A novelist who gives a character grey eyes in chapter four and blue eyes in chapter thirty can fix chapter four. Nobody has read it yet. A serial writer cannot. The wrong detail went out on a Tuesday, some number of people read it, and the options that remain are to contradict it in public, to write around it, or to have noticed it before it happened.
That is the argument for keeping a written record while you write a serial, and it is not the same argument usually made for planning. Planning is about deciding what happens next. A record is about remembering what you have already said. The two look similar on the page and solve opposite problems.
A series bible is a record, not a plan
The useful version of this document is closer to a receipt file than an outline. It holds things you have committed to in public and can no longer change for free. If a fact appears in a published chapter, it is now load-bearing, and the cost of getting it wrong later is a reader writing a comment that begins "wait, I thought".
Everything else - what the magic system implies, where the war is going, who the traitor turns out to be - belongs somewhere else, or nowhere. Those are decisions, and decisions are still yours to revise. Planning a novel is its own craft, and mixing the two documents is the fastest way to abandon both.
Names, spellings, and the details you said out loud
Start with the things a reader can check. Every character name and its exact spelling, including the ones you invented and typed twice differently. Place names. Job titles. Any physical detail you put in a sentence: height relative to someone else, a scar, a limp, the hand someone writes with. Ages, if you ever gave one. Family relationships, if you ever named one.
This list is boring to maintain and it is the part that earns its keep every single week. It is also the part that gets harder as the cast grows, which is the same pressure that makes a growing serial cast hard to keep legible for the reader. Your record and their memory are under the same strain, roughly ten chapters apart.
The timeline is the thing you cannot re-derive
Most continuity facts can be recovered by re-reading one chapter. The timeline cannot, because it is distributed. How many days have passed since the fire? Was the festival before or after she left? How long has the journey been taking?
Keep a running list of events in story order with an elapsed count beside each one. Not dates unless your story uses dates - just a number that goes up. The moment you write "three weeks later" you have made a claim that has to survive forty more chapters, and the only way it survives is if something outside your head is counting.
This gets sharper still if you are writing across two timelines, where every scene has to be placed twice and the reader is doing arithmetic you have to have done first.
Track the promises you have not paid off
The second thing worth its maintenance cost is a list of open promises: the sealed letter, the stranger who recognised her, the debt, the thing in the cellar. Each entry is one line - what was promised, which chapter, and whether it has been paid.
Nobody forgets the big ones. What gets lost is the small hook dropped in chapter eleven to hold a slow stretch together, which is exactly the hook a reader who binges the archive will be waiting on. This list is also the most useful document you own when you start planning how the serial ends, because the ending is largely the set of promises still unpaid, and you cannot schedule what you have not listed.
What to leave out
A series bible that takes an hour a week will be abandoned by chapter twenty, and an abandoned record is worse than none, because you will trust it after it stops being true.
So leave out anything you could re-derive by re-reading a single chapter. Leave out elaborate history nobody can contradict you about, which is the pleasurable part and the part that does no work. Serials build their world in fragments as they go, much the way short fiction builds a world in very little space, and the reader only ever holds the fragments you published. Depth that never reaches a chapter does not need a record. Leave out plot plans. Leave out anything you are still choosing.
Test each entry with one question: if you get this wrong, can a reader prove it? If not, it is not continuity, it is worldbuilding, and it can live in your head.
Starting one at chapter forty
Most people who need this already have forty chapters published and no record, and the size of the backfill is why they never start. Do not backfill. Open the file today, write down what you know off the top of your head, and add facts as you use them.
From then on, one rule: when you write a chapter, anything new and checkable that goes into it goes into the file before you publish. That is a two-minute pass while you already have the chapter open. The record fills forward at the speed you write, and within a dozen chapters it covers the material you actually reach for, which is never the whole archive.
Backfill only on demand. When you need to know whether the brother was older, look it up once, get it right, and write it down while you have it. The gaps that never come up were never worth an evening.
It only has to be faster than re-reading
The format matters less than any argument about format suggests. Plain text in one file is enough. What it has to be is searchable and one keystroke away, because it is competing with the alternative of scrolling your own archive, and it only gets used if it wins that race.
Keep one file, not five. Sort it in whatever way lets you find a name in three seconds. Append rather than reorganise - a tidy record you maintain monthly is worth less than a messy one you touch every chapter.
The payoff is not that your serial becomes airtight. It is that continuity stops costing you drafting time. Checking a fact becomes a three-second lookup instead of a twenty-minute re-read that breaks the rhythm you need to keep a serial moving. That is the entire return, and over a long serial it is a large one.