Three stages, and permission to stop after the second.
Fifty thousand pounds and a licensing conversation, then a quarter of a million and a web game, then two to five million and a native app — and the third one only happens if the second clears simulation-tier retention. Written down in advance so that nobody has to be brave about it later.
The three stages, side by side.
| Stage 1 — licence in | Stage 2 — the Chapter Game | Stage 3 — native mobile | ||
|---|---|---|---|---|
| What it is | A branded content pack inside an existing farming simulation | A web and HTML5 game on this platform | A free-to-play native app on iOS and Android | |
| Cost | £50,000–200,000 | £250,000–600,000 | £2m–5m | |
| Duration | 3–9 months | 6–12 months | 18–30 months | |
| Team | A producer and a licensing lawyer | 6–10 people | 20–40 people | |
| Who builds it | Somebody else, under licence | A studio, or this platform’s own team | A studio with live-ops experience, which is a different skill | |
| Server model | Theirs | Server-authoritative, mandatory | Server-authoritative, with a live-ops layer | |
| Risk | Lowest by a distance | Real but survivable | Company-threatening | |
| What it proves | That the audience turns up at all | Retention, the economy, and whether the loop works | Nothing new. It scales what Stage 2 already proved | |
| Gate to the next stage | Any measurable uptake | D1 45% and D30 20%, simulation tier | None. There is no stage four |
Stage 1 — licence into an existing farming simulation.
The shape
- Cost
- £50,000–200,000Mostly legal, production management and asset creation. Very little engineering.
- Duration
- 3–9 monthsDriven almost entirely by the licensing negotiation rather than the build.
- Team
- A producer, a licensing lawyer, an art contractorNo studio, no engineers, no live operations.
- What ships
- A map, a set of buildings, a vehicle livery, a voice packInside a game that already has millions of players and an existing economy.
Why it is first
- It answers the only question that matters
- Does this audience play farming games?Nobody knows. Ten million people watching a programme about a farm is not evidence that they will play one.
- It costs a fifth of the next stage
- And produces a real numberDownloads, playtime, and whether anybody finished the map.
- It is reversible
- A licence expiresIf it does not work, the licence lapses and nothing has been built that has to be maintained.
- It builds the relationship
- With a studio that already shipsWhich is worth more than the revenue.
What it cannot do
- Test the loop
- NoIt is somebody else’s loop. Cocking It Up does not exist in it and could not be added.
- Own the player
- NoThe relationship, the data and the store are all the licensee’s.
- Prove retention for Stage 3
- NoRetention inside somebody else’s game says nothing about retention in yours.
Stage 2 — the Chapter Game, on the web.
The shape
- Cost
- £250,000–600,000Six to ten people for six to twelve months, plus servers and a year of running.
- Duration
- 6–12 monthsIncluding a closed test with real members before anything is public.
- Team
- 6–10 peopleTwo engineers, a server engineer, a designer, an artist, a producer, a QA, and a live-ops person from month four.
- Platform
- Web and HTML5No app store, no review cycle, no thirty per cent, and a URL that can be linked from a chapter page.
- Server model
- Server-authoritative, mandatoryEvery yield, price, failure and currency movement is computed on the server. The client renders and nothing else.
Why server-authoritative is not negotiable
- The economy is the product
- A client-authoritative economy is cheated within a weekAnd a cheated economy in a game about honest farming is a particularly embarrassing way to fail.
- The failure rate must be verifiable
- One cycle in six, provablyIf the rate lives on the client, nobody can be told it is the same for paying players, because it would not be.
- The vote must be auditable
- Server-side, logged, immutableA vote whose tally lives on a client is not a vote.
- It costs money
- Roughly 20–30 per cent of the buildAnd it is the line item that gets cut in every project that later has to be rebuilt.
What Stage 2 has to prove
- D1 retention
- 45 per cent or betterThe simulation-tier benchmark. Below it, the first session is not working.
- D30 retention
- 20 per cent or betterThe one that actually decides whether a native build is defensible.
- Session length
- 3–6 minutes, several times a weekLong sessions in this design would mean the pacing is wrong.
- Paying share
- Any number at allA cosmetic-and-time economy converts a larger share at a lower value. If nobody pays anything, the economy is wrong rather than the game.
Stage 3 — native free-to-play mobile, conditionally.
The shape
- Cost
- £2m–5mBuild, launch and the first year of live operations, which is the half everybody forgets.
- Duration
- 18–30 monthsTo a soft launch, in two or three markets, before anything global.
- Team
- 20–40 peopleIncluding a live-ops team that does not stand down after launch, ever.
- Store terms
- 15–30 per centWhich is the reason Stage 2 is on the web and this one is not first.
The gate
- It happens only if
- Stage 2 clears D1 45% and D30 20%Those are simulation-tier benchmarks rather than aspirations, and they are written down before the money is spent.
- If it does not clear them
- StopKeep the web game running, keep the licence income, and do not spend two million pounds proving the same thing twice.
- Who decides
- The numbers, not the roomThe point of writing the gate down in advance is that nobody has to be brave in a meeting eighteen months from now.
What it adds, and what it does not
- Adds
- Reach, push notification, store discoveryAnd a substantially larger audience of children, which is why the Children’s Code work has to be done before this, not during it.
- Does not add
- Any new mechanicStage 3 is Stage 2 with more people playing it. If a mechanic needs Stage 3 to work, it does not work.
- Requires
- A fresh DPIABefore launch, and before any significant change afterwards. Not a formality.
Stage 3 is the only stage that could damage the estate financially, and it is the only one with a written condition attached. That is not a coincidence.
The estimate, and how much to trust it.
These are ranges built from published industry figures for team size, duration and day rate, applied to a scope described on these pages. They are not a quote, nobody has priced this, and no studio has been approached.
The number most likely to be wrong is Stage 3, and it is most likely to be wrong in the direction of being too low, because live operations for a free-to-play mobile game is a permanent cost that most estimates treat as a launch cost. Two to five million is a build and a first year. The second year is not free.
The number most likely to be right is Stage 1, because a branded content pack inside an existing simulation is a well-understood transaction with a known shape and a lot of public precedent.
What each stage would cost per month, and what it buys.
| Stage | Monthly burn | For how long | The thing it buys |
|---|---|---|---|
| Stage 1 | £8,000–25,000 | 3–9 months | An answer to whether this audience plays farming games at all |
| Stage 2, build | £35,000–60,000 | 6–12 months | A working loop, a real economy, and two retention numbers |
| Stage 2, running | £12,000–20,000 | Indefinitely | Servers, live-ops, and the vote settling every season |
| Stage 3, build | £90,000–180,000 | 18–30 months | Reach, and nothing mechanical that Stage 2 did not have |
| Stage 3, running | £60,000–120,000 | Permanently | The half of a free-to-play game that everybody underestimates |
The plan, questioned.
Because it would spend between two and five million pounds to find out whether the loop works, when a quarter of a million answers the same question on the web with no store cut and no review cycle.
Then Stage 3 does not happen. The web game keeps running, it keeps costing twelve to twenty thousand a month, and nobody spends millions proving the same point twice.
Because the economy is the product and a client-authoritative economy is cheated within a week. It is also the only way to be able to say truthfully that the failure rate is identical for paying and non-paying players.
Nobody, about anything. This is an unsolicited third-party document. No studio, no publisher, no rights holder and no person named on this site has been contacted.
And the layer that joins it to everything else.
A narrow, versioned API that carries six things and refuses to carry four.
