Game development stages shown through 3D environment design software
Back to top

Game development stages: a full production cycle

Dmytro Lunov

Written by

Dmytro Lunov Verified author

Head of Delivery and Program Director at Game-Ace

Dmytro leads Game-Ace delivery teams on game development, art production, game design, MVP prototyping, and Unity and Unreal Engine projects.

Published November 1, 2023 Updated June 16, 2026

Game development stages are the structured phases a studio moves through to turn a concept into a released, supported product. The widely used framework has six phases: pre-production, production, alpha QA, beta, launch, and post-launch live ops. Each phase has its own deliverables, team composition, and acceptance criteria.

What are the stages of game development?

Game development stages infographic from pre-production to post-launch live ops

Most studios organize a project around six recognizable stages, each ending in a reviewable artifact. Pre-production produces the Game Design Document (GDD), a proof of concept (POC), and a playable prototype. Production builds the vertical slice and the alpha build, where core systems are feature-complete. The QA-driven alpha and beta phases harden the build through functional, regression, compatibility, and performance testing. Launch covers the release candidate, soft launch in a limited region, and full release. Post-launch live ops keeps the title alive with patches, balance updates, seasonal content, and platform-certification follow-ups. The six-stage view is helpful because it lets producers map budget, team size, and risk to a known timeline. Studios that skip pre-production usually pay for it later in rework during production, and studios that compress beta tend to ship the kind of stability issues that erode reviews in the first week.

Six-stage framework with duration, deliverables, and lead roles:

Stage Typical duration Key deliverables Lead roles
Pre-production 2–6 months GDD, POC, prototype, art bible, tech spec Game designer, art director, tech lead, producer
Production 6–18 months Vertical slice, alpha build, content pipeline Programmers, artists, animators, level designers
Alpha QA 1–3 months Feature freeze, alpha test report, bug triage QA leads, automation engineers, gameplay programmers
Beta 1–3 months Content freeze, beta build, performance pass QA, performance engineers, compliance specialists
Launch 2–8 weeks Release candidate, soft launch, gold master Producer, release engineer, publishing, marketing
Post-launch live ops 6–36 months Patches, seasonal content, analytics dashboards Live ops team, community, backend engineers

Pre-production: GDD, POC, and prototype

Pre-production is the planning phase where a studio decides whether the game is worth building and how it will be built. The phase ends with three artifacts: a Game Design Document (GDD) that describes mechanics, systems, monetisation, and scope; a proof of concept (POC) that validates the riskiest gameplay loop in code; and a playable prototype that lets stakeholders feel the core fantasy. For a mid-budget mobile title this stage runs two to four months, for a multiplayer PC title four to six months. The team is small at this point, usually a game designer, a tech lead, an art director, and a producer. Decisions made here have outsized weight: a vague monetisation model or an undefined target frame rate at the end of pre-production almost always becomes a production bottleneck later. A clear GDD also speeds vendor selection if any production work will be outsourced, because external teams can scope from documented requirements rather than verbal briefs.

Typical pre-production deliverables:

  • Game Design Document covering mechanics, progression, monetisation, and live-ops plan
  • Proof of concept build that validates the riskiest gameplay loop
  • Vertical-slice or whitebox prototype with placeholder art
  • Art bible covering style targets, palette, and reference boards
  • Technical specification with engine choice, target platforms, and frame-rate budget
  • Production plan with milestones, risk register, and cost estimate

Production: vertical slice to alpha

Production is the longest phase and the one that consumes the largest share of budget. The team grows to its full size, often three to five times the pre-production headcount, and content pipelines are pushed to capacity. The phase typically begins with a vertical slice, a polished representative segment of the game that proves the production pipeline can hit the target quality bar. From there, teams build out levels, characters, animations, and systems in iterative sprints. The phase closes when the build reaches alpha, the point at which every planned feature is in the game in some form, even if not yet polished. Production is also where engagement models matter most for outsourcing. A Co-development model lets an external team own a defined system (for example, multiplayer netcode or character animation), while a Team Extension model adds named specialists to the internal pipeline under the studio's own producers. Engine choice (Unity or Unreal Engine) is usually locked at the start of production, because changing later is expensive.

