The short answer: how many characters is too many
Two leads, five supporting characters, and eight minor names add up to fifteen names at eighty thousand words. That total sits above the mystery and thriller convention of six to twelve named characters, so it signals bloat for that shelf. Three or so minor names should go or fold into stronger roles, point-of-view voices should stay at one or two, and each name that remains must perform a plot job no one else performs. This convention comes from craft practice, not from sales records or reader surveys, and it shifts upward as books grow longer and as genres widen their scope.
The question behind the question is practical. Writers rarely ask for a perfect number. They ask because a draft feels crowded, because beta readers mix up names, or because a subplot stalls every time a new face walks in. The tracker answers that worry with arithmetic: sort names by importance, divide by book length, compare the result with a genre band, and act on the excess. The rest of this guide explains each move in plain steps.
What counts as a named character
Count every person the reader must remember by name. That includes the protagonist and antagonist, the friend who recurs, the neighbor who appears twice, the shopkeeper with a name tag in chapter three, and the ancestor mentioned by full name in four scenes. If the reader needs the name to follow later events, include it in the count.
Do not count unnamed walk-ons. A guard who says one line and never returns is not a named character. A nurse who hands over papers without a name is not one either. The distinction matters because unnamed functionaries cost almost no memory while names promise the reader that this person will matter later. Every name is a small promise. Too many promises dilute attention.
Also count names that appear only in dialogue or messages. A sister who never enters a scene but receives six phone calls still occupies memory. Readers track her. The tracker treats her as a minor named character unless she starts making decisions that alter the plot, at which point she belongs in the supporting tier.
Pets and creatures follow the same rule. A named horse that carries supplies across three chapters counts. An unnamed street dog that barks once does not. Be consistent: if you gave it a name on the page, add it to the tally, then decide whether the name earns its place.
The three tiers and how to sort them
Leads carry plot threads from beginning to end. Most novels hold one to four leads. A lead wants something large, acts toward it across the whole book, and changes because of the outcome. When you remove a lead, at least one full thread collapses. That collapse test is the fastest way to confirm lead status.
Supporting characters recur across several scenes with visible goals of their own. They help or hinder a lead, they hold information the lead needs, or they embody a cost the lead must pay. A supporting character can survive the removal test for one thread yet still matter: cutting them weakens the book without destroying it. Typical counts run three to ten depending on genre and length.
Minor named characters appear once or twice to perform a single function. They deliver a message, witness an event, or supply one skill at a decisive moment. They rarely hold goals beyond the scene. Typical counts run two to fourteen depending on genre, and this tier is where bloat gathers first because single-function names are easy to add during drafting.
Sort borderline cases by function, not by affection. A character you love who appears twice with no goal is minor. A character you dislike who recurs with demands is supporting. The tracker rewards honest sorting because merge and cut guidance reads the tier totals, not your feelings about each person.
How the tool math works
The tracker runs four calculations in order. First it adds the three tiers into a total cast number. Two leads plus five supporting plus eight minor equals fifteen. That sum is the headline figure everything else interprets.
Second it divides the total by book length in units of ten thousand words. Fifteen names across eighty thousand words means fifteen divided by eight, which is one point nine names per ten thousand words. The same fifteen names across one hundred and forty thousand words means fifteen divided by fourteen, which is one point one per unit. Length dilutes density, so longer books carry more names comfortably.
Third it compares the total against a genre band stated as craft convention. Mystery and thriller conventionally run six to twelve names at eighty thousand words. Romance runs six to twelve. Literary fiction runs seven to fourteen. Science fiction runs twelve to twenty. Epic fantasy runs eighteen to twenty-eight. These bands describe what working writers and editors commonly treat as manageable, not what market data proves optimal. No sales dataset sits behind them, and no accuracy promise attaches to them.
Fourth it computes excess by tier. Each genre carries caps for leads, supporting, and minor names. Mystery and thriller cap minor names at five, supporting at five, and leads at three. Epic fantasy caps minor names at fourteen, supporting at ten, and leads at four. When a tier exceeds its cap, the surplus becomes guidance: surplus supporting names become merge candidates, surplus minor names become cut candidates, and surplus leads become demotion candidates because leads are the most expensive tier to carry.
