Last month I opened an episode a host had cut himself before sending it over for a polish. Fifty eight minutes. I counted 214 camera changes. That is a switch every 16 seconds, metronome steady, whether anyone was talking or not.
He was proud of it. He told me he had set himself a rule: never stay on one angle longer than 20 seconds, because attention spans. I watched ten minutes of it and felt slightly seasick.
That episode is the reason I keep coming back to this topic. Multi-cam switching is not a pacing tool. It is a listening tool. The moment you start cutting on a clock instead of on the conversation, the audience stops believing anyone is in the room.
The timer rule is the most common mistake I see
I understand where it comes from. Somebody watched a big YouTube show, noticed it never sits still, and reverse engineered a rule out of it. Cut often. Keep it moving. Do not let the viewer get bored.
The problem is that the big show is not cutting every 16 seconds because of a timer. It is cutting when the reason to cut arrives, and on a good episode the reasons arrive often. Those are two completely different processes that happen to produce a similar cut count on a busy day.
Run the timer version on a slow, thoughtful interview and the seams show immediately. Someone is halfway through a hard sentence about their father and the frame jumps to a wide because 20 seconds elapsed. Nothing happened. The cut is a lie about what mattered.
Here is the test I use, and it is brutally simple. Watch any cut with the audio muted. If you cannot guess roughly what was going on in the conversation from the picture changes alone, the switching is decoration.
What actually earns a cut
After enough episodes you stop thinking about angles and start thinking about triggers. These are the ones I trust:
- A speaker change. The most honest reason there is. Somebody new starts talking, we go look at them. Not always, but this is the default.
- A reaction worth seeing. The guest says something and the host's face does something. That is the shot. This is the single most undervalued cut in podcast editing.
- A shift in energy. The conversation drops into something quieter and more serious, so we go to the tighter frame and stay there. Or it opens up into laughter and we pull to the two shot so the room feels bigger.
- A new thought. New topic, new subject, new question. The angle change works like a paragraph break.
- Covering a fix. An edit point that would otherwise show a jump gets hidden behind an angle change. Boring, functional, necessary.
Notice that four of those five are about the content and one is about craft housekeeping. That ratio is roughly right. When the housekeeping cuts start outnumbering the content cuts, the episode has a bigger problem than switching.
The cut should feel like a person in the room turning their head. Nobody turns their head every 16 seconds unless something is wrong with them.
Hold longer than feels comfortable
The hardest thing to teach a new editor is patience with a good take. When somebody is telling a story well, the correct number of cuts is often zero. Let the shot run for 40 seconds. Sometimes 90.
This feels wrong in the timeline because you are watching it for the fourth time and you are bored. The audience is watching it for the first time and they are not bored, they are listening. Your boredom is not data.
The flip side is real too. When someone rambles, when the answer is circling, cutting more will not save it. That is a trimming problem, not a switching problem. Adding angles to a weak stretch is like adding hot sauce to a bad sandwich. You notice the sauce, not the sandwich.
I would rather cut 90 seconds of meandering out of the episode entirely than switch cameras six times to distract from it. Our editors watch the whole episode, start to finish, so we can actually tell the difference between a slow moment that is building to something and a slow moment that is just slow. Silence detection cannot do that, which is one of several reasons I do not let a machine make that call.
Cameras should have jobs, not numbers
A lot of switching problems are really coverage problems. If your three angles are basically the same shot from slightly different chairs, of course the cuts feel arbitrary. There is no new information in any of them.
Give every camera a job before you record. Something like:
- A close single on the host. Reaction machine. Used most often for listening, not talking.
- A close single on the guest. Where the story lives. This one will carry the majority of the runtime on a good interview.
- A wide two shot. Establishes the room, resets the eye, covers laughter and crosstalk, and gives you somewhere safe to land when both people talk at once.
The moment each camera has a purpose, switching stops being a choice between three options and starts being a question with an answer. Do I need a reaction, a story, or a reset? That is why I think three cameras beat two for almost every interview format, and why the third camera is worth more than a fourth one would be.
Coverage also has to be physically consistent. If the guest is a foot taller than the host, your two shot fights you all episode and you will end up cutting away from it just to avoid the awkwardness. Sorting out seating and eyelines for guests of different heights before you roll saves more editing time than any switching rule ever will.
Crosstalk is where rhythm gets decided
Every interview has moments where both people talk at once. Those five seconds tell you whether an editor was thinking or clicking.
The wrong move is to chase it, cutting back and forth on every half sentence. You get four cuts in three seconds and the viewer feels the panic. The right move, almost always, is to go wide and stay there until the traffic clears. Let the two shot hold the mess. It reads as a real room where two people got excited, which is exactly what happened.
Same with laughter. Cutting during a laugh chops it in half and kills it. Land on the wide before the laugh peaks, or land on the reaction and hold through the whole thing. Do not cut in the middle.
This is also the place where sync problems announce themselves. Nothing exposes drift like a fast crosstalk exchange between two angles. If your angles were rolling separately, check them here first, and if something feels a frame or two off, fix the desync properly rather than nudging by eye before you cut a single frame of the episode.
Graphics and the switching rhythm are the same problem
When a lower third fires at the same moment as a camera change, you have asked the viewer to process two new things at once. It reads as busy even when each element is fine on its own.
My rule is to separate them by at least a second. Cut to the new angle, let it settle, then bring the name in. Or bring the name in on a held shot and switch after it clears. The general principle behind lower thirds that stay out of the way applies to every overlay you add, including chapter cards and B roll.
The same logic runs through multi-cam editing generally. Every element competes for the same attention. Switching is just the loudest one.
What this means for vertical clips
A clip cut for phones is a different animal. Thirty to sixty seconds, one idea, and the frame is tall so your wide two shot mostly does not survive the crop.
I do not reuse the long form switching pattern in clips. I recut them from the source, usually with far fewer angle changes than people expect, because in a tall frame a cut costs more. The eye has less room to recover. One clean angle change at the turn of the idea usually beats five.
And if the clip is heading to YouTube as a full episode, the packaging around it matters as much as the switching inside it. Our sister studio ClickTheory handles the thumbnail side, because a beautifully cut episode that nobody clicks is still a beautifully cut episode that nobody watched.
Lighting deserves a mention here too. Angles that do not match in exposure or color make every cut feel like a bump, no matter how well motivated it is. Most of what makes switching invisible gets decided at setup, which is why I keep pointing hosts at the podcast lighting guide before we talk about anything else.
The rhythm I actually use
If you want a starting point rather than a philosophy, here is roughly how a normal hour comes out of our timeline:
- Somewhere between 60 and 120 angle changes across a 60 minute episode, not 214.
- Long holds during stories, faster movement during back and forth.
- The guest single carrying the largest share of runtime on an interview show.
- The wide used as a reset, not as filler.
- Reaction cuts placed on the reaction, one or two frames before the sound of it, never after.
Those numbers move a lot by show. A debate format runs faster. A solo monologue with a second angle might have 20 changes in an hour and be perfect. The count is an output, not a target. The second you make it a target you are back on the clock.
One last thing. When you finish an edit, watch the first five minutes at full speed without touching anything. Not scrubbing, not skipping. If you notice the cutting at all, you cut too much. Good switching is the thing nobody mentions, which is a little unfair to editors and completely correct for the audience.
If you want a look at how we handle this on real episodes, the how it works page walks through the process. Or just send me an episode you are unsure about and I will tell you honestly where the rhythm is fighting the conversation.