Alpha and beta QA

QA is a continuous activity, but two formal milestones sit between production and launch: alpha and beta. Alpha is feature-complete: every system planned for the release version is in the build, and QA moves from smoke testing to structured test plans. The team runs functional testing of gameplay systems, regression suites after every merge, compatibility testing across the device matrix, and performance testing against the frame-rate and memory budgets. Beta is content-complete: art and audio assets are final, localisation strings are locked, and certification testing for first-party platforms (PlayStation, Xbox, Nintendo, App Store, Google Play) begins. Beta is where build stability is measured against a crash-free session rate target, usually 99.5 percent or higher for a mobile soft launch. Studios that treat alpha and beta as overlapping phases rather than sequential ones ship with fewer last-minute regressions, because performance and stability issues surface while the team still has time to address them.

Test scopes typically covered before launch:

  • Functional testing of gameplay systems and UI flows
  • Regression testing on every merge into the release branch
  • Compatibility testing across the agreed device matrix
  • Performance profiling against frame-rate and memory budgets
  • Localisation testing for each shipping language
  • Certification or store-submission testing for target platforms
  • Load and soak testing for online and multiplayer systems

Launch: release candidate, soft launch, full release

Launch is the shortest stage but the highest risk. It begins with a release candidate (RC), a build that QA and production both believe could ship. From there, most modern projects run a soft launch in one or two regions to validate retention, monetisation, and server load with real players before the worldwide release. A soft launch typically runs four to twelve weeks. Data from this window drives the final tuning pass: economy balance, onboarding flow, ad placement, and server capacity. Once the soft launch hits its target metrics, the team locks the gold master and pushes the global release. Launch coordination also covers store assets (screenshots, trailers, store descriptions), platform certification sign-off, day-one patch planning, and customer-support readiness. A well-rehearsed launch has a documented rollback plan, an on-call rotation for the first 72 hours, and dashboards that show install rate, crash rate, ARPU, and retention in real time.

Post-launch live ops and content updates

Post-launch is the longest stage by elapsed time, often longer than pre-production and production combined. The live-ops team owns patches, balance changes, seasonal content, events, and platform-certification updates. For free-to-play mobile and PC titles, content cadence usually runs two to six weeks between meaningful updates, with smaller hotfixes between them. The team composition shifts: many production specialists rotate off, while live-ops engineers, community managers, data analysts, and a smaller content squad rotate in. Post-launch is also where outsourcing patterns differ from production. A live-ops engagement is usually long-running and lean, with a stable team of named specialists who learn the codebase deeply. Backend reliability, anti-cheat, A/B testing infrastructure, and analytics pipelines become the dominant engineering surface. A studio that documents the production codebase well during the previous stages spends much less effort onboarding live-ops engineers, which is a hidden but real production cost saving.

Engagement models mapped to project stages:

Stage Full-cycle delivery Team Extension Co-development
Pre-production Vendor owns GDD, POC, prototype Specialists join client GDD work External team scopes a system
Production Vendor builds full game to alpha Named developers in client sprints External team owns a feature area
Alpha/beta QA Vendor runs full QA cycle QA engineers embedded in client team External QA owns the test plan
Launch and live ops Vendor coordinates release Specialists support release rotation External team owns live-ops slice

Roles, tools, and risks across game development stages

Each stage has a recognizable role mix and a recognizable risk profile. Pre-production risk is scope ambiguity, production risk is pipeline throughput, QA risk is regression debt, launch risk is platform certification and server capacity, and live-ops risk is content cadence slipping. Mapping roles and risks to stages early lets producers staff and budget against known failure modes rather than reacting to them.

