One-person game studio: the funnel, in public
Every idea I collected, every prototype that got built, every one that survived play, and exactly how far the pipeline has reached — including the rows that are still zero. Updated as the numbers move.
Live: I am working this direction right now. The board updates whenever a real result arrives. I am actively building and shipping in this direction, so this board changes. Every entry in the update log below is a real event, not a content schedule.
19 / 40
ideas that became something playable
Forty raw ideas went in. Nineteen exist as playable prototypes. Everything after that is a much smaller number — which is the honest shape of a studio funnel run by one person.
Ideas captured
40
Raw, unfiltered, in my own words
Playable prototypes
19
12 of them built in half a day
In full production
3
Tutorials, biomes, polish, soundtrack
Submitted to a store
1
Internal testing, not public yet
Public releases
0
Nothing has shipped to players yet
Revenue to date
€0
No public release means no income
The pipeline, stage by stage
Each bar is how many things reached that stage. The line under each bar is how many survived the gate above it. A stage at zero is drawn as zero — the whole value of a board like this is that it cannot flatter itself.
Written down without filtering. Ideas get graded later, never discarded at capture.
50% of the stage above made it through this gate
Graded on buildability multiplied by fun-potential. About half made the cut.
95% of the stage above made it through this gate
A self-contained build: one mechanic, one input axis, playable on desktop and phone.
BublDog · TwoBallz · Sewer Worm · Zombie Steve · H2GO · FlutterCat · Laser Crystals
37% of the stage above made it through this gate
Played properly, written up, and given a tier. Playtesters included my own children.
43% of the stage above made it through this gate
Judged fun enough to deserve tutorials, difficulty curves, biomes, polish and an original soundtrack.
BublDog · TwoBallz: Super Rush · Zombie Steve
33% of the stage above made it through this gate
Wrapped for Android, signed, uploaded, with a paid remove-ads product configured.
TwoBallz: Super Rush
0% of the stage above made it through this gate
Live where a stranger can find and install it. Internal testing does not count.
Money actually received. Ad impressions and paid unlocks both count; neither has happened.
View the same data as a table
| Stage | Count | Survived the gate | State |
|---|---|---|---|
| Ideas captured | 40 | — | Done |
| Shortlisted to prototype | 20 | 50% | Done |
| Playable prototype built | 19 | 95% | Done |
| Graded after real play | 7 | 37% | Done |
| Promoted to production | 3 | 43% | Done |
| Store build submitted | 1 | 33% | In progress |
| Released publicly | 0 | 0% | Not started |
| Earning revenue | 0 | — | Not started |
Update log
Newest first. Only real events — no content schedule.
- Milestone 4 Sep 2026 Board published. Baseline numbers locked so every later change is visible against them.
- Shipped Aug 2026 First game wrapped for Android and uploaded to an internal testing track, with the paid remove-ads product configured and active.
- Milestone Aug 2026 Developer account verified after identity and website checks — the step that gates every store release.
- Lesson Aug 2026 Playtesting with my own children changed real features, not just polish. The most useful feedback on the board came from people who cannot read a design document.
- Milestone Aug 2026 Three prototypes promoted to full production after grading; four graded prototypes held back as "seed there, not enough yet".
- Milestone Aug 2026 Twelve playable prototypes built in half a working day, while an album was uploading in a parallel project.
How to read this board
- 01
The top of the funnel is cheap and the bottom is expensive. Collecting forty ideas cost an evening. Getting one game to a store took weeks of packaging, policy, billing and console work that has nothing to do with making a game.
- 02
Building stopped being the bottleneck; judging did not. Twelve prototypes in half a day is real. Deciding which three deserved production took longer than building all nineteen.
- 03
The last two rows are zero, and they are the only rows that pay. Every claim about a solo studio should be checked against its bottom two rows, including mine.
- 04
The drop from 19 to 7 is not failure. It is the funnel working: a prototype exists to answer one question honestly, and "no" is a valid answer that cost twenty minutes.
What it has cost, in cash
The surprising part of an AI-era solo studio is not the speed — it is that the marginal cash cost of a game is close to nothing. The real cost is attention, and it is spent almost entirely after the game is finished.
| Item | Cash cost | Why |
|---|---|---|
| Store developer registration | $25 one-off | A single lifetime fee for the account, not per game |
| Ad network account | €0 | Free to open; it pays out rather than charges |
| Engine, framework, licences | €0 | Plain HTML5 canvas, one self-contained file per game, no engine |
| Art and sound assets | €0 | Drawn in code, sound synthesised in the browser; no asset purchases |
| Original soundtracks | €0 marginal | Produced in-house by my own music project, so nothing new was bought |
| Hosting for the test playground | €0 marginal | A folder on a server I already run for other projects |
| AI coding assistant | shared | One subscription used across the whole portfolio; not attributable to games alone |
Total cash spent that exists only because of the games: one $25 store registration. Everything else was already paid for, free, or produced in-house — which is exactly why the interesting constraint is no longer money.
