Structure follows strategy, and constrains it
Organisational design is treated as an administrative matter more often than it should be. In practice, structure determines what an operator can execute, because it determines who owns what and how quickly decisions get made.
An operator pursuing a market-specific strategy through a purely functional organisation will struggle, because nobody owns a market. An operator pursuing scale economics through a fully devolved market structure will duplicate work and lose the advantage that made multi-market operation viable. Neither problem is solved by effort.
The two basic shapes
Functional structure organises by discipline. Marketing, product, trading, compliance, payments, customer operations and technology each exist once and serve every market.
Its advantages are real. Capability is not duplicated. Standards are consistent. Specialists work alongside other specialists, which develops depth. Scale economics are captured, which matters enormously given the fixed cost dynamics covered earlier in this course.
Its weakness is that no one owns a market outcome. When a market underperforms, the cause sits across several functions and the accountability is diffuse. Market-specific requirements compete for attention against every other market's requirements, and the smaller markets lose consistently.
Market structure organises by geography. Each market holds the capabilities it needs and owns its results.
Its advantages are ownership, responsiveness and local knowledge. Someone is accountable. Decisions can be made quickly by people who understand the market.
Its weakness is duplication and inconsistency. The same capability is built several times, standards drift, and the scale economics that justify multi-market operation are eroded. In a sector where fixed cost per market is already the dominant constraint, adding avoidable duplication is expensive.
Where operators actually land
Most operators of any size end up with a hybrid: functional ownership of capability, with market-facing roles owning outcomes, which is a matrix whether or not it is called one.
Matrices are widely disliked and are frequently the correct answer here, because the underlying tension is genuine. The industry requires both scale economics and market-specific execution, and no structure delivers both without some form of dual accountability.
What distinguishes matrices that work from those that do not is not the diagram. It is whether decision rights are explicit.
In a functioning matrix, it is written down who decides what. Which decisions belong to the market owner, which to the functional lead, which require both, and what happens when they disagree. People know before the conversation who is deciding.
In a dysfunctional matrix, decision rights are negotiated case by case. Every significant choice becomes a question of whose view carries more weight, resolved by escalation or by persistence. This is where the reputation of matrix organisations comes from, and it is a failure of specification rather than of the structure itself.
What to centralise
A reasonably consistent set of principles applies.
Centralise what benefits from scale. Platform and technology, data infrastructure, group-level supplier relationships and specialist capability that no single market could justify.
Centralise what requires consistency. Group compliance standards, brand, risk appetite and financial control. Inconsistency here is a liability rather than a local adaptation.
Centralise what carries group-level risk. Anything where a failure in one market damages the group, which in this sector includes regulatory conduct, since licensing authorities assess applicants globally.
Devolve what requires local knowledge. Market-specific compliance interpretation, local partnerships, local marketing execution, and product decisions that depend on local customer behaviour.
Devolve what requires speed. Decisions that must be made faster than a central function can serve, provided the decision-maker is competent and the boundaries are clear.
The recurring error is centralising for control rather than for benefit. A central function that must approve routine local decisions creates a bottleneck without improving quality, and the usual consequence is that markets work around it, which produces the inconsistency centralisation was meant to prevent.
The compliance case
Compliance deserves separate treatment because both principles apply simultaneously and neither can be sacrificed.
Local knowledge is non-negotiable. Obligations differ by market in detail that matters: verification timing, permitted products, advertising rules, reporting formats, responsible gambling tooling, complaint escalation routes. A central function without local expertise will apply the wrong requirements confidently.
Central consistency is also non-negotiable. Group standards, a consolidated view of regulatory risk, and the recognition that conduct in one market affects licences in others all require a group perspective. Markets left entirely to themselves produce divergent standards and no aggregate visibility.
The workable arrangement is generally a central function owning standards, aggregate risk view, group policy and the relationship with the board, with market compliance roles owning local interpretation and application, reporting into the central function rather than into market commercial leadership.
That last point matters. A compliance officer reporting to the person whose revenue targets compliance decisions may constrain has an incentive problem structurally identical to the VIP account manager case covered in the Customer Service course. Independence of reporting line is what makes the function able to do its job.
Symptoms of an outgrown structure
Operators generally outgrow their organisation before anyone proposes changing it. The symptoms are recognisable.
Decisions take longer than the value of what is being decided. Routine choices requiring several meetings and an escalation indicate coordination cost disproportionate to the work.
Nobody can say who owns an outcome. Asked who is accountable for a market's performance or a product's success, the answer is a list.
The same work happens in several places. Duplication that nobody planned, usually because markets or teams built capability rather than waiting for a central function.
Escalation is the normal path. Where most significant decisions reach senior leadership, authority has not been delegated adequately, and senior capacity is consumed by choices others should be making.
Handoffs dominate. Work spends more time waiting between teams than being done.
Local workarounds proliferate. Markets bypassing central processes is evidence that the process does not serve them, regardless of how well it is designed.
None of these is fixed by exhortation. They are structural symptoms requiring structural responses.
Decision rights as the practical tool
The most useful intervention available in most organisations is not restructuring but clarifying who decides.
A simple approach names, for each significant decision type, who decides, who must be consulted before the decision, and who is informed after it. The value is not the framework but the requirement to be specific, because ambiguity about decision rights is what generates most coordination cost.
Two disciplines make this stick. Decisions should be pushed down as far as competence allows, since authority sitting above the information produces slow, poorly informed choices. And disagreement should have a route, so that an unresolved conflict escalates promptly rather than persisting as friction.
This is considerably cheaper than reorganising and addresses a large proportion of the problems that prompt reorganisation.
Scaling stages
Operators pass through recognisable stages, and the structure that works at one is wrong at the next.
At small scale, structure barely matters. Everyone knows everyone, decisions are made in conversation, and formal process adds cost without benefit. The failure at this stage is premature bureaucracy.
At growing scale, informal coordination stops working. People no longer know what others are doing, decisions get made twice or not at all, and consistency drifts. The failure is denial, continuing to operate informally past the point it functions.
At multi-market scale, the functional against market tension becomes acute, specialist capability is required, and compliance becomes a substantial function. The failure is structure lagging strategy, with the organisation still shaped around the markets it had three years ago.
At large scale, coordination cost becomes the dominant constraint. The failure is accumulating process, where every past problem has generated a control and the aggregate makes the organisation slow.
The general observation is that structure needs revisiting periodically rather than when it breaks, for the same reason market portfolios do. Reviews triggered by dysfunction arrive after the cost has been paid.
Governance and the review rhythm
Structure determines who decides. Governance determines when and on what basis, and it is the part most often left implicit.
A functioning operator has a small number of recurring forums with clear purposes. An operational review examining performance against plan and dealing with immediate issues. A risk and compliance forum covering regulatory matters, incidents and emerging obligations, with genuine independence from commercial pressure. An investment or prioritisation forum allocating scarce engineering and capital across competing demands. And a strategic review on a longer cycle covering the structural decisions described in the first lesson of this course.
The failure modes are recognisable. Forums that review without deciding, producing discussion and actions rather than conclusions. Forums whose agenda is reporting rather than decision, consuming senior time on information that could have been circulated. Forums with no clear owner for the decisions they take. And the proliferation of forums generally, where every past coordination failure has generated a standing meeting and the aggregate consumes the capacity it was meant to protect.
The specific requirement in this sector is that compliance and risk governance must be genuinely independent. A risk forum chaired by the person whose commercial targets the risks constrain is not providing challenge. This is the same structural principle that applies to VIP account management and to compliance reporting lines, and it recurs because the underlying problem recurs: where the person identifying a problem is the person whose results it affects, the problem tends not to be identified.
Talent and the capability question
A dimension of organisational design that sits alongside structure.
This sector has specific capability requirements that are not abundantly available: trading and quantitative pricing, gambling compliance, high-risk payments, game mathematics, and safer gambling practice. These skills are developed largely within the industry, which means the pool is limited and mobile.
