A podcast with two people is built for a wide screen. The host sits on one side, the guest on the other, and the horizontal frame holds both of them comfortably. Turn that into a 9:16 short and the frame suddenly holds about a third of the picture. One person stays, the other disappears, and if the clip is a back-and-forth, half the conversation is now coming from someone you can’t see.
I covered the general layouts in why a vertical crop isn’t always the right edit. This post goes deeper on the case I see most often: two people talking, recorded either in a studio with several cameras or remotely with each person on their own webcam. These are the three layouts I actually use for it, how I pick between them, and a couple of mistakes I made along the way.
First question: who carries the moment?
Before picking a layout, I watch the clip once and ask who the viewer needs to see at each point.
Most podcast moments fall into one of two shapes. In the first, one person does nearly all the talking: the guest tells a story, or the host goes on a rant, and the other person mostly listens. In the second, the moment is the exchange: a disagreement, a joke that lands on the other person, a question and a fast answer. The first shape wants the speaker big. The second wants both people visible, or at least a clean switch between them.
Getting this wrong is where most bad podcast shorts come from. A story crammed into a split screen makes the storyteller small for no reason. A heated exchange cropped to one face turns into one person arguing with a voice off screen.
Layout 1: follow the speaker
Use it when the recording has a separate camera on each person, and one person talks at a time.
How it works: each camera shot gets its own full-bleed vertical crop, centred on whoever is talking, and the edit cuts between them as the conversation moves. It’s the most native-looking option, because every frame is a big, clear face.
The rule I hold to here is simple: the camera is on the person speaking. Reaction shots are allowed, but only when the reaction adds something, like a laugh or a look of disbelief, and only for a moment.
A few things I’ve learned doing this on multi-camera shows:
- Don’t trust the studio’s own vertical edit blindly. One multi-host sports show I cut for supplied both a horizontal master and a vertical version. The vertical cut sat on the wrong person often enough that I went back to the horizontal footage and cropped each shot myself.
- Watch for late switches. Program feeds are edited live, and the switch to the new speaker sometimes lands a few frames after they start talking. In a short, that reads as a stutter, so I fix it in the edit rather than copying it.
- Avoid flash shots. When two people talk over each other, the edit can end up with shots that last a fraction of a second. Those look like mistakes on a phone. I’d rather hold one person a little longer.
Layout 2: the stack
Use it when both reactions matter at the same time, or the show was recorded remotely with each person in their own window.
How it works: one person goes in the top half, the other in the bottom half, each cropped to fill their panel. The captions usually sit on the seam between them, where they’re close to both faces and away from the platform buttons at the bottom of the screen.
The stack is at its best when it’s built from separate recordings of each person. Remote recording tools such as Riverside save an individual video and audio track for every participant, which means each panel can be a clean, full-quality crop instead of a cut-out from a small window in a combined feed. If your show is recorded that way, sending those tracks to your editor makes a real difference.
A few practical details:
- Pick an order and keep it. An HR interview podcast I made sample shorts for puts the guest on top and the host on the bottom in its own shorts, so I kept that. Whatever the order, switching it mid-clip makes viewers reorient themselves.
- Keep captions off faces. In the first version of a sample set for a motorsport podcast, the captions sat over a guest’s chin. I caught it in review and moved them, and now I check caption placement against each face, not just the canvas.
- Don’t stack a monologue. If the bottom person is just listening for forty seconds, the stack halves the speaker’s size for nothing. That clip should follow the speaker instead.
Layout 3: the band
Use it when both people are in one wide shot and the moment depends on seeing them together: a shared laugh, a glance, someone leaning in.
How it works: the original 16:9 frame stays intact as a horizontal band across the middle of the vertical canvas, usually over a blurred or darkened copy of the same shot. There’s room above for a title and below for captions.
The band is also the safe choice for more than two people. On sample shorts for a business podcast with three people at the table, I noticed their own shorts used a fairly tight band that sometimes cut the third person off during crosstalk. For those moments I used a wider band so nobody was lost.
The trade-off is size. Faces in a band are small on a phone, so it needs big, readable captions, and it shouldn’t be the default for a whole clip if a closer layout would work.
Mixing layouts inside one clip
These layouts aren’t a choice you make once per clip. Many podcast feeds switch between single shots and a side-by-side view, and the short can follow that: a full-bleed crop when the feed shows one person, a stack when it shows both. That was the plan on the HR podcast samples, and it keeps the short moving with the original edit instead of fighting it.
What should stay the same is the styling. Same caption font, same colours, same position logic.
The off-camera interjection problem
Here’s a mistake I made that’s worth sharing. In one clip, the guest was mid-answer and the host interjected a quick two-word question. The original feed stayed on the guest, so the host’s line came from off screen.
My instinct was to show him. First I cut to a shot of the host taken from another moment. Then I tried a split screen with both of them. Both versions looked wrong in review, and the reason was obvious once I saw it: the borrowed shot of the host wasn’t from that moment, so his lips never matched what he was saying.
The version that worked was the plain one. Keep the guest on screen, play the host’s line over the shot, and caption it like any other line. Viewers understand off-camera voices perfectly well. What they don’t forgive is a face that’s clearly saying something else.
Is showing both people worth the effort?
I wanted to know whether any of this shows up in results, so when I studied the best and worst posts from 33 creators who clip their own long-form videos, I included what’s on screen in the first frame. Clips that opened with two or more people in frame were more common among the hits than the flops: about one in five hits, against about one in ten flops. Opening on a lone face close-up leaned the other way.
That was software labelling a few hundred posts, and the effect was modest, so I treat it as a hint, not a rule. But it matches what I see while editing: an exchange reads as a conversation when you can see it happen, and that’s often what makes a stranger stop. I wrote more about that study in the cold open.
A quick checklist
- Decide who carries the moment before choosing a layout.
- One talker: follow the speaker. Two reactions at once: stack. Shared moment in a wide shot: band.
- Camera on whoever is speaking; reactions short and only when they add something.
- Captions on the seam or in the lower middle, never over a chin or a mouth.
- Never fake a face from another moment to cover an off-camera line.
- If you record remotely, give your editor the separate tracks.
If you’d like to see your own two-person show cut this way, I can make a free short from one of your episodes. You keep it either way.