Worked example one: a thriller with two leads, five supporting, and eight minor at eighty thousand words. Total fifteen. Density one point nine per ten thousand words. Band six to twelve. Excess total three. Minor cap five against eight actual gives three surplus minor names, so cut candidates equal three. Supporting cap five against five actual gives zero surplus, so merge candidates equal zero. One point-of-view voice against two leads is manageable. Verdict: over the band, trim minor names first.
Worked example two: an epic fantasy with four leads, eight supporting, and twelve minor at one hundred and forty thousand words. Total twenty-four. Density twenty-four divided by fourteen, which is one point seven per ten thousand words. Band eighteen to twenty-eight. Total sits inside. Minor cap fourteen against twelve actual gives zero surplus. Supporting cap ten against eight actual gives zero surplus. Three point-of-view voices against four leads is manageable. Verdict: fits the band, protect the balance.
Worked example three: a romance with two leads, four supporting, and three minor at seventy-five thousand words with five point-of-view voices. Total nine. Density nine divided by seven point five, which is one point two per unit. Band six to twelve. Total sits inside. Yet five voices against two leads means voices outnumber leads, which the tracker flags as heavy. Extra voices multiply cast because each voice arrives with its own circle of contacts. Verdict: total fits but voice load needs reduction.
Illustrative hypothetical example one: the crowded village mystery
Illustrative hypothetical: imagine Mara, who writes a cozy village mystery at seventy-five thousand words. Her draft names two leads, six supporting villagers, and nine minor names including a named postman, a named baker, a named vicar, a named taxi driver, two named pub regulars, a named doctor, a named florist, and a named retired teacher. Total seventeen. Density seventeen divided by seven point five, which is two point three per ten thousand words. Band six to twelve for mystery. Excess total five. Minor cap five against nine gives four surplus minor names.
Mara works the plan tier by tier. She lists the single function of each minor name. The taxi driver exists to move the sleuth once, so the vicar, who already drives to the village, absorbs that trip. The florist exists to deliver one clue, so the baker, who already gossips, delivers it instead. Two pub regulars become one regular with a sharper tongue. Three minor names leave, one merges, total falls from seventeen to thirteen. Still one over the band, so she folds the retired teacher into the doctor, reaching twelve. Nothing plot-bearing is lost because each removed name performed a job another name now performs.
Illustrative hypothetical example two: the sprawling second-world fantasy
Illustrative hypothetical: imagine Tomas, who drafts an epic fantasy at one hundred and fifty thousand words. His cast holds five leads, eleven supporting, and sixteen minor names. Total thirty-two. Density thirty-two divided by fifteen, which is two point one per unit. Band eighteen to twenty-eight. Excess total four. Minor cap fourteen against sixteen gives two surplus. Supporting cap ten against eleven gives one surplus. Lead cap four against five gives one surplus lead.
Tomas resists cutting plot threads first. He demotes the fifth lead, a captain who commands three good scenes but carries no thread to the ending, into a supporting ally. That move drops leads to four and raises supporting to twelve, which is two over cap, so merge candidates rise to two. He combines two quartermasters with overlapping supply scenes into one. He folds two minor heralds into a single messenger who recurs. Minor falls to fourteen, supporting falls to ten, total lands at twenty-eight, the top of the band. The book keeps every thread because jobs moved rather than vanished.
Edge cases the count handles awkwardly
Ensemble casts break simple bands. A heist novel with six equal leads at ninety thousand words may total twenty names yet read cleanly because the structure promises many protagonists and rotates them in disciplined turns. The tracker will flag twenty as over the thriller band, and the flag is still useful: it tells the writer to check that each of the six voices earns its chapters through distinct goals rather than shared banter.
Very short books punish names fast. A forty-thousand-word novella with twelve names runs three per ten thousand words, which reads crowded in almost any genre. New writers often port a novel-sized cast into a novella without adjusting. The fix is steep: aim for the bottom of the band or below it, and convert half the minor names into unnamed functionaries.
Very long books forgive more than writers expect. A two-hundred-thousand-word epic with thirty names runs one point five per unit, which sits comfortably inside fantasy practice. Length alone is not permission, though. Thirty names spread across ten threads still confuse. Density is a screen, not a verdict on clarity.
Returning series characters complicate the tally. Book three may name twenty people, but eight are beloved returns the reader already knows. The tracker counts them anyway because new readers meet them fresh, yet the writer can weight the guidance: returning names cost less memory for followers and full price for newcomers. When the series targets newcomers, honor the full count. When it rewards followers, tolerate the top of the band.