The practical consequences for organisational design are that key person dependency is a real risk, particularly in trading and compliance where individual expertise can be substantial and undocumented; that geographic clustering of talent shapes where operators locate functions, since certain cities hold concentrations of relevant experience; that development matters because hiring alone cannot fill these roles at the rate the sector needs; and that retention in specialist functions is worth more attention than headline turnover figures suggest, since replacing a competent compliance officer or trader takes far longer than the average role.
Structures that concentrate specialist capability centrally generally develop it better than those that scatter individuals across markets, because specialists learn from other specialists. This is a genuine argument for centralisation in the specialist functions, separate from the scale economics argument, and it is frequently the stronger one.
Reorganising, and when not to
A final practical caution, because restructuring is a common response to problems it does not solve.
Reorganisation is disruptive. It consumes months of attention, unsettles staff, breaks working relationships that were functioning, and produces a period of reduced output while people work out their new responsibilities. It is sometimes necessary and is frequently reached for too readily.
The questions worth asking before restructuring are whether the problem is genuinely structural, whether clarifying decision rights would resolve it more cheaply, whether the issue is capability rather than arrangement, and whether the proposed structure solves the identified problem or simply differs from the current one.
Many problems attributed to structure are actually decision rights problems, which can be fixed by specification. Others are capability problems, where the right structure exists and the people in key roles are not equipped for them, and reorganising around that avoids the harder conversation. Others are strategy problems, where the organisation is unclear about priorities and no arrangement of boxes resolves that.
Where restructuring is genuinely warranted, the things that make it work are being explicit about the problem it addresses, moving quickly rather than allowing prolonged uncertainty, being clear about decision rights in the new arrangement rather than assuming they follow from the diagram, and accepting that the transition period will be less productive rather than pretending otherwise.
The change management dimension of this is substantial enough to warrant its own treatment, which the final lesson of this course provides.
Structures for specific functions
A brief note on where particular functions tend to sit, since the general principles resolve differently for each.
Trading is almost always centralised, because pricing capability benefits enormously from concentration and because liability must be managed at group level rather than market by market.
Compliance requires the hybrid described above: central standards and risk view, local interpretation and application, with reporting lines running to the centre.
Payments is usually centralised for provider relationships and orchestration, with market-specific work on local methods. The compliance interface argues for close central coordination.
Marketing is typically split, with brand and central capability held centrally and acquisition execution devolved, since channel effectiveness and creative resonance are strongly local.
Product varies most. Core platform product is centralised by necessity. Market-specific product work has to sit somewhere, and where it competes for central attention against every other market it consistently loses, which is an argument for some local product ownership.
Customer operations is generally structured by language and time zone rather than by market, which mostly works and creates the multi-market compliance complication described in the Customer Service course, where an agent must apply different rules depending on the customer's jurisdiction.
Data and analytics benefits from centralisation for infrastructure and standards, with embedded analysts serving functions and markets. Fully centralised analytics tends to become a request queue disconnected from the decisions it should inform.
None of these is prescriptive. The point is that the functional-against-market question has a different answer for each function, and applying a single organisational philosophy uniformly across all of them produces a structure that suits some and constrains others.
The remote and distributed question
One further consideration that has become material and did not exist in the same form a decade ago.
This sector has always been geographically distributed, with operators frequently licensed in one jurisdiction, headquartered in another, and serving markets across several more. Distributed working has extended that further, and the organisational implications are worth stating.
The advantages are access to specialist talent that is scarce and geographically clustered, and the ability to place market-facing roles in the markets they serve rather than at a central office.
The costs are real and specific to certain functions. Trading, particularly in-play, benefits from co-location because rapid coordination during live events is difficult remotely. Incident response across payments, platform and compliance benefits similarly. Development of specialists, where junior staff learn by proximity to experienced ones, is measurably harder when distributed. And cross-functional problem solving, which in this industry means payments talking to product talking to compliance, happens less spontaneously when nobody shares a space.
The workable arrangements generally concentrate functions where co-location genuinely matters, distribute those where it does not, and invest deliberately in the coordination that used to happen by accident. What does not work is assuming that a structure designed for co-located teams will function identically when distributed, which is how operators end up with the coordination cost of a matrix and none of the informal resolution that made it bearable.