From 25 mechanic prototypes to 14 live titles: the CMZ puzzle pipeline

Challenge

• 1

  In puzzle, the studios that win are the ones that can test the most ideas with the least waste. A small team cannot afford to rebuild progression, boosters, ads, IAP and analytics for every concept — yet every prototype needs those systems before its numbers mean anything.

• 2

  By mid-2025 the studio had a library of mechanic prototypes (the m-series: screw jam, jigsaw, block legends, berry blast, water sort, cake sort, cupcake sort, block fit …) but each lived in its own project. Turning any of them into a store-ready title took months of duplicated work.

Objective

1. 1

   One Unity codebase (the CMZ Puzzle Framework) that any mechanic can plug into.

2. 2

   A level editor and content pipeline shared across titles.

3. 3

   Ads, IAP, remote config and analytics wired in from the first playable build.

4. 4

   A release cadence measured in weeks, with every title test-launched on Google Play to collect real cohorts.

Approach

1. 1

   Separate the mechanic from everything around it: a small contract (load level, move, win/lose, hint, booster port) is all a mechanic implements; the framework handles meta, monetization and tracking.

2. 2

   Make content a data problem: the level editor (Dec 2025) exports level data the framework loads directly, so designers ship chapters without engineering time.

3. 3

   Ship to the store early: instead of long internal alphas, each title goes to Google Play as a test launch so retention, funnel and creative data come from real players.

4. 4

   Keep the m-series alive as a library — killed mechanics are archived with their code, levels and learnings and revisited when the market moves.

Execution

1. 1

   October 2025: Blast Jam: Block Away, Color Blast Saga and Cupcake Sort become the first three titles on Google Play, proving the shared booster and progression systems.

2. 2

   December 2025 – March 2026: the level editor and the core framework repository are established; Arrows Lace Puzzle and Marble Away start on the new base.

3. 3

   June – August 2026: eleven titles ship in three months — Wool Loop, Fill The Car, Block Flow, Arrows Lace, Marble Away, Zen Garden, Cube Tidy, Hole Crowd, Coffee Man Run, Vacuum Conveyor — across sort, conveyor, arrow-logic, crowd-jam and even a runner.

4. 4

   September 2026: Jelly Jam launches as the fourteenth title; Cube Tidy receives a hybrid meta layer as the template for the next generation of products (pd-series).

Experiments

• 1

  Conveyor loop variants: marbles (Marble Away) versus wool (Wool Loop) versus cars (Fill The Car) — same core timing, different themes and pacing, compared on D1 and session length.

• 2

  Arrow-logic depth: Arrows Lace (2D ropes), Block Flow (3D chains + conveyor) and Vacuum Conveyor (sand physics + collection) tested how much layered feedback a logic puzzle can carry.

• 3

  Genre control: Coffee Man Run, a stack runner, was shipped on the same framework to check that the systems hold outside board puzzles.

Result

• 1

  Fourteen puzzle titles on the framework — sixteen studio titles in total — live on Google Play in under twelve months, from a team small enough to fit around one table.

• 2

  New concepts reach a store-ready test launch in weeks rather than months because progression, monetization, remote config and tracking are inherited.

• 3

  Nine product builds (pd-01 … pd-09) are in the pipeline on the same base, each with an explicit KPI gate before further investment.

• 4

  Per-title KPIs (installs, retention, monetization) are shared with publishers and investors under NDA and intentionally not published here.

What we learned

• 1

  A framework earns its keep on the third title, not the first. Design the mechanic contract for the game you have not built yet.

• 2

  Test launches beat internal opinions: three weeks of real cohorts settle debates that months of playtests cannot.

• 3

  Theme is cheap, mechanic is expensive — reskinning a proven loop (marbles → wool → cars) is a legitimate experiment when the question is audience, not fun.

• 4

  Killing fast is a feature. Most of the 25 mechanics did not ship, and the studio is healthier for it.