Beat-em-up game design explained

Subgenres and what makes a brawler

Beat-em-ups share a side or isometric view, melee-focused combat, and progression through discrete stages. Subgenres diverge on movement freedom, combo depth, and how strictly the camera locks forward:

• Classic belt-scrollers — linear left-to-right stages, two attack buttons plus jump, enemies spawn from screen edges. Examples: Double Dragon, Final Fight, Streets of Rage.
• Free-roam brawlers — wider arenas, full eight-direction movement, environmental weapons and throws. Examples: River City Ransom, Scott Pilgrim vs. the World.
• Modern revival brawlers — retro aesthetics with deep combo systems, parries, and online co-op. Examples: Streets of Rage 4, Teenage Mutant Ninja Turtles: Shredder’s Revenge.
• Hybrid action brawlers — brawler stages inside a larger action game. Examples: God of War arena segments, Bayonetta set-piece waves.
• Horde / survivor brawlers — auto-aim or simplified movement; hundreds of enemies, upgrade picks between waves. Examples: Vampire Survivors (distant cousin), Castle Crashers wave rooms.

Pick the subgenre before tuning enemy density. A classic belt-scroller lives on readable three-enemy-at-a-time fights; a survivor-style brawler lives on upgrade dopamine and screen-filling chaos. Mixing them without signaling the shift confuses players who expect one rhythm.

The core loop: walk, fight, advance

Every beat-em-up iteration reduces to three beats:

1. Walk — move through a stage segment with light ambient threats (pickpockets, breakable props, tutorial prompts).
2. Fight — a gate triggers; enemies spawn; the screen may lock until the wave is cleared.
3. Advance — the camera unlocks; players collect drops, choose a path fork, or enter a shop before the next gate.

The loop succeeds when fight segments feel like punctuation, not endless slog. A stage with ten identical wave rooms back-to-back fatigues; a stage that alternates tight choke fights, weapon pickups, and a miniboss breather keeps momentum. Stage length for a classic brawler: six to twelve minutes solo, with a boss every three to five minutes.

Screen lock and flow

Screen locking (invisible walls at the edges until enemies die) is the genre’s pacing lever. Lock too early and players feel trapped before they see threats; lock too late and runners kite off-screen. Spawn enemies slightly inside the visible frame so players read the threat before the lock engages. Unlock the moment the last enemy dies — or one frame after the death animation starts if you want snappy flow.

Combat: hits, juggles, and crowd control

Brawler combat is simpler than fighter frame data but must stay readable at eight-on-one density. Core verbs:

• Light / heavy chains — fast jab strings into a launcher or knockdown finisher. Three to five hits per chain is the classic ceiling before combo decay.
• Grabs and throws — beat blocking enemies and clear space by tossing one into others. Throws should have invincibility startup so they are not interruptible by cheap hits.
• Special moves — cost health, meter, or cooldown. Screen-clearing supers are crowd-control relief valves, not spammable win buttons.
• Air juggles — launch, keep aloft, slam down. Tune juggle gravity and hitstun decay so skilled players extend combos without infinite loops. See combo systems for scaling and reset rules.
• Invincibility frames — dodge rolls, backdashes, or parries give breathing room. Without a defensive option, eight enemies become unavoidable damage sponges.

Depth vs accessibility

Classic arcade brawlers used two buttons; modern revivals add combos, parries, and air dashes. Layer depth behind optional inputs — a tap-tap-special still clears normals, while execution-heavy routes reward style points or bonus drops. Never require a 10-hit juggle to progress the campaign.

Environmental interaction

Barrels, bikes, and knives break monotony and give weaker characters a power spike. Environmental weapons should spawn before difficulty spikes, not after players are already overwhelmed. Breakable scenery also signals where hidden paths or bonus rooms hide.

Enemy design and wave choreography

Enemies are the beat-em-up’s level design. Each archetype teaches a different response:

• Grunts — low health, walk into range, exist to be juggled and thrown. Fill the screen without threatening a one-shot.
• Rushers — fast approach, low health, punish players who only face forward. Force sidestep and rear awareness.
• Brutes — high health, armor on heavy attacks, grab attempts. Require kiting, throws, or co-op focus fire.
• Ranged — knives, guns, magic from the back row. Must be prioritized or they chip health during brawls. Cap simultaneous ranged units at two to keep readability.
• Elites / minibosses — unique patterns, phase transitions, arena hazards. Bookend stages and test everything learned.

Wave scripting

Script waves as compositions, not random spawns. A good three-beat wave: two grunts approach (warm-up), a brute enters while grunts distract (priority shift), a rusher flanks from behind (rear check). Stagger spawn timing by 0.5–1.5 seconds so players parse threats sequentially. Random spawn at full density reads as unfair noise.

Difficulty scaling

Scale difficulty by composition complexity, not only stat inflation. Late stages add elite modifiers (fire aura, shield) rather than 10x health grunts. If enemies become damage sponges, combat stops being a brawl and becomes a DPS check.

Co-op: the couch-king genre

Beat-em-ups are among the strongest co-op experiences because shared screen space creates natural teamwork — juggle setups, throw collisions, and aggro splitting happen without voice chat. Co-op design rules:

• Shared or split camera — classic belt-scrollers use one camera at the trailing player; modern titles teleport laggers or split briefly. Never let one player scroll four screens ahead alone.
• Enemy HP scaling — multiply enemy health 1.6–2x per additional player, not linear 2x, or fights end too fast. Add elite spawns instead of only inflating grunt HP.
• Friendly fire policy — off by default in casual modes; optional in hardcore runs for tension.
• Revive windows — downed players should be rescuable during a brief bleed-out. Instant permadeath kills couch sessions.
• Combo synergy — partner launches, you juggle; throw into partner’s waiting uppercut. Synergy rewards feel better than passive aura buffs.

Online co-op needs rollback or solid lockstep netcode; input delay ruins juggle timing. If online is a stretch goal, ship excellent local co-op first.

Level pacing and stage structure

A six-stage classic brawler often follows an intensity curve:

• Stage 1 — tutorial density; one enemy type at a time; introduce grab and special.
• Stage 2–3 — mix archetypes; first miniboss; optional shop or branching path.
• Stage 4 — difficulty spike; environmental hazards; forced co-op gate (heavy + ranged combo).
• Stage 5 — breather stage with spectacle (vehicle chase, auto-scroller) before finale.
• Stage 6 — boss rush or gauntlet; final boss with two to three phases.

Vary horizontal vs vertical scroll segments. A rooftop climb or elevator descent breaks the monotony of flat streets. Set-piece segments (motorcycle chase, falling scaffolding) are memorable but should not replace core combat for more than 60 seconds.

Rewards between stages

Shops (River City style), score-based rank upgrades, or character unlocks give players a reason to replay on higher difficulty. Tie upgrades to mechanics players already used — longer juggle, faster dodge — not raw +10% damage everywhere.

Worked example: Harbor Alley stage three

You are building Harbor Brawl, a four-player co-op belt-scroller. Stage three is a dockyard at dusk — six minutes target, one miniboss, one shop before the stage boss.

Segment A: arrival (45 seconds)

No screen lock. Three grunts wander; players practice the new parry move introduced in the stage-two interstitial. A breakable crate hides a pipe weapon. Teaches parry without punishment.

Segment B: crane yard fight (90 seconds)

Screen locks. Wave 1: four grunts. Wave 2: two grunts plus one brute (spawn behind crates). Wave 3: two rushers from left and right while a ranged enemy on a crane platform throws barrels (telegraphed shadow). Players must prioritize the ranged unit or use the pipe to knock down the crane ladder as a shortcut.

Segment C: shop alley (30 seconds)

Unlock camera. NPC sells health pizza and a one-use throwable fish (area stun). Optional side room with bonus score if both players stand on pressure plates — rewards co-op coordination.

Segment D: miniboss — Forklift Foreman (75 seconds)

Arena lock. Foreman charges in a straight line (dodge sideways), summons two grunts at 50% HP, and enters enrage (faster charge, no grunt adds). Throwing grunts into the forklift deals bonus damage — teaches environmental synergy under pressure.

Segment E: boss approach (60 seconds)

Auto-scroller on a moving barge. Grunts jump aboard from speedboats; density stays at three active enemies max due to narrow deck. Transitions into the stage boss without a loading screen.

Playtest metric: 70% of co-op groups reach the boss with at least one life remaining on Normal; 30% wipe on the Foreman and use a continue — that is the intended difficulty sweet spot.

Subgenre decision table

Choose classic belt-scroller for a tight arcade scope with strong couch co-op. Choose modern revival if your team can invest in combo depth and online play. Choose survivor-style only if you are deliberately departing from traditional brawler readability.

Common pitfalls

• Off-screen spawns — enemies attacking from outside the camera feel cheap. Spawn inside frame or telegraph with audio cues.
• Infinite juggles — one player locks an enemy in the air while others idle. Cap juggle hits or add gravity scaling per hit.
• Samey waves — ten rooms of grunts only. Vary composition and arena shape every two fights.
• Damage sponge bosses — five-minute HP bars without phase changes. Add patterns, adds, or vulnerability windows.
• Co-op camera chaos — players on opposite screen edges get hit by off-screen enemies. Rubber-band the camera or teleport stragglers.
• Unclear forward direction — players walk left into a dead end. Use arrows, NPC exits, or obvious lighting on the path.
• Special move dominance — screen-clear supers trivialize every wave. Gate supers behind meter, health cost, or boss-only effectiveness.
• No defensive option — eight enemies with no dodge or block forces passive trading hits. Give a reliable escape.
• Input eating — when eight enemies hit at once, inputs drop. Cap simultaneous hits on the player or grant brief hit invulnerability after knockdown.
• Ignoring audio — rushers need a distinct spawn shout; ranged enemies need charge cues. Visual-only telegraphs fail in co-op chaos.

Production checklist

• Prototype one arena with three enemy archetypes before building six stages.
• Define juggle limits, hitstun decay, and throw invincibility in a tuning doc.
• Script waves as timed compositions; avoid pure random spawn tables.
• Playtest solo and two-player — solo must be viable, co-op must scale.
• Lock screen only after threats are visible; unlock on last enemy death.
• Place environmental weapons before difficulty spikes, not after.
• Cap active ranged enemies at two for readability.
• Target six to twelve minute stages with a boss every three to five minutes.
• Build local co-op first; add online only with stable juggle netcode.
• Record a no-death speedrun path to verify skip and sequence-break risks.

Key takeaways

• Beat-em-ups are forward-marching crowd combat — clear the screen, advance, repeat.
• Enemy archetypes and wave composition are your real level design.
• Co-op synergy (juggles, throws, revives) is the genre’s strongest differentiator.
• Scale difficulty through composition and patterns, not HP inflation alone.
• Classic simplicity and modern combo depth both work — pick one and commit.

Related reading

• Hack-and-slash game design explained — loot ARPG combat with deeper build systems
• Fighting game design explained — one-on-one duels, neutral, and frame data
• Co-op game design explained — shared goals, scaling, and couch-session flow
• Combo systems explained — juggle limits, scaling, and reset rules