Every channel with more than one host eventually has the same slightly awkward conversation, right, usually about six months in, when someone notices that one face has been on most of the thumbnails and someone else feels invisible, and the question of whose face leads stops being a design question and becomes a team question. Some version of that conversation comes up with a lot of the duos and groups who come to ClickTheory, and the honest answer is that there is a system for it, but the system has to be built on what each video is actually about rather than on keeping everyone equally happy.
Permanent hosts are a different packaging problem from anything else with more than one person in it, and getting it right is mostly about turning the cast into something the audience recognises, so that whichever face leads this week, the viewer still knows exactly whose channel it is.
Why is a channel with permanent hosts different from a collab?
Because a collab is a one-off meeting of two audiences, while a multi host channel is one audience that knows the whole cast, so the packaging is about rotation and recognition rather than introducing anyone.
In a collab, the job is to make both audiences click on the same video, and the thinking for that is in packaging collab videos so both audiences click. On a channel with a permanent duo, a couple, a group of friends, a family or a podcast with fixed co-hosts, the viewers already know everyone, and many of them have a favourite. That changes everything, right, because the thumbnail is no longer a handshake between strangers, it is a signal about which part of the show this episode belongs to, and the viewer reads that signal before the title.
The other difference is time. A collab happens once, so you can design it from scratch. A permanent cast shows up in every upload for years, so whatever system you choose gets repeated hundreds of times, and a small mistake in that system adds up across the whole channel.
Whose face should lead the thumbnail?
The host whose story the video is, the one doing the challenge, making the claim, taking the risk or getting the surprise, not whoever led last week.
This sounds obvious, right, and yet a lot of the group channels I look at default either to the most popular host on every thumbnail or to a strict rotation where everyone takes a turn, and both of those break the connection between the frame and the video. If one host is attempting something and the others are watching, the person attempting it leads, and the others become the reaction in the background. If the video is a debate between two hosts, both faces lead with equal weight, because the tension between them is the story. If the whole group is doing something together, the thumbnail can show the group, but it still needs one face with the strongest expression to anchor the eye.
For instance, a duo channel where one host is teaching the other something should put the learner's face forward, because the viewer relates to the person who does not know yet, and the expert in the background reads as the promise of an answer. I'm pretty sure most hosts would agree with that rule in the abstract, it is just harder to accept when it is your face being moved to the back.
Two faces, one small rectangle
The geometry is the part I see most multi host thumbnails get wrong, right, because YouTube gives you one 16:9 rectangle (the custom thumbnail specs are the same for one host or five), and every extra face in it makes every face smaller, and at phone feed size small faces stop carrying emotion.
Two faces can work well if they are big, close together and both showing a clear expression, but three or four faces almost always need a hierarchy, one large and the others smaller or partly cropped, or the frame turns into a team photo that says nothing about the video. The direction the faces are looking matters even more with a cast, because if one host is looking at the other, the viewer's eye follows, and that gaze line can do the job of an arrow without any graphic at all. The detail behind that is in faces, eyes and where viewers actually look first, and it applies twice over when there are two sets of eyes in the frame competing for attention.
The audience subscribed to the cast, so the thumbnail's job is to tell them which part of the cast this video belongs to, in under a second.
How do you rotate hosts without confusing the audience?
By keeping everything except the face constant, the colours, the type, the logo position and the overall layout, so the channel stays recognisable whichever host is in the lead that week.
In my experience rotation confuses viewers when it comes with a change in style, right, when one host's videos look one way and another host's videos look completely different, because then the grid reads like two or three separate channels sharing an account. What works is a strong shared frame and a flexible face slot. The audience learns the frame, and the face tells them who this video is about. Over time viewers who have a favourite host start spotting that host's videos in the feed, which is exactly what you want, but they only do that if the rest of the thumbnail is consistent enough that the face is the thing that stands out.
Shorts add their own wrinkle for a cast. The frame is vertical, so two hosts usually end up stacked one above the other rather than side by side, and the cover gets read even faster than a long form thumbnail, which pushes you toward one face per Short almost every time. YouTube's A/B test is not available for Shorts either, so you cannot let a test settle which host leads, and the decision has to come from the story of the clip and from what your comments tell you about who the audience follows.
The top of that map matters more than people expect. When a video is aimed at people searching a topic who have never heard of you, the cast means nothing to them yet, so the thumbnail should lean on the topic and one clear face rather than the whole group, and the group can show up in the video itself.
Should every host get a signature look?
Often yes, a small recurring cue for each host, like a colour, a text position or a typical expression, helps returning viewers spot their favourite without breaking the channel's shared style.
The trick, I think, is keeping the cue small. A different background colour for every host can work on a channel with two hosts, but with four or five it starts to look like a rainbow grid and the channel loses its own identity. A lighter touch, a coloured outline or the colour of the text, usually does the job. What we write down for multi host channels is a short guide that says what never changes, what changes with the host and what changes with the topic, which is the same idea as our thumbnail style guide service, just with a section for the cast.
Building a cast library at every shoot
The single most useful habit for a multi host channel is shooting thumbnail faces for every host on every shoot day, because the best expression for a given video is rarely in the footage of that video.
Have you noticed how often your thumbnails use a blurry frame grab of one host mid word? That is usually the sign that the library does not exist yet, and it is the easiest thing on this list to fix.
What should the title say when there are several of you?
I usually want the title to talk about the video's idea, not the cast, because the face already tells the viewer who is in it, and names in titles only help when the audience already knows them.
Titles like "we tried" or "I made my co-host" work for a known audience because they set up the dynamic between hosts, and those are often strong for returning viewers. For search, the title should name the topic plainly and leave the cast out of it. One thing I would avoid is inventing conflict between hosts in the title to get a click, because the audience knows the cast well, and a title promising a fight that the video does not deliver does more damage on a group channel than almost anywhere else, as the viewers feel misled about people they care about.
What happens to the packaging when a host leaves or joins?
The old thumbnails stay as they are in most cases, and the new packaging shows the change gently, by keeping the frame constant and introducing the new face inside the system the audience already knows.
When a host leaves, there is a temptation to repackage every old video to remove them, and I rarely think that is worth it, right, because those videos are part of the channel's history and the audience knows what happened. What does matter is the first few videos after the change, where the frame should feel familiar so the audience knows the channel has not changed underneath them. When a new host joins, the first videos with them should usually pair them with an established host, so viewers see a known face next to a new one and the new host borrows some of that recognition.
When does a group channel not need a studio?
If one of the hosts is a capable designer and enjoys it, doing it in house is often the right choice, because they know the cast and the inside jokes better than anyone, and the main risk is that the designing host leans toward their own face.
Where we tend to help, right, is when the cast is too busy filming to design, when the grid has drifted into a mix of styles, or when the internal argument about whose face leads has become a real problem and an outside person deciding by the story of each video takes the politics out of it. Our usual pack is $250, which gives you eight custom thumbnails with titles worked through for each, and the details are on services. If your channel is a podcast with fixed co-hosts, the same ideas apply (and if you want the episodes and clips edited too, that is what the Pixel Samy podcast team does), and the extra issues that come with that format are in why podcast thumbnails underperform and our thumbnail design for podcasters.
A cast is the most valuable thing a group channel owns, and the packaging should treat it like that, one shared frame the audience trusts, one face at a time telling them whose story this is, so that the rotation becomes part of why people keep watching. So yeah. That's my way of saying it.