AWS Cloud Operations Blog

Delivering Business Value with Cloud Platform Teams

Imagine a world where software developers build applications quickly on the cloud, focusing on innovative features, unburdened by complex infrastructure and intricate configurations. Welcome to the era of the cloud platform team.

Cloud platform teams build internal tools, automation, and self-service infrastructure to free developers from commodity tasks and enable them to innovate faster. However, the undeniable ROI of platform teams remains opaque as their impact is often indirect.

This blog post dives deep into platform teams to understand how they can deliver business value, the challenges they face, and the best practices for being successful.

Understanding platform teams, their benefits, and pitfalls

Platform teams

Let’s first understand how platform teams are different from other technical teams. There are application development teams that build end-user products and internal apps for business units. In this blog, they are referenced as “stream-aligned teams” (a term borrowed from Team Topologies). Platform teams build foundational platforms for these other teams to build upon. In other words, stream-aligned teams are customers of the platform team. Platform teams are centralized and serve multiple stream-aligned teams across business units. The cloud platform team is also known as the cloud center of excellence, the cloud engineering team, or the cloud enablement team. Regardless of name, the charter is similar. Platform teams build self-service-enabled internal platforms. Think app stores for developers, offering pre-configured infrastructure, tools, and services. Developers can easily deploy and scale environments or services without manual setup.

Benefits of platform teams

Platform teams alleviate the cognitive load on stream-aligned teams. Building modern applications requires managing a complex ecosystem of infrastructure, tools, deployment pipelines, testing, networking, monitoring, and security. Having stream-aligned teams manage this detracts from their core work and introduces risks. Moreover, multiple teams need the same tools, automation, and infrastructure. Instead of creating silos, the platform team addresses this by establishing standardized cloud infrastructure, automating tasks, and providing centralized tools. This ensures security, promotes reuse, and optimizes resources to adopt cloud services at scale. Furthermore, platform teams streamline operations and maintenance with automation while optimizing cloud service costs.

Specialized platform teams

As cloud adoption scales, monolithic platform teams struggle to manage diverse needs. Enter specialized platform teams. A data platform team simplifies data complexity with data lakes, governance, and access controls, empowering faster data-driven decisions. A security platform team develops security tools and automation to improve vulnerability management and compliance. Machine learning platform teams accelerate AI/ML innovation by offering pre-built tools and automated frameworks, streamlining the journey from idea to implementation.

Pitfalls

While platforms provide value, implementing them well can be challenging. Platform teams risk undermining the autonomy of stream-aligned teams if they introduce bureaucracy and excessive controls that create friction when consuming their services. Ignoring the developer experience when building platforms is a leading cause of failure. As platform teams serve internal organizations, stream-aligned teams often complain that platform teams lack accountability and treat them as end-users, not customers. Platform value, derived from saved developer time, risk mitigation, and accelerated innovation, is indirect, unlike stream teams. This often results in platforms not getting adequate staffing, limiting their effectiveness.

Best practices for delivering business value with platform teams

Work backwards from the needs of the stream-aligned teams

Working backwards from stream-aligned teams’ needs and pain points ensures that platform features solve real problems, boosting adoption and team value. Consider direct feedback loops, user research, and early prototypes to earn trust. For example, platform teams can offer pre-built and customizable continuous integration and deployment (CI/CD) pipelines for automated testing, deployment, and delivery, speeding up development cycles. Mature platform teams foster communities of practice where teams collaborate and share knowledge. Through this close collaboration, a continuous feedback loop is established. This loop informs platform feature prioritization, enabling iterative delivery for ongoing improvement and refinement. Stream-aligned teams should include platform teams in their strategic priorities and key initiative planning discussions.

Centralized/Decentralized tradeoff – What will the platform team own?

Deciding what features are built centrally on the platform and what stream-aligned teams build themselves is important. Decentralize by default to empower stream-aligned teams. Centralize capabilities that speed up stream-aligned teams and where governance is critical. Ensure centralized solutions are automated and lightweight to avoid hindering agility. For example, platform teams can use Infrastructure as Code (IaC) to automate provisioning, allowing developers to self-provision standardized infrastructure stacks instead of waiting.

Leadership support and organizational placement

As platform teams are the unseen foundation that empowers developers to build efficiently, they need leadership support to sponsor their work and justify staffing them. Platform teams can start small and scale as needed. These teams typically sit within the broader centralized engineering or IT department (often in CTO/CIO organizations), bridging development and operations. As an organization grows larger, platform teams can be federated into business units but still operate centrally within them. As complexity increases and demands diversify, specialized platform teams such as those for data, security, and machine learning can be formed.

Ease of adoption

Platform teams should focus on ease of adoption by allowing self-service onboarding and providing detailed documentation that includes quick-start guides, code samples, and best practices for usage. Changes to platforms, such as widely used platform APIs like authentication, should not force stream teams to modify code and redo testing. A backward-compatible API version is a good strategy to allow seamless updates. Major platform upgrades should be automated without needing extensive effort from the stream teams to implement. Platform teams are either funded at the top or through a usage-based chargeback model. If the latter, making the chargeback process simple and transparent is important to build trust.

Product operating model, experimentation, and incremental delivery

Platform teams fail when they deploy technology capabilities instead of aligning with business outcomes. A shift to a product-based operating model instead of the traditional “activity-based” model is crucial. Examples of products are a container-orchestration platform for supporting microservices-based teams or a real-time dashboard for application performance and health. This customer-centric approach starts with the needs of internal teams, fostering iteration and experimentation. Instead of rushing to “make it as quickly and cheaply as possible,” the focus is on learning, adaptation, and minimizing waste. This deliberate model drives greater innovation, smarter decisions, improved products, and ultimately, happier, more empowered teams. Platform teams can operate as two-pizza teams with single-threaded ownership over specific products, ensuring agility and accountability.

Balancing between standardized and bespoke platform solutions

Platform teams often struggle to decide between standardizing their offerings and creating bespoke solutions. Standardization ensures efficiency, while bespoke solutions address unique team needs. Strategies include building a robust foundational layer of standardized infrastructure and tools, with optional customizable “layers” on top. Provide configurable options for teams to tailor to their needs. Set clear standards and governance principles to guide standardization and customization. Remember, it’s not a binary choice between one or the other. It’s finding the right mix of pre-built and customizable options that empowers developers and drives optimal results. Allow developers to experiment with emerging technologies in a controlled, low-risk manner, ensuring continuous innovation.

Platform accountability

The business impact of defects in a platform is wