Before-and-after is the most reliable format a service business has. It needs no script, no presenter and no acting. The proof is the pictures. A landscaper, a renovator, a detailer, a dentist, a hairdresser and a personal trainer are all selling the same thing — the distance between two states — and the format shows that distance in six seconds.

It is also the format people get wrong most consistently, and almost always in the same three ways.


The four beats

Every good before-and-after reel is the same four beats. Change the job, keep the structure.

BeatShotOn-screen textHold
1. The state it was inWide, from the spot you will shoot the afterName the problem plainly2–3s
2. The workMid-job — tools, mess, progressWhat you actually did3–4s
3. The revealSame frame as beat 1, finishedThe payoff line4–6s
4. The askOne detail shot of the finished workOne specific action2–3s

That is twelve to sixteen seconds. It does not need to be longer, and every second past about thirty is a second of people leaving.

The three ways it goes wrong

1. The caption lands on the wrong photo

This is the big one, and it is subtle enough that people ship it without noticing.

You write four lines of copy. You have nine photos. The editor — you, or a tool — pins the copy to the first four photos in the order they came off the camera. But your camera roll is in the order you took them, which means the first four are all before and during shots. So the word "paradise" appears over a photo of the drained, filthy pool, and the actual finished pool plays later with nothing written on it at all.

The viewer reads the mismatch instantly, even if they couldn't tell you what was wrong. The reel feels off. Check every caption against the frame it is actually sitting on, not against the line above it in your script.

2. The reveal is buried

Nine photos of a job, and the best one — the golden-hour wide of the finished thing — is number seven. If the reel plays them in camera order, the payoff arrives after most viewers have gone.

Sequence for the story, not for the timestamp. The finished shots are not the tail of the reel; the single best one is the climax, and the rest are the encore.

3. The CTA fires before the payoff

If "DM me for a quote" appears at second ten and the reel runs another twenty seconds, those twenty seconds are dead air. The ask marks the end of the reel whether or not the file has stopped.

Put the call to action after the reveal, and put nothing consequential after the call to action.

4–6sScreen time for the reveal. It should be the longest beat in the reel — it is the only frame anyone screenshots, saves or sends to their partner.

The shot list — take these on the job

The reel is decided at the job site, not in the editor. Four photos, taken deliberately:

Take more than four if you can — eight to ten gives you room to pick — but these four are the reel.

Matched framing, specifically

Same standing position, same phone height, same orientation, same zoom. Vertical, always. If you can, same time of day — a before shot at noon against an after at golden hour reads as a lighting trick rather than a transformation, and sharp-eyed viewers will say so in the comments.

Writing the four lines

Plain beats clever. The copy's job is to make the pictures legible, not to be admired.

Pacing, music and captions

Pacing. Beats one and two move; beat three sits. A common instinct is to cut faster toward the end for excitement — it's wrong here. The reveal needs room.

Music. One track, and if it has a lift, put the lift on the reveal. If your library track is a flat loop with no build, don't write "and then everything changed" over it — the audio will contradict the copy. Match the promise to the track you actually have.

Camera motion. Still photos held still look like a slideshow. A slow push on the before builds a little tension; a slow pull on the reveal opens the space out. Subtle is the whole point — if a viewer notices the motion, it's too much.

Captions. Large, high contrast, in the middle third. Most viewing is muted, and the bottom of the frame is covered by the platform's own interface.

By trade

Landscaping and outdoor. The strongest vertical for this format — the changes are large and legible at a glance. Shoot the before on the quote visit, before anything is moved.

Renovation and building. Matched framing matters more here than anywhere, because interiors are hard to orient in. Same doorway, same height.

Salons and barbers. Fastest turnaround of any trade — before and after are ninety minutes apart. Get written consent before anything goes out; the salon guide covers the waiver.

Detailing and restoration. Split-frame works unusually well here because the surface is the subject and the boundary between the two states is visible in one shot.

Food. Before-and-after is a weaker fit — nobody wants the "before" of a kitchen. Process and reveal works better; see the food reel hooks pillar.

Frequently asked

How long should a before-and-after reel be?
Twelve to twenty seconds. Four beats, with the reveal held longest. Past thirty seconds you are losing viewers without adding proof.

Can I make one from photos alone, with no video?
Yes — that's the normal case. Four stills with camera motion and timed captions reads as video. No filming required.

Should the before come first, or should I hook with the after?
Before first for most trades — the contrast needs a baseline to land against. Leading with the after works when the finished result is genuinely arresting on its own, but you then have to return to the before, which costs you a beat.

What if I forgot the before photo?
You have a process reel, not a before-and-after. Shoot the work and the reveal and write it as "how this gets done." Then set a rule: photo before you unload the van.

Do I need permission to post a client's property or face?
Property, generally check — many clients are happy, some aren't, and asking costs nothing. Faces, always, in writing. Cropping to hands and work avoids the question entirely.

How often should I post them?
One a week is plenty and sustainable. You have a new one every job; the constraint is editing time, not material. Cadence data here.


Related reading