Most games die before they ship. Not because the idea was bad, but because the process was invisible. You can't manage what you can't see. This guide maps the five stages every game passes through — concept, pre-production, production, polish, launch — so you can spot the bottlenecks before they kill your timeline.
Stage 1: Concept — Prove the Fun in 2 Weeks
Don't write a 50-page design doc. Build a vertical slice that answers one question: is the core loop fun? Use Unity or Godot for speed. Target 2 weeks max. If you can't make it fun in 14 days, the concept won't survive 14 months. Kill it early. Save your runway.
💡
TipPrototype with programmer art. Ugly is fine. Boring is fatal.
Stage 2: Pre-Production — Scope Before You Build
This is where indies win or lose. Define every system, asset, and milestone. Lock the feature set. Create a production schedule with buffers. A 6-month indie project needs 2 months of pre-production. AAA spends 1-2 years here. Skip this, and you'll rewrite code you already shipped.
| Phase | Indie Timeline | AAA Timeline | Key Deliverable |
|---|---|---|---|
| Concept | 1-2 weeks | 3-6 months | Vertical slice |
| Pre-production | 1-3 months | 12-24 months | Design doc + schedule |
| Production | 3-12 months | 2-4 years | Alpha build |
| Polish | 1-3 months | 6-18 months | Gold master |
| Launch | 2-4 weeks | 3-6 months | Live ops ready |
Stage 3: Production — The Grind Where Games Get Made
Daily builds. Weekly playtests. Zero tolerance for broken main branch. Use Git with trunk-based development. Automate testing in CI. Unity teams: addressables for content streaming. Unreal teams: gameplay ability system for scalable mechanics. Outsource art only after pipelines are proven. Never outsource core gameplay.
Stage 4: Polish — Where Good Becomes Great
No new features. Only bugs, juice, and balance. Screen shake. Sound timing. Input latency. Difficulty curves. This stage separates shipped games from abandoned repos. Allocate 20% of total time minimum. Use telemetry to find where players quit. Fix those spots first.
"Polish isn't the last 10%. It's the second 90%.
— Anonymous dev, shipped 12 titles
Stage 5: Launch — The Day One Checklist
Steam page live 30 days prior. Wishlist campaign running. Press keys distributed. Build uploaded to all platforms 72 hours before launch. Rollback plan documented. Discord mod team briefed. Post-launch patch schedule written. Launch isn't the finish line. It's the start of live ops.
⚠️
WarningDon't launch on Friday. No weekend support. Tuesday-Thursday only.
Engine Choice in 2026: Match Tool to Team
Unity: best for 2D, mobile, small teams, rapid iteration. Unreal Engine 5: best for 3D, high-fidelity, teams with C++ experience. Godot 4: best for open-source purists, 2D/3D hybrid, zero licensing risk. Don't choose based on hype. Choose based on your team's strongest language and target platform.
✦
Your next step: pick one stage you're weakest at. Spend this week fixing its process. Not the game. The process. A better pipeline ships better games. Start with a 2-week vertical slice. Prove the fun. Then build the machine that delivers it.
