Elite sport has moved from buying everything to trying to build everything. Neither end is where it will settle.
Executive Summary
For a long period, performance and medical departments relied on large, all-in-one systems that were expensive, slow to change and difficult to adapt to how an individual department worked. As AI and modern development tools have lowered the cost of building software, many departments have responded by building their own AMS platforms and rehab tooling in-house.
The motivation is legitimate. Departments want systems that reflect how they work and data they control. But building everything has become an over-correction. Development and data teams are spending significant time and budget rebuilding the foundational layers that every return-to-play process requires: the data model, permissions, audit trail and compliance, with practitioners advising along the way. These layers are difficult to build well, expensive to maintain, and rarely the source of a department's competitive advantage.
The mature position separates the stack into two parts. Departments own the layers specific to them, such as dashboards, reports, workflows and agents. They rely on established systems for the shared core. This paper sets out how the market arrived here, what the core layer involves technically, what rebuilding it costs, and how heads of department can decide which layers to own.
1. The Era of Fixed Platforms
The first extreme was defined by rigidity. Departments bought large systems designed around a general model of how a performance or medical department should operate. Configuration was limited, change requests moved through a vendor roadmap, and the cost of switching was high enough that most departments adapted to the system rather than the reverse.
The difficulty was the approach, not any one product. Every department runs return to play differently. Criteria, sign-off processes, the split of responsibility between physiotherapy, strength and conditioning and medical staff, and the way progress is reported to coaches and leadership all vary from one organisation to the next. A fixed system cannot reflect that variation, so it captures the parts of the process it was designed for and leaves the rest outside.
In practice, the rest moved into spreadsheets, shared documents and messaging threads. Rehab planning, phase criteria, handovers and stakeholder reporting, some of the most operationally important work in the department, ended up in files that were maintained manually, owned by individuals and disconnected from the system of record. The department paid for a platform and then ran a parallel, informal one alongside it.
2. The Case for Building
The second extreme is a direct response to the first. AI-assisted development, modern frameworks and accessible cloud infrastructure have reduced the time and skill required to build functional software. Departments that spent years adapting to their tools can now build tools that adapt to them.
The reasoning deserves credit. Departments are pursuing three legitimate goals:
• Fit. Workflows, criteria and reporting that reflect how the department actually operates.
• Ownership. Control over how data is structured, where it lives and how it is used.
• Connectivity. The ability to connect athlete data to analytics, models and AI agents on the department's own terms.
These goals are correct, and they will shape how performance departments operate over the next decade. For a meaningful share of the stack, building is the right way to achieve them. The question is whether building is the right way to achieve them for every layer.
3. The Over-Correction
When a department builds everything, it inherits responsibility for the full stack, including the foundational layers that every return-to-play process depends on. These layers share three characteristics. They are required regardless of how the department works. They are technically demanding to build correctly. And they carry ongoing obligations long after the initial build.
The Data Model
Return to play is a relational problem. An athlete has injuries. Each injury has a rehab case with phases, objectives and criteria. Each objective has measurements, achievement standards and dates. Each session links interventions to the objectives they serve, and each outcome links back to the athlete, the injury and the methodology used.
Modelling those relationships correctly is what makes the data useful later. A flat or loosely structured schema can support day-to-day logging, but it limits what can be asked of the data afterwards: how long a given criterion typically takes to achieve, which interventions preceded progression, how rate of progress compares across similar cases. The data model also has to evolve as criteria, test batteries and workflows change, without breaking historical records. Schema design and migration are specialist work, and errors made early compound over every season of data collected.
Permissions
A performance department is a multi-role environment. Physiotherapists, strength and conditioning coaches, doctors, sports scientists, department heads, coaches and external practitioners all need different levels of access to the same athlete record. Medical detail appropriate for clinicians is not appropriate for every stakeholder. External clinicians, national team staff and off-season providers may need access to one athlete's plan and nothing else.
Permissions therefore need to operate at the level of role, team, athlete and, in some cases, individual field. They need to be maintained as staff join, leave and change roles, and they need to fail safely. Access control is straightforward to prototype and difficult to get right under real organisational conditions.
Audit Trail
Return-to-play decisions carry clinical, competitive and financial consequences. When a status changes, a milestone date moves or a player is cleared, the department needs to know who made the change, when, and why.
A reliable audit trail records changes immutably, captures the reason alongside the change, and allows the state of a rehab case to be reconstructed at any point in time. It supports accountability across a multi-practitioner team, internal review after a setback, and any external scrutiny the department may face. Retrofitting an audit trail onto a system not designed for one is considerably harder than building it in from the start.
Compliance
Athlete health data is regulated. Depending on jurisdiction, departments operate under frameworks such as GDPR and HIPAA, alongside league, institutional and organisational policies. Compliance covers encryption in transit and at rest, data residency, retention and deletion policies, subject access requests, breach handling, vendor agreements and security review.
Compliance is also not a one-time task. Regulations change, security expectations rise and every new integration or AI connection introduces new data flows that have to be governed. For an in-house build, that responsibility sits permanently with the department.
The Common Thread
None of these four layers are specific to one department. Every team needs them, and none of them are what a department set out to improve when it decided to build. They are prerequisites. They also take the longest to get right, which means a significant share of an in-house build is spent reaching the starting line.
4. The Cost of Rebuilding the Core
The cost of building everything appears in four places.
Labour. Engineering and data staff are a scarce resource in most departments. Time spent on schema design, access control and compliance work is time not spent on analysis, modelling or tools that reflect the department's own methods.
Budget.
The initial build is the smaller part of the cost. Hosting, security review, maintenance, bug fixes and updates as requirements change continue for the life of the system.
Practitioner time. Clinicians and coaches are drawn into specifying requirements, testing incomplete versions and working around gaps while the system matures. That time comes from the same staff responsible for athletes currently in rehab.
Continuity.
An in-house system is often closely tied to the people who built it. When those individuals leave, knowledge of how the system works, and why it was built a certain way, can leave with them. Departments already experience this with spreadsheets and rehab plans owned by individual practitioners. A custom platform can repeat the same pattern at a larger scale.
Together, these costs delay the outcome the department wanted in the first place: clearer rehab plans, shared visibility across disciplines and better-informed return-to-play decisions.
5. A Layered Model for Return-to-Play Systems
The mature position treats the stack as two distinct layers with different ownership.
The core layer contains what every return-to-play process requires: the relational data model, permissions, audit trail, compliance and the operational structure for planning and tracking rehab. This layer benefits from being built once, built to a high standard and maintained by a dedicated team. Departments should expect it to be stable, secure and fully accessible to them through open interfaces such as APIs and AI connectivity.
The owned layer contains what makes a department distinct: its dashboards, reports, analytical models, workflows and AI agents. This is where a department's methodology, priorities and competitive thinking live. It is also where in-house build effort produces the greatest return, because the work reflects knowledge no external system can supply.
The condition that makes this model work is data ownership. A department relying on an established core must still own its data, on its own terms, with the ability to pull it into any environment it builds. Without that condition, the model drifts back towards the rigidity of the first extreme.
6. Deciding What to Own
For heads of department, the decision is about where build effort produces a return. Four questions help separate the layers:
• Is this layer specific to how we work? If another department would need the same thing built the same way, it belongs in the core.
• Does building it create an advantage? Layers that encode the department's methodology or analytical approach are worth owning. Layers that meet a universal requirement are not.
• Who maintains it in three years? Any layer the department builds becomes a permanent responsibility, including through staff changes.
• What is the cost of getting it wrong? Layers with clinical, legal or regulatory consequences carry the highest risk when built under time pressure by a small team.
Applied consistently, these questions direct build capacity towards the layers that reflect the department's expertise, and away from rebuilding foundations that already exist.
Conclusion
The swing from fixed platforms to building everything was a reasonable reaction to systems that could not adapt. The next stage is a more deliberate allocation of effort: an established core for what every return-to-play process needs, and department-owned layers for what makes each department different. Departments that make that distinction will spend their time and budget on the work that improves return to play, and reach those improvements sooner.
Gameplan is built as the support layer for teams improving their return-to-play systems.
Head to www.GameplanPerformance.com to schedule a discovery call.
- Build Vs Buy
- Return-To-Play Systems
- Athlete Management Systems
- AI-Assisted Development
- Relational Data Modeling
- Role-Based Access Control
- Clinical Audit Trails
- Athlete Data Compliance
- Data Ownership
- Layered Software Architecture