The actual stack
What is genuinely in use, not a list of everything that exists in the category.
Build
-
HTML5 canvas, hand-rolled
One self-contained file per game — no engine, no build step, instant to publish anywhere
-
AI coding agent
Writes the prototype from a tight design brief; I direct, playtest and tune the feel
-
Capacitor
Wraps a finished web game as an Android build when it earns a store release
Distribute
-
Private web playground
Where every prototype lives first, noindexed, for play and judgement
-
Google Play Console
Store listing, internal testing track, billing product
-
Browser game portals
The planned next channel: they bring the audience and share ad revenue
Monetise
-
Ad network, interstitial on game over
The main free-tier income model for casual games
-
One paid unlock
A single small purchase that removes ads and unlocks the full game
Cross-project leverage
-
My own music project
Original soundtracks come from the label instead of a licence fee
-
My own playtesters
My children play everything and give the least polite, most useful feedback available
What happened, when
-
August 2026
Forty ideas captured, half shortlisted
Collected raw over a short period, then graded on buildability against fun-potential.
Done
-
August 2026
Twelve prototypes in half a working day
Each one a single-mechanic playable build, made to answer one question: is this fun enough to deserve production?
Done
-
August 2026
Seven graded, three promoted
The three favourites went into full production: in-game tutorials, difficulty curves, biomes, polish and original soundtracks.
Done
-
August 2026
Store account verified, first build uploaded
Developer account verified, the first game wrapped for Android and pushed to an internal testing track with a paid remove-ads product configured.
Done
-
Next
First public release
Moving the first game from internal testing to a public listing — the step that turns every number below it from zero into something measurable.
In progress
-
Then
Web portals and a channel
The same finished games distributed through browser portals that bring their own audience and share ad revenue, plus gameplay video as the marketing layer.
Not started
Behind the board
The detailed layer
Everything above is the public board and stays public. The layer below is the working material behind it — the part that is only useful if you are actually building something and want to skip the months of figuring it out.
The idea bank and the grading sheet
🔒All forty ideas as written, plus the scoring that sent twenty forward and left twenty behind.
- Every idea in its original one-line form, ungroomed
- The buildability × fun-potential grid and how each idea scored
- Why the four rejected favourites were rejected, in specifics
- The tier system I use to grade an idea without discarding it
Mechanics teardowns of the production three
🔒The design decisions that turned a prototype into something people finish, written up per game.
- Core loop, input model and the one question each prototype existed to answer
- The difficulty curve and biome structure, with the numbers I tuned them to
- The polish pass that makes a build feel finished, item by item
- Playtest feedback from real players and exactly what changed because of it
The prompt and process pack
🔒How a design brief becomes a playable build in a session, including what fails.
- The design-brief template a prototype is built from
- The prompting sequence that produces a working single-file game
- The specific failure modes I hit and the checks that catch them
- The refactor discipline that stops a working game breaking silently
The store and monetisation runbook
🔒The unglamorous half: packaging, policy, billing and ads, in the order it has to be done.
- Web build to signed Android release, step by step
- The console setup: testing tracks, billing product, the fields that block a review
- Ad placement that pays without wrecking a casual game
- The traps that cost me days, so they cost you none
The revenue log
🔒Real income figures per game and per channel, updated as they arrive — starting from the honest zero above.
- Income by game, channel and month, as received
- What each channel actually paid against what it promised
- Cost per release once the time is counted honestly
- The point at which any of this pays for itself, if it does
How access and price work
-
While the direction is live
The current version of the detailed layer is at full price, because it keeps growing: every new result, number and lesson lands in it while I am still doing the work. Buy once and the updates to that version are included.
You are here
-
When the next version ships
Each version is sold up to a point, then replaced by a better one built on everything learned since. When that happens the previous version becomes free, or nearly. You can start on the old one, see whether the material is worth anything to you, and upgrade only if it is.
-
When I stop working the direction
The board is marked closed, the numbers freeze, and the price drops again. What you get is a complete, finished record rather than a live feed — worth less, and priced accordingly.
-
About a year after that
Everything opens completely and becomes free to read. Nothing here stays behind a wall forever; the point of an archive is that somebody learns from it.
Bundled with
The AI music label board — The soundtracks in these games come from that project — buying one case gets you the other at a discount.
This layer does not exist as a product yet, and I am not going to pretend it does. It is being written from the working material as the pipeline moves. Leave your email and you will hear when it is ready and what it costs — nothing else.