into Vizard Agent?

Write the sequence as a numbered list. Vizard Agent identifies the clips from their content rather than their names, and giving it the order plainly is what stops it having to guess which of two similar shots is the before and which is the after.

Prompt

Variants worth knowing:

• Split screen. A side-by-side before and after is a different and equally strong structure.
• Keep the working sound. Say so; the noise of the job is more convincing than music alone.
• A set of them. Send several jobs and ask for one Reel each in the same style.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real service Reel. Two things are worth noting: it studies how the object moves before deciding the cuts, and it checks its own render command before running it.

1. Probes every uploaded clip for dimensions and duration.
2. Extracts and organises frames from each one to identify what it shows.
3. Reads the resulting grid to see the clips visually.
4. Checks which clips carry usable audio tracks.
5. Searches the music library for an elegant modern track and downloads it.
6. Analyses specific frames of the object and the work to understand the movement in each clip.
7. Builds a second frame grid and reads it to see how the pieces move.
8. Lists the installed fonts and writes the styled title and call-to-action file.
9. Prints the ffmpeg filter graph to check it for errors, fixes it, renders, then extracts quality-control frames and looks at them.

Step 9 is the habit that saves time rather than showing craft. Printing the command before running it catches a channel-mapping mistake in a second, where running it and watching the output catches the same mistake several minutes later.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 25.37 seconds, AAC audio. Vertical, twenty-five seconds, the original job audio mixed under a licensed track, with styled titles and a call to action burned on.

Vizard Agent pulled a grid of frames from the finished render and looked at it to confirm the titles were positioned and styled correctly before delivering.

Twenty-five seconds across three stages gives each about eight seconds. The before needs the least, the work needs the most, and the result wants a beat of stillness at the end to be looked at.

How long it takes was not measured separately here. Comparable work runs a median of 28 to 38 minutes end to end, and the upper quartile is several times the median. Across all projects the median cost by tier is Flash 47, Pro 55, Max 242, Ultra 263 credits.

When does this not work well?

This format is proof, and proof is only ever as good as what was actually filmed on the day. Vizard Agent sequences, scores and titles whatever arrives and reviews every clip before cutting, and the shots nobody took are exactly the ones the finished Reel will be missing.

• A missing before shot ruins it. Film the mess before you start. It is the half people care about.
• Different lighting fakes the result. A before in shadow and an after under a lamp reads as a trick even when it is not.
• Customers' homes are private. Faces, addresses, personal belongings in shot — get permission before posting.
• Twenty-five seconds is three or four clips. More footage means a longer Reel, not a denser one.
• A Reel is not a portfolio. One job is one post; the case is made by posting them regularly.

How do you fix a result that came back wrong?

Say which clip or which caption. Vizard Agent keeps the analysed frames, the downloaded track and the styled title file separately, so reordering the clips or rewording the call to action is a re-render rather than a fresh review of the footage.

• "Show the before for longer." A timing change on that clip.
• "Change the booking line." The title file re-rendered over the same cut.
• "Keep the vacuum sound louder." A mix change against the same assembly.

How does Vizard Agent compare to doing it yourself?

By hand this is a phone editing app at the end of a working day: trimming four clips, finding music that is not copyright-claimed, and typing a caption with a finger. It is twenty minutes you do not have, repeated per job.

Common questions

Do I need to label the clips? No. Vizard Agent identifies them from their content; just say what order the sequence should run in.

Should I keep the working sound? Usually yes. The noise of the job is convincing, and Vizard Agent mixes it under the music.

How long should it be? Twenty to twenty-five seconds. Vizard Agent will hold to whatever length you name, and this format makes its point quickly or not at all.

Can it add my phone number? Yes. Give the exact wording and Vizard Agent sets it as a call to action at the end.

Can I get a square version? Yes. Vizard Agent re-lays the titles for the new shape rather than cropping the vertical cut.

Can it do several jobs in the same style? Yes. Send them together and ask Vizard Agent for one Reel each, kept consistent with one another.

Do the clips need to be the same resolution? No. Vizard Agent probes each one and normalises them during the assembly.

Can it work from a split screen instead? Yes. Say so and the before and after run side by side rather than in sequence.