top of page
264293.jpg

Designing a Scalable Live Service System for a Persistent Dungeon Game

A seasonal live service model built to drive long-term retention without resets, power creep, or economic instability

Role: Live Service Design

Focus: Live service Architecture • Player Retention • Seasonal System

The game context

This project explores the live service architecture of a systemic dungeon-based action RPG built around long-term player progression and emergent gameplay designed in Case studyI (Systemic game design).

 

The game is based on:

  • mastery-driven progression (no fixed classes)

  • a persistent vertical dungeon

  • modular systems interacting at runtime

 

Unlike traditional live games, the system does not rely on seasonal resets or power

 

Why This Changes Live Ops Design

 

Traditional live service models rely on:

  • seasonal resets

  • power escalation

  • content replacement

 

This system requires a different approach:

  • progression must persist

  • systems must remain stable

  • the world must evolve without breaking the player journey

multi level dungeon
264293.jpg

Design challenge

How design a live service that:

  • sustains long-term engagement without resetting players

  • evolves the game without introducing power creep

  • maintains a stable economy over time

  • encourages players to return without relying on FOMO

My Role

  • Designed the full live service architecture

  • Defined seasonal structure and content cadence

  • Built daily, weekly and long-term retention loops

  • Integrated economy protection into live rewards

  • Ensured alignment between gameplay systems and live ops strategy

Game design
264293.jpg

Design principles

01

No Player Resets
Progression is persistent. Players build identity over time.

02

No Power Creep
Progression is horizontal (efficiency, control, mastery), not vertical inflation.

03

Economy-First Design
Live rewards never destabilize the economy.

04

Discovery Over FOMO
Players return to explore and optimize, not because they are forced to.

05

World Evolution, Not Content Replacement
Seasons evolve the world instead of replacing it.

Live service architecture

ChatGPT Image 31 mar 2026, 16_25_22.png

Duration and structure:

  • 4 Seasons per year

  • 12 Weeks per season

 

Each season follows a structured progression:

  • Phase 1 — Discovery
    Lore introduction, anomalies appear, early speculation

  • Phase 2 — Expansion
    Meta shifts, spawn mutations, new craft unlocks

  • Phase 3 — Climax
    Server-wide events and unique monster encounters

  • Phase 4 — Closure
    Narrative resolution and controlled resource conversion

 

What Changes vs What Persists:

Changes

  • environmental anomalies

  • spawn behavior

  • craftable categories

  • seasonal materials

  • world events

Persists

  • player progression

  • mastery systems

  • core mechanics

  • economic structure

 

This ensures long-term investment without invalidating player effort.

ChatGPT Image 31 mar 2026, 16_27_59.png

Weekly Loop

  • Rotating unstable dungeon layers

  • Guild bounties with targeted objectives

  • Environmental gameplay modifiers

Outcome

  • drives player movement across the dungeon

  • creates controlled resource consumption

  • evolves the meta without breaking balance

 

Daily Loop

  • 3 lightweight objectives

  • first-run XP bonus (no drop boost)

  • capped rewards

Design Goal

Encourage frequent engagement without creating inflation or mandatory grind.

 

Long-Term Retention

  • unique monsters (non-farmable, condition-based)

  • deep dungeon progression

  • seasonal system unlocks (not direct power)

 

Insight

Retention is driven by:

  • discovery

  • mastery

  • optimization not by repetition or artificial scarcity.

ChatGPT Image 31 mar 2026, 16_31_35.png

Economic Stability Strategy

  • No gold trading between players

  • Controlled resource sinks (repair, consumables, optimization)

  • Craft-driven value instead of direct loot injection

  • Diminishing returns on repetitive farming

Outcome

  • stable long-term economy

  • no inflation spikes

  • no veteran advantage collapse

 

accessible progression for new players

ChatGPT Image 31 mar 2026, 16_32_14.png

Unique Monster System

  • One instance per shard

  • No respawn

  • Triggered by hidden conditions

  • Uncertain reward outcomes

Impact

  • creates emergent social events

  • encourages cooperation

  • prevents content farming

 

Seasonal Material Model

  • unlocks crafting categories (not items)

  • integrates into existing systems

  • converts at controlled value post-season

Impact

  • avoids power spikes

  • preserves economic balance

  • maintains long-term progression consistency

264293.jpg

Product Thinking

Key Insight

  • Retention is driven by system depth, not content volume

  • Stability is more valuable than short-term engagement spikes

  • Player identity persistence increases long-term investment

  • Live ops must be aligned with economic systems, not layered on top

Metrics and KPIs

To validate and balance this system, I would track:
01 - Retention
  • D1 / D7 / D30 retention

  • Weekly Active Users

02 - Progression
  • Average dungeon depth reached

  • Mastery progression distribution

03 - Economy
  • Resource inflation rates

  • Material circulation vs sink ratio

04 - Engagement
  • Event participation rate

  • Weekly loop completion rate

05 - Meta Health
  • Build diversity

  • System usage distribution

264293.jpg

Final Takeaways

This model demonstrates how a live service can:

  • sustain long-term engagement

  • protect player investment

  • maintain economic stability

  • avoid common pitfalls like power creep and forced retention

The goal is not to increase playtime artificially,but to create a system players want to return to.

bottom of page