Build or buy

What “build” actually includes here

Seven things, and the list is worth reading slowly because most build estimates in this category are estimates of the first two.

• A dual-currency ledgerTwo balances with hard separation at every write, not a column on a transaction.
• Wallet callback semanticsDebits and credits arriving mid-round, retried safely, and reconciled when a round does not finish.
• A free entry route that keeps workingUsable rather than documented, and running for as long as the product runs.
• Per-state access controlRegistration, play and redemption gated independently, per state, changeable without a release.
• Verification gating at redemptionA check somebody performs and a gate something enforces, and they are not the same component.
• Redemption records that survive an auditWhat was redeemed, under what conditions, verified by whom, retained long enough to answer a question you have not been asked yet.
• Content relationshipsStudios, certificates, currency contexts, and somebody whose job is the roadmap.

The full split of what exists on each route, row by row, is on the platform page.

The part that is not a project

Most of that list is not a project. It is an operation. A ledger, a free entry route and a state-access surface do not get shipped and finished — they run, and they keep needing answers when a state moves, when a studio leaves, when a payout fails and when somebody asks what happened in March. Teams costing a build almost always cost the building and not the running, and the running is the part with no end date.

• A state changes position and something has to be closed by the end of the week, including on a public holiday.
• A player disputes a redemption from four months ago and somebody has to produce the record.
• A title behaves differently in one currency context and somebody has to prove whether it should.
• An acquirer asks for a change to the product before they will lower the reserve.

What buying does not save you

Five costs stay yours whichever way this decision goes, and they are the ones that most often turn a comfortable budget into a tight one.

• Acquiring. Nothing on either route supplies the merchant account behind a coin purchase, and nobody can hand you an underwriting decision.
• The reserve. It is held against your business by somebody who is not us.
• Legal. The structure has to be right on your facts, in the states you open, and that is counsel's work whoever writes the software.
• Identity verification. Performed by you on both routes, with your provider.
• Marketing. Buying a platform does not buy an audience, and this is the cost most launch budgets underestimate.

How underwriting reads this model is on the acquiring page, and what a launch actually costs — including the four costs nobody quotes — is on the cost guide.

When building is right

Five conditions. They are written so that a reader who meets them can conclude build without feeling talked out of it, because for some teams that is the correct answer and a page pretending otherwise would be worth nothing.

• You already run a ledger you trust, and the sweeps layer is a small part of a larger product rather than the whole of it.
• You carry engineering capacity you are paying for anyway, and it has room.
• You have a strategic reason to own the stack that survives being said out loud to a finance director.
• Your requirements are genuinely unusual, and you can name which one no supplier meets. If you cannot name it, this is not the reason.
• You are prepared to own the operation as well as the build, including the parts that arrive at inconvenient times.

A middle option people miss

There is a third answer and it is not a compromise. An API integration is the hybrid people describe as if it did not exist: you keep the ledger, the accounts, the front end and the product, and buy the catalogue, the launch layer and virtual-currency support. It is ours, so read it as a fact about the shape of the market rather than as a recommendation — and it is why build-or-buy is a false binary in this category.

Which of the two products that is, and how it differs from the delivered version, is on the comparison of the three names. What the ledger has to do either way is on the dual-currency wallet.

How to decide

Five questions. If the room cannot answer the first two, the decision is not ready to be made.

• Who holds the wallet of record in the version you are imagining, and does everyone in the room agree?
• Which of the operating obligations has a named owner on your side already?
• What is the cost of the operation in year two, not the cost of the build in year one?
• Which requirement is the one no supplier meets, and can you name it?
• If you build it and it works, what have you got that a competitor cannot buy?