Can I tell it who to follow?

Not by naming a person. What you can control is the template, and split screen removes the choice entirely by keeping both.

Does this happen more with three or more people?

Yes, because there are more candidates and more chances for two of them to talk at once.

Would separate microphones help?

Considerably. One microphone per speaker is the single biggest improvement available for multi person recordings.

What about a panel on a stage?

Wide stage shots are difficult because everyone is small in the frame. Choosing moments with a single speaker holding the floor works better.

Is a reaction shot ever the right choice?

Sometimes it is the better clip. If the reaction is the moment, the wrong person is actually the right one.

Does seating position matter?

It does. People seated at the ends of a table are harder to isolate than people in the middle, and it is worth thinking about before recording rather than afterwards.

Why the Wrong Person Is in Frame

What makes a clip follow the wrong person?

Three things, usually together. One microphone picking up the whole room, so nobody is clearly louder. A speaker looking away from the camera while another faces it. Or overlapping speech, where two people genuinely are talking. Each is a property of the recording rather than of the clip.

Recording setups and how they behave

Almost every case of the frame following the wrong person traces to the recording rather than to the clip.

Setup	What happens	Improvement
One microphone in the middle of a table	Nobody is clearly louder, so the frame hesitates	One microphone per speaker
Lapel microphone each	Works well, the speaker is unambiguous	Nothing needed
Talker in profile, listener facing camera	The listener reads as the subject	Seat people facing the camera
Genuine crosstalk	Two people really are speaking	Split screen keeps both
One camera per person, cut together	Only one is ever on screen	Keep a wide two shot as well

Recording habits that help

Clear recordings help every part of the process. Give each speaker good light on their face, seat people so both are visible to the camera and record in a quiet room. Ask participants not to talk over each other during important points, so each moment has one clear speaker.

These habits improve every clip, not just framing.

Choosing moments with one voice

Moments where one person speaks at length make the strongest clips anyway. Quick crosstalk rarely stands alone in a feed. Prefer clips where one speaker holds the floor.

Those moments are easiest to follow for viewers too.

When a reaction is the point

Sometimes showing the listener is right: a surprised face or a laugh can be the most engaging part of a moment. If the reaction carries the story, keep it.

Consider a layout showing both people for these moments.

Panels and larger groups

With three or more people, choose clips where one person speaks for a sustained stretch. Short exchanges between many people are difficult to present in a narrow frame.

A wider shape can help for group moments.

Checking before posting

Watch every clip once on a phone and make sure the person on screen matches the voice. If it does not, choose a different moment or shape.

A quick check prevents confusing clips.

Improving the next recording

Note what made the confusing moments difficult and adjust the setup next time.

Small changes in seating and lighting help a lot.

Keep the talker in the picture

Check where the microphones were. One shared microphone in the middle of a table is the most common cause.
Watch who is facing the camera. A talker in profile and a listener face on is a hard combination.
Use split screen when both matter. On paid plans it keeps two people in shot instead of choosing between them.

Send a two person recording and see who it follows.

Questions people ask

Can I tell it who to follow?: Not by naming a person. What you can control is the template, and split screen removes the choice entirely by keeping both.
Does this happen more with three or more people?: Yes, because there are more candidates and more chances for two of them to talk at once.
Would separate microphones help?: Considerably. One microphone per speaker is the single biggest improvement available for multi person recordings.
What about a panel on a stage?: Wide stage shots are difficult because everyone is small in the frame. Choosing moments with a single speaker holding the floor works better.
Is a reaction shot ever the right choice?: Sometimes it is the better clip. If the reaction is the moment, the wrong person is actually the right one.
Does seating position matter?: It does. People seated at the ends of a table are harder to isolate than people in the middle, and it is worth thinking about before recording rather than afterwards.