Every operator makes this decision at least twice, and most make it badly the second time — not because the analysis is hard, but because the answer that was right at one scale is quietly wrong at the next, and nobody schedules a review.
The three positions
Turnkey. A supplier provides platform, trading and content. Fast to market, low fixed cost, minimal control. Margin is capped by the revenue share and differentiation is capped by what the supplier offers everyone else.
Hybrid. Third-party platform, proprietary layers on top — usually pricing, risk and front end. Most mid-size operators live here, and for good reason.
Proprietary. Own the stack. Full control, full cost, full responsibility. Justified when scale makes the revenue share larger than the engineering organisation it would replace, and when differentiation genuinely requires it.
The principle that actually decides it
Own what produces your margin. Rent everything else.
For a sportsbook, margin comes from pricing and risk management. That is where an edge is possible and where the difference between good and average shows up in the number. Payments, account management, compliance workflow, bonus engines and content aggregation are commodity infrastructure — necessary, undifferentiated, and maintained better by a supplier at scale than by your team.
This principle is easy to state and consistently ignored, because engineering organisations find infrastructure satisfying to build and because "we own our whole stack" sounds stronger in a board meeting than "we own the part that matters".
The result is common enough to be a pattern: an operator two years into a full-stack rebuild, with a competent payments orchestration layer, a solid account management system, and pricing that is still a third-party feed with manual overrides on top.
When each answer is right
Turnkey works below roughly €20m in gross gaming revenue, or in a market you are testing rather than committing to. Below that scale, the revenue share is cheaper than the fixed cost of doing it yourself, and speed matters more than margin.
Hybrid works across a wide middle. The economics support a pricing and product team, but not a platform organisation. Most operators should be here longer than they think, and should be spending their engineering capacity on trading models rather than on infrastructure.
Proprietary works at real scale, in multiple markets, with a differentiation thesis that a supplier genuinely cannot serve. It is also the only answer that makes sense if the plan is to sell technology to others, which is a different business with different economics.
The decision has a shelf life
The most expensive version of this is not choosing wrong. It is choosing right and then not revisiting.
An operator who signed a turnkey deal at €10m GGR and is now at €80m is paying a revenue share that would fund a substantial engineering organisation. The contract renews quietly. The switching cost grows every year. Nobody wants to own the migration.
A reasonable discipline is a formal review at each doubling of revenue, at each material market entry, and at every contract renewal. Not a migration — a review, with the revenue share modelled against the equivalent internal cost, and the differentiation question asked honestly.
Migration cost is not an engineering number
When groups do decide to move, the estimate is almost always built by engineering, and almost always wrong in the same direction, because the dominant cost is not technical.
It is trading continuity. Moving a pricing and risk operation without a period of degraded margin is genuinely difficult. Models need retuning against new data structures. Traders need to relearn tooling under live conditions. Risk limits behave differently. The margin dip during a platform migration is frequently the largest line in the true cost, and it rarely appears in the business case at all.
Second is customer disruption — account migration, payment re-verification, and the churn that follows both. Third, and only third, is the engineering.
A credible migration plan prices all three, sequences trading last, and holds a rollback position for longer than feels necessary.
What to do with this
If you are below the middle of the range, resist building. If you are in the middle, check what your engineering team is actually spending its time on and whether it maps to where your margin comes from. If you are above it and still on a revenue share, model the alternative properly — not as a migration decision, just as a number.
And whichever position you hold, put a date in the calendar to look again.