In a novel, a subplot is a spacing problem. A thread opens on page ninety and returns on page two hundred and ten, with a hundred and twenty pages between them that most readers cross in a few sittings. The reader still has the thread. You are managing proportion, not memory.
In a serial, the same two beats are separated by time. A weekly serial that opens a thread in chapter seven and returns to it in chapter nineteen has put twelve weeks between them. Three months. The reader finishing chapter nineteen last held that thread in early summer. The thread is not spaced out. It is gone.
That one difference changes how often a secondary thread has to appear, what its return beat must accomplish, and how many threads you can honestly carry.
The gap is measured in weeks, not in pages
Stop counting chapters and start counting calendar. Twelve chapters means nothing on its own: at three a week it is a month, at one a fortnight it is most of half a year. Your thread's survival depends on which you are writing, which is why this and the update schedule you picked are the same question wearing two hats.
Do the arithmetic once, in weeks. Published weekly, every chapter of absence costs your reader seven days, so a B-plot surfacing every fifth chapter asks someone to hold a fact for over a month between reminders, unassisted. That is a tall order, and not because readers are careless. It is a real month.
Re-establish the question, not the person
The instrument here is not new to this blog. Folding a reminder into forward motion, so the returning reader is caught up and the archive-bingeing reader is never lectured, is already the house answer for two smaller problems: bringing back a character who has been offstage, covered in keeping a growing serial cast legible, and opening a chapter after a gap, covered in chapter hooks that survive the wait.
What changes at thread level is the object. Re-installing a person restores who they are and what they want. Re-installing a thread restores an unanswered question, and questions decay differently. A reader who has forgotten a face recognises it when they see it again. A reader who has forgotten a question does not know there is anything to remember, so nothing on the page triggers recognition. There is no prompt to lean on; the beat has to carry the question inside it.
Practically: the return beat states the open question again as part of doing something else. Not "she still had not heard back about the loan." Instead: "The bank had not written in five weeks, so she stopped checking and started pricing the van." That restores what is unresolved, says how long it has been unresolved, and moves. The reader who remembers gets plot. The reader who forgot gets the question handed back without being told they lost it.
Surface a live thread at least once every three chapters
Writers want a range here, and a range is useless: the only value of a cadence rule is that you can check a chapter against it. So take a number. A thread you intend to keep live should surface at least once every three chapters. Four is the outer edge. At five and beyond readers stop treating it as pending and start treating it as finished, and a thread filed as finished cannot be picked up later without feeling like a retcon.
Surfacing is a low bar on purpose. It is not a scene. It is one concrete mention that costs the reader nothing: an unopened letter on the hall table, a name in a list, a refusal to answer a question. What it cannot be is a status update from the narration, because that is exposition and reads like it. The same restraint that keeps a serial's rhythm from flattening applies: threads stay warm through detail, not updates.
What the return beat has to do in a single scene
A novel's B-plot return beat assumes the reader arrives with the thread loaded. A serial's cannot, so it does two jobs at once, every time: re-establish the question and advance it in the same scene. A beat that only reminds is a recap, and readers going straight through feel it as a stall. A beat that only advances leaves half the audience watching a payoff for a setup they no longer hold.
The test takes a minute. Read the scene as though you last saw this thread three months ago: can you tell what is unresolved? Then as though you saw it yesterday: did anything change? If either answer is no, the beat is doing one job and needs the other added, not a better version of the one it has.
How to tell a thread has gone cold
The tell is what readers reach for when they describe your story to someone else. Whatever they leave out is not part of the story they think they are reading, however many words you have given it. A cold thread disappears from those retellings first, well before anyone says they have lost track of it. Nobody announces that they have forgotten something.
The second tell is in your own drafting. If you write a sentence whose only job is to remind readers a thread exists, the thread is already cold and the reminder will not warm it. The sentence is a symptom: the last few chapters had no natural place for the thread, which usually means it is not tied tightly enough to the main line to surface on its own.
Bringing a cold thread back without an announcement
The wrong move is the one that feels most responsible: a scene that formally reopens the thread, reminds everyone where it stood, and over-explains by way of apology. It tells the readers who let go that they were right to, and the ones still holding on that you did not trust them.
Re-enter through consequence instead. Do not return to the thread; return to something it quietly caused while it was offstage. The loan that never came through is now a van that was never bought, and the character is on a bus. The reader who forgot the loan learns it as a fact about the present, not a correction of their memory. The reader who remembers gets a payoff they saw coming. Neither is told that time has passed.
One caveat. A thread you cannot re-enter through consequence, because nothing changed while it was away, was never load-bearing, and reviving it is an expensive way to find that out. Keeping the list of what is genuinely still unpaid is record-keeping rather than craft, and it belongs in the series bible you keep while you write, not in your head.
Carry one fewer thread than you want to
The cast side of this is settled: hold the number of characters live at any moment to something small, and put people to sleep deliberately. Threads need their own, lower number, because a thread costs more than a person does. A person can stand in a scene and be recognised. A thread has to be re-established every time it appears, so each live thread is a recurring tax on every chapter that touches it.
Two live threads beside the main line is comfortable. Three is work. Four is where chapters start opening with a paragraph of quiet reloading. The one to cut is rarely the weakest idea; it is the one least connected to the main line, because that is the one that never surfaces naturally and needs the reminder sentences that give the problem away.
Cut it in the middle, where you still have room. Near the end the arithmetic hardens, and the rules for the last stretch of a serial get much stricter about what you may open at all. The middle is where you can still choose. Choose there, rather than discovering forty chapters in that you have been paying rent on a thread nobody has thought about since spring.
Subplots keep a serial breathing between beats. For the finish line, see how to end a serial without losing the ending.