How a small studio decides which game to build next

June 27, 2026

The most expensive decision a small studio makes isn’t a line of code. It’s choosing which game to spend the next few months of your only lives on. Pick wrong and no amount of polish saves it. So before we committed to our next project, we did something unglamorous: we put six concepts on a spreadsheet and forced ourselves to score them.

This is that framework — and the uncomfortable finding it produced.

The six candidates

We had six ideas on the table, each written up with a concept doc, a rough GDD, a level-design sketch, and — crucially — a business-model estimate:

Ragdoll — assemble parts, and a creature is born.
GravitySort — a sorting puzzle where physics, not taps, moves the pieces.
Sock Pair — match mismatched socks.
Splika — a “reverse Suika,” slicing shapes instead of merging them.
StackCrash — the hook is destruction, not construction.
Muddy — an ASMR color-mixing toy.

Six plausible games. Only one worth building first. The spreadsheet’s job was to tell us which — and to stop us falling for the idea that felt coolest.

The five axes we scored

Every concept got a score on the same five questions:

Can a stranger understand it in five seconds? If the core can’t be grasped from a first screen or a short clip, it dies in the feed. We scored this 1 to 5, and we were strict.
Is there an addiction hook? Not “is it fun once” — is there a reason to tap one more time.
Is it different enough to defend? A clone-able idea with no moat gets buried by whoever has the bigger user-acquisition budget.
What’s the production risk? The single biggest flag here was one word: physics. A physics-based feel is a multi-week tuning problem, not a feature you bolt on.
What’s the realistic revenue ceiling? Annual net, honestly estimated — not the fantasy number, the likely one.

That last axis is the one small studios skip, and it’s the one that hurts most when you skip it.

What the scores said

Ragdoll came first. Comprehension 5/5 — “parts become a creature” reads instantly. No physics, so a 2-to-3-week prototype in a 2D engine was realistic. The merge sub-genre it sits in grew about 80% year-over-year in 2025, and the honest revenue ceiling landed at roughly $1M-$3M a year. Its one real risk was content: the loop needs 50-plus creature types before novelty stops carrying it, so the work is front-loaded into variety.

Sock Pair came second on the strength of the same 5/5 comprehension and a viral hook (“mismatched socks”) that works with zero game context — you’d get the joke from a thumbnail. A one-week HTML5 demo was enough to test it. What held it back was the ceiling: the entire sorting niche it competes in is only around $3M a year globally. That makes it a “ship fast, validate, move on” project, not a flagship.

Then came the interesting part.

The most technically impressive concept, StackCrash, ranked fifth. Its reverse hook — destroy instead of build — is clear and clip-friendly. But the honest revenue ceiling was $670K-$1.1M a year, and our team’s operating cost is $320K-$390K a year. A game that, at its ceiling, barely clears the cost of the team building it is not a business — it’s a hobby with extra steps. Add a four-to-eight-week window before the big casual publishers could clone it, and the math stopped making sense.

Muddy ranked last for the same reason, only starker: a median revenue ceiling of $234K a year — below what it costs to keep the lights on. The doc’s own verdict was blunt: a personal portfolio piece, not a team business.

The rule that saved us weeks

Three of the six — GravitySort, Splika, StackCrash — were physics-based. For those, we wrote down a hard rule before scoring anything else:

No UI. No content. No production. Not until ten players play the raw physics prototype and seven of them say “one more.”

The feel of a physics game is the game. Tilt sensitivity, restitution, friction — those aren’t polish, they’re the whole product, and no amount of art rescues a feel that doesn’t land. Entering full development before the feel test passes isn’t ambition. It’s waste. Writing that rule down meant a promising-on-paper physics idea couldn’t drag us into a two-month build on the strength of a mockup.

Why the boring answer was the right one

The concept that won wasn’t the flashiest. It was the one that a stranger understood instantly, that we could prototype in weeks instead of months, that sat in a growing market, and whose honest revenue ceiling comfortably cleared the cost of making it. Every one of those is a small, checkable fact — not a gut feeling.

We didn’t always work this way. An earlier project, a food-sorting puzzle we built as a deliberate reverse-design study of an existing hit, taught us the value of validating a market and sharpening our craft before betting the studio on an original. The spreadsheet is what that lesson looks like written down.

Taste still matters. But taste tells you which ideas are worth scoring — it shouldn’t be the thing that picks the winner. For a studio with a finite number of shots, the framework isn’t bureaucracy. It’s how you make sure the shot you take is one worth taking.

Want to know the moment our next game is playable? Get launch updates →