GTM Master Plan
Status: living doc. Owner: Oleg + Mariana. Audience: public — we publish our launch thinking in the open.
This is the operational companion to the public essay "An empty world is a sad world." The essay argues the thesis in public. This document commits us to the bets, grades each by how much the evidence backs it, and names what would prove us wrong.
Positioning
Cinevva World is a browser-native, persistent, shared 3D world where you claim a piece of territory and build it by hand, fast. The one-line public pitch: open a link, you're in a shared world in five seconds, claim a spot, and assemble real structures on it with snap-together kits, terrain you sculpt, and one-click building. AI is an optional helper that does the tedious parts, not a requirement. Owning and building your territory is the retention engine; the shared ground around it is the meeting place.
Our honest framing is different from the GDD's "platform, not product." We are launching a game first and a platform second. The platform (creator economy, marketplace) is the second act we earn after the game retains people. Leading with "platform" is the exact mistake that killed Dreams, Core, Project Spark, and Sansar.
The five bets, graded
Bet 1 — Game first, not platform. Confidence: very high. Every survivor (Minecraft, Roblox, Fortnite) led with something fun to do and grew creation on top. Every creation-first sandbox without a game died. The first session must contain a loop with a goal, playable solo, with zero human-made content present. This is the single most evidence-backed decision we have.
Bet 2 — AI manufactures the supply others died waiting for. Confidence: medium, this is the real wager. No studied platform was bootstrapped with generative AI, so we have a mechanism, not a proof. Mechanism: ~67% of two-sided platforms die on the supply side; the classic fix is to fake supply until it's real (DoorDash drove the food, Reddit seeded posts, Roblox hand-built early games for years). We can do that with generation, cheaply and fast. Hard guardrails: cap synthetic seed at ~30% of visible content and convert to real creator content within ~60 days; and remember Decentraland was full of generated stuff and drew ~30 DAU, so AI does the assets, humans design the fun. We are NOT betting AI makes the world fun. We are betting AI frees all our scarce human effort to go into the loop. To be precise about the mechanism: the seed is assembled by hand from the toolset we already ship (terrain, modular kits, scatter) plus a pre-vetted asset library, which is fast and reliable; AI is an accelerant on top, not a throughput dependency, because on-demand generation today takes minutes and the quality is inconsistent. Players and creators assemble; generation is a background ingredient.
Bet 3 — Publish and obey our own kill-criteria. Confidence: high (it's a discipline, not a guess). The metric that predicted survival in every case was whether the retention curve flattens. See the Metrics & Kill-Criteria doc. If the alpha curve does not flatten, we change the thesis or stop.
Bet 4 — Start narrow (atomic network). Confidence: high. Launch to the smallest group dense enough to be alive on its own, not to a market. Ours already exists: Cinevva users who have generated 3D/music/SFX assets and have inventory to place. Fifty of them in one world at the same hour beats fifty thousand scattered signups.
Bet 5 — Tell the truth about emptiness (frontier framing). Confidence: low, a hunch. A polished quiet world reads as dead. Frame the first cohort as founding citizens of a world that is empty because they're early. Least proven of the five; treat as a positioning experiment, not a load-bearing pillar.
What is genuinely unproven (do not pretend otherwise)
- That the loop is fun. It does not exist in shippable form yet. This is the assumption every dead platform also held. Until alpha retention says otherwise, everything else is scaffolding on an unverified foundation.
- That AI-seeded content reads as worth-doing, not as filler. Pretty is not playable.
- That browser/no-download is a net advantage. It is double-edged: frictionless to enter, frictionless to leave. The burden is entirely on the first 60 seconds.
- That the creator economy retains anyone early. At 50 to 1,000 users, tips and marketplace earnings round to zero. Early retention must come from the loop, not money. Tipping is a Phase 2+ signal, never a Phase 1 retention lever.
Dual-funnel model
Stop optimizing everyone into creation (that rebuilds Dreams). Roblox's real shape is ~90-99% who only play sitting on a thin layer who create. Track two separate funnels with two separate aha moments:
| Funnel | Aha moment | Primary metric |
|---|---|---|
| Player | "There's something fun to do here" | session length, D1/D7/D30 return |
| Creator | "I made a thing and someone saw it" | first object < 60s, creator D7/D30, return-to-edit |
The GDD's headline "30% of visitors place an object" is a creator-funnel metric. It should not be the top-line success measure. The top line is player retention.
Phase map (GTM motion per phase)
- Phase 1 (private alpha, ~50): prove the loop retains. See Alpha Playbook. No public marketing.
- Phase 2 (public beta): Product Hunt + HN + targeted subreddits + creator partnerships + existing email list. Mobile visitors enabled. Tipping live as a signal, not a lever.
- Phase 3 (GA + marketplace): only after retention curve flattens in beta. Marketplace, voice, uploads.
Do not advance a phase until the prior phase's kill-criteria pass.