Unnamed-but-recurring figures are a quiet trap. A bartender who appears in twelve scenes without a name still occupies memory through face and habit. The tracker excludes unnamed entries by design, so apply judgment: name the recurring figure and count them, or reduce their appearances. Do not hide a supporting role behind the absence of a name.
Children, crowds, and historical inserts inflate minor tiers. A school scenes chapter can add six named pupils in four pages. A battle chapter can add five named officers. A family saga can add eight named ancestors. Each addition may serve one chapter and burden twenty. Ask whether the scene works with two names and unnamed remainder. Usually it does.
Failure modes: how cast fixes go wrong
Cutting plot jobs instead of names is the most common failure. A writer deletes a minor messenger and loses the only delivery of a clue, then patches the hole with coincidence. The correct sequence is to reassign the job first and remove the name second. Every cut should name the surviving carrier of each lost function.
Merging opposites creates a muddled person. Combining a comforting mentor with a cruel rival yields dialogue that swings without cause. Merge names whose functions already overlap: two informants, two rivals, two helpers. Shared territory makes the union believable.
Demoting a lead without reassigning their thread strands the plot. A demoted lead still wants the large thing. Either the thread moves to a remaining lead with visible adjustments, or the thread closes early with consequences. A thread that evaporates reads as abandonment.
Reducing voices without reducing circles wastes the effort. Cutting a viewpoint chapter while keeping that voice's six exclusive contacts preserves the bloat. When a voice goes, audit its circle: keep one or two contacts who serve other voices, release the rest.
Adding names mid-draft without retiring any is slow bloat. Each new helper feels small at insertion and permanent by revision. A working rule helps: one name in means one name out until the draft is complete. The tracker supports this by making the total visible before each addition.
Chasing the bottom of the band flattens texture. A thriller with only four names can feel airless if every scene holds the same faces. Bands are ceilings with floors for a reason. Six names at the bottom of the mystery band is lean, not starved. Two names is a different book.
Genre-specific guidance
Mystery and thriller reward lean casts. Six to twelve names at eighty thousand words keeps suspicion legible. Readers must hold every suspect in mind, so each extra name dilutes the suspect pool. Keep two or three leads including the investigator and the central threat, cap supporting at five, cap minor at five, and run one or two voices. When a witness matters once, prefer an unnamed witness with a sharp detail over a named one who never returns.
Romance centers the couple. Six to twelve names protect page time for the relationship. Two leads carry the bond, up to five supporting supply friends, family, and rivals, and up to five minor names handle single functions. Extra ex-partners and extra confidants are the usual bloat: each one steals scenes from the central pair. One or two voices dominate the genre, and dual voice remains common. Three or more voices need strong justification because each added voice removes pages from the couple readers came to follow.
Literary fiction tolerates interior density but not clutter. Seven to fourteen names at eighty thousand words fit the convention. The genre often studies a small circle closely, so supporting characters carry thematic weight beyond plot function. A minor name can still earn its place through a single vivid scene that alters the protagonist's understanding. Voices stay low, commonly one, sometimes two. When the cast grows past fourteen, check whether ideas are multiplying through people instead of through pressure on a few.
Science fiction widens the crew. Twelve to twenty names fit the convention at eighty thousand words, rising with length. Ships, stations, and expeditions justify ensembles because the setting organizes them: rank, department, or mission groups the faces. Use the setting as a memory aid by introducing groups together and reusing them as units. Voices run one to three. Each added planet or faction tempts new names, so assign each location a standing cast rather than fresh faces per visit.
Epic fantasy carries the largest bands. Eighteen to twenty-eight names at eighty thousand words, expanding with length, reflect courts, armies, and journeys. Readers of the genre accept many names when houses, banners, and titles group them. Introduce the groups before the individuals, repeat the groupings, and keep a consistent naming style within each house so memory has handles. Voices run one to four, with multi-voice structures conventional. Past four voices, the circles multiply faster than pages can sustain, even at high word counts.
Point-of-view load and the voice circle rule
Each voice brings a circle. A new viewpoint character needs a home, colleagues, opponents, and personal stakes, which means three to six names that exist mainly for that voice. Four voices therefore add twelve to twenty-four associated names even before the main plot staffs up. That is why the tracker flags four or more voices as heavy and flags any count where voices outnumber leads.
One voice suits mysteries, romances, and many literary novels. The reader bonds deeply, suspicion stays controlled, and the cast stays lean because everyone is defined by relation to the single lens. Two voices suit romances with dual perspectives, thrillers with investigator and threat, and fantasies with split quests. Three voices suit quest fantasies and family sagas where threads genuinely diverge. Five or more voices belong to long epics with disciplined rotation, and even there each voice should command a full thread rather than commentary duty.
A practical test: list the chapters per voice. Any voice holding fewer than fifteen percent of chapters is a candidate for demotion. Convert that voice into a strong non-viewpoint lead whose scenes are witnessed by remaining voices. The cast shrinks twice, once from the voice and once from its orphaned circle.
The merge-or-cut method in full
Start with the tier totals and the excess figures. Suppose the tracker reports cut candidates three and merge candidates two. That is the work order: cut first, merge second, demote only if leads exceed cap.
Cutting proceeds name by name. For each minor name, write its single function in plain words: delivers the map, witnesses the quarrel, treats the wound. Then ask who else could perform that function without a new scene. The best carrier is a supporting character who already appears nearby. Move the function, delete the name, and check that no other scene references the removed person. Repeat until the surplus clears.
Merging proceeds pair by pair. List supporting characters with their two strongest traits and their plot jobs. Pair the two whose jobs overlap most. Combine their jobs into one person, choose the stronger name, and carry over the single best trait from the absorbed character. Read the merged scenes in sequence to confirm the combined person behaves consistently. One careful merge often improves the book beyond the count because the survivor gains depth from the union.
Demotion applies to surplus leads. Choose the lead whose thread resolves earliest or matters least to the ending. Reassign their remaining thread beats to a surviving lead, adjust motives in the affected scenes, and rewrite their viewpoint chapters through another lens. Demotion is revision work, not deletion, so schedule it before line editing.
After each round, recount. Writers often clear the flagged excess and discover the book reads better two names leaner still. The band is a ceiling, not a target. Ending near the middle of the band leaves room for the inevitable late-draft addition without re-entering warning territory.
Frequently asked depth: nearby questions writers actually ask
How do I handle a cast that feels confusing even inside the band. Bands screen density, not distinctiveness. Confusion inside the band usually comes from similar names, similar roles, or similar voices. Vary name shapes and lengths, separate introductions across chapters, and avoid giving two characters the same job. A cast of ten with five mentors confuses more than a cast of fourteen with clearly separated functions.
Should family members each count. Count the ones named on the page. Family sagas often name eight relatives where three would carry the story. Consider naming the three who act and referring to the rest by role: the aunt, the cousin. Role references preserve the sense of family without charging memory per person.
What about characters who die early. A death in chapter two still costs a name if the reader must remember the loss. Count the casualty. Early deaths can justify leaner later counts because grief occupies attention the way a living character would. Do not treat a dead name as free.
How do suspects differ from ordinary supporting roles. Suspects demand more memory because readers compare them. Five suspects plus three investigators plus four minor names total twelve, which touches the thriller ceiling. That is workable if suspects are introduced with distinct motives and kept visible on a steady rotation. A suspect who vanishes for a hundred pages returns as a stranger.
What if beta readers love a minor character I planned to cut. Affection signals potential, not obligation. Promote the beloved minor into a supporting role by giving them a goal and recurring scenes, then cut a weaker supporting character to hold the total. Swaps preserve the band while honoring reader response.
Where drafting happens: Co-Writer
The tracker outputs a cast roster sorted by tier plus a merge-or-cut plan with counts. That roster needs a drafting home where each surviving name gains scenes, goals, and voice. Co-Writer at /write is that home in this product: open a project there, create one document per surviving lead and supporting character, and record the plot job each one alone performs. Then draft or revise the manuscript in the same place, working through the merge and cut list name by name until the total rests inside the band.
A closing checklist in prose
Sort honestly, weigh length, respect the genre band as convention, clear minor surplus before touching supporting depth, merge overlaps before cutting jobs, demote surplus leads with thread reassignment, hold voices at one or two unless the structure truly demands more, and recount after each change. Cast size is not a prize for imagination. It is a budget for attention. Spend it where the story pays back.
The tracker exists to make that budget visible. Use it early enough that cuts are cheap, use it at the midpoint while merges are still easy, and use it at the end to confirm the published cast is the one the story needs. Readers will never praise the count directly. They will say the book moved, the people mattered, and no one blurred. That response is what a well-sized cast buys.