Adding a third character to a two-person story feels like an increment and is actually a structural change. Two people have one relationship; three have three, plus whatever exists between all of them at once. That is four things to keep alive rather than one, and most work that attempts it does not plan for the increase.
The visible symptom is familiar: one character quietly becomes furniture, present in the images and absent from the story. That is a writing failure before it is a drawing failure, and it is entirely avoidable with a small amount of structural attention.
Three pairs plus a group
Between three characters there are three pairwise relationships, and each has its own history, tension and unresolved questions. There is also the dynamic of all three together, which is not the sum of the pairs and frequently contradicts them: two people who are easy alone can be awkward with a third present.
Writing all four is more work than most short pieces can support, which is a practical argument for using three characters in longer forms and keeping shorter ones to two, as suggested in the piece on short-form work. Attempting four relationships in eight pages produces four sketches rather than one story.
Attention is finite and visible
In any scene, characters are attending to each other in some distribution, and readers track it closely. A character who receives no attention from the other two for a long stretch drops out of the reader's model of the scene, even if she is drawn in every panel.
This means attention has to be allocated deliberately, rotating so that each character is the object of somebody's attention regularly. In drawn work that is a staging problem as much as a writing one, and it connects directly to the grouping and hierarchy questions in the piece on composing groups.
The furniture problem
The standard failure is that one character has no wants of her own, reacts to nothing in particular, and exists to complete an arrangement. Readers notice this immediately, and it is the three-character version of the fault described in the piece on what makes an explicit scene work, where one participant is present only as a body.
The test is to ask what each character wants from this specific scene, and what would be different if she were not there. If one of them has no answer, the scene has two characters and a prop, and it will read that way regardless of how carefully she is drawn.
Rotate the pairings
Across a story, deliberately give each pair its own scenes without the third present. That is how three relationships get developed rather than one relationship with an audience, and it gives the group scenes something to draw on. It also solves pacing, since two-person scenes are structurally simpler and provide relief from the harder ones.
The structure that emerges is usually alternating: pair scenes establish, group scenes complicate, and the changes carry forward. Each scene should still pass the change test from the piece on what makes a scene, which is harder with three people and more necessary.
Jealousy is not the only engine
The reflex when writing three characters is to make the tension about exclusion, and that works, once. It is also the least interesting of the available engines and it tends to flatten one character into an obstacle. Differences in what each person wants from the arrangement, or in how much each is prepared to say, produce more durable material.
Asymmetry of knowledge works particularly well with three, since information can travel through one person to another, which is impossible in a two-hander. That is a structural advantage of the form and it is regularly left unused.
Keeping three characters all present
Plan the three pairs plus the group, rotate attention deliberately, give each character a want in every scene, write pair scenes without the third, and find an engine other than jealousy. That is enough structure to keep a three-person story from collapsing into two people and a bystander.
All characters are fictional and unmistakably adult, and every participant must remain individually legible, as set out in the reader agreement.