Lead roles and tools by stage:

  1. Pre-production roles: game designer, tech lead, art director, producer. Tools: Confluence, Miro, engine prototypes, paper design
  2. Production roles: gameplay programmers, technical artists, animators, level designers. Tools: Unity or Unreal Engine, Perforce or Git LFS, Jira
  3. QA roles: QA lead, automation engineer, performance engineer, compliance specialist. Tools: TestRail, Jenkins, device labs, profilers
  4. Launch roles: release engineer, publishing producer, community lead. Tools: store back-offices, dashboards, on-call rotation tools
  5. Live ops roles: live-ops engineers, data analyst, community manager. Tools: A/B testing platforms, analytics pipelines, CMS for content

About Game-Ace

Game-Ace has been delivering custom production work since 2005, with 120+ in-house specialists and 200+ delivered games across mobile, PC, web, and iGaming. The studio supports projects at any stage of the framework above, from GDD authoring in pre-production to live-ops engineering after release. For studios that want a single accountable vendor across the cycle, Game-Ace offers full-cycle game development. Selected work across genres and platforms is available in the Game-Ace portfolio, with further case studies under NDA. To discuss scope, milestones, or a specific stage of the production cycle, contact our team.

Game development stages: frequently asked questions

The widely used framework has six stages: pre-production, production, alpha QA, beta, launch, and post-launch live ops. Some studios merge alpha and beta into a single QA phase, which gives the five-stage view often seen in older references.

Pre-production typically runs two to six months. A mid-budget mobile title sits closer to two to four months, while a multiplayer PC title with custom backend systems usually needs four to six months to reach a validated GDD, a working POC, and a playable prototype.

Each stage closes with reviewable artifacts that drive the milestone gate. A typical delivery list across the cycle looks like this:
  • Pre-production: Game Design Document, art bible, technical specification.
  • Pre-production: proof of concept build validating the core loop.
  • Production: vertical slice that hits the target quality bar.
  • Production: feature-complete alpha build under version control.
  • Alpha and beta QA: test reports, performance pass, certification submission.
  • Launch: release candidate, soft-launch metrics, gold master.
  • Post-launch: patch cadence, seasonal content drops, analytics dashboards.
Documenting these artifacts during production also makes live-ops onboarding faster once the title is live.

The most common entry points are late pre-production and early production. Joining at the end of pre-production lets an external team scope from a finalised GDD and shape its own backlog with the studio's producers. Joining at the start of production lets the external team grow with the pipeline rather than catching up to a moving target. Late entries are possible but cost more in onboarding time, because the external team has to absorb decisions that are already locked. Game-Ace typically scopes either a Team Extension model with named specialists or a Co-development model where an external squad owns a defined system end to end.

Scope is locked through a feature-complete gate. At alpha, every system planned for the shipping version exists in the build, even if not polished. After that gate, new features are blocked unless a formal change-request process accepts them, usually with a matching cut elsewhere in the backlog. The gate is enforced by the producer and tracked against the original milestone plan from pre-production.

Soft launch validates four things that internal QA cannot: real-world retention, monetisation against live players, server capacity under organic load, and store-funnel conversion. It runs in one or two regions for four to twelve weeks and feeds a final tuning pass on economy balance, onboarding, and ad placement before the global release. Studios that skip soft launch usually carry retention or server risks into worldwide release, where the cost of a fix is much higher.

Post-launch runs as long as the title earns its operating cost. For a free-to-play mobile or PC title, that is often 24 to 36 months of active live ops, with a smaller maintenance tail after that. For a paid premium release, post-launch is shorter and usually focused on bug-fix patches, balance updates, and one or two content expansions. The live-ops team is leaner than the production team but stays consistent, because deep codebase knowledge is the main driver of patch velocity.
Average rating 4.7 / 5. Votes: 188
Related posts
Metahuman game character standing near realistic cars in unreal engine scene How MetaHuman changes game character development Game development for startups rocket in space Game Development for Startups: How to Build a Game with a Small Budget in 2026 Tower defense games image preview Engineering the Next Generation of Tower Defense Games Idle game development preview img How to Turn Idle Game Development into a Scalable, Long-Term Revenue Product Gamification in recruitment preview Key Trends Shaping Gamification in Recruitment for 2026 and Beyond
Futuristic game robot running through a purple portal
Get in touch
menu
Get in touch
Game-Ace logo loader