# Evidence-led Concept Development

Use for finding new game ideas, producing a test shortlist, or evaluating an existing proposal.

## Entry routes

**Evaluate a proposal:** Preserve the owner's intended proposition. Translate it into falsifiable assumptions about players, demand, experience, distribution, monetization, and execution. Compare substitutes and failed analogues; do not search only for confirming examples. Offer a revised concept only when the evidence warrants it and label the change.

**Discover new concepts:** Begin with the business mandate and confirmed studio constraints. Identify distinct player needs and opportunity spaces from market signals. Compare them with balanced light scans; do not require a predefined genre and do not default to the host project's mechanics.

## From opportunity to candidates

1. Link the opportunity to specific insight/evidence records, competing solutions, counter-evidence, and why the gap persists. Missing competitors alone is not demand.
2. Generate meaningfully different solutions: change the player action, session structure, accessibility, strategic depth, social relationship, or delivery model in response to that need.
3. Explain why any mechanic combination improves the intended experience. "Genre A + mechanic B" is not enough. Cosmetic reskins belong to the same concept family unless audience/proposition evidence distinguishes them.
4. Preserve separate install, stay, and pay hypotheses. Do not improve monetization by contradicting the player promise without flagging the trade-off.
5. Apply hard feasibility constraints before ranking. Reusable technology may lower costs but is not proof of demand.
6. Select the smallest useful portfolio of tests. Prefer distinct learning over several superficially different tests of the same assumption.

Generate only as many candidates as are useful; for a broad brief, 3-5 test candidates can be a working target, never a quota. Keep unsupported creative ideas in an explicitly unvalidated idea pool rather than the evidence-backed shortlist. If none qualifies, report that outcome and the next research question.

## Concept Card

- ID / title / version / status:
- One-sentence player proposition:
- Target player, context of play, and unmet need:
- Linked opportunity, insight IDs, sources and snapshot dates:
- Existing substitutes and closest comparables:
- Concrete session: starting situation, player actions, meaningful choices, feedback, payoff, ending:
- Core loop and lightweight meta loop:
- Visible differentiation and reason to switch/try:
- Install hypothesis and credible distribution hook:
- Stay hypothesis and repeat-engagement driver:
- Pay/ad-acceptance hypothesis and value exchange:
- Counter-evidence and reason this gap may be unattractive:
- Confirmed execution fit, dependencies and unknowns:
- Test build: must include / explicitly out of scope:
- Test cost and duration range, assumptions behind the estimate:
- Production implications: content volume, technical needs, LiveOps and UA:
- Most consequential falsifiable assumption:
- Next experiment, success/failure/inconclusive criteria, decision enabled:
- Confidence by demand / concept / execution / distribution claim:

Keep cost estimates assumption-based until the relevant owner validates them. Avoid turning early concept cards into full GDDs.

## Prioritized shortlist

| Priority / concept | Evidence-backed opportunity | Visible distinction | Feasibility | Confidence / main unknown | Next test / cost-time cap | Why this test now |
|---|---|---|---|---|---|---|

Include selected candidates, a concise rejected/held list with reasons, shared dependencies, and total portfolio resource demand. Do not recommend parallel prototypes that require the same unavailable people. Separate "best market opportunity" from "best next test".

A concept is eligible for a test shortlist when there is traceable opportunity evidence, a clear player experience, a plausible bounded execution path, and a meaningful experiment. This eligibility does not mean commercial viability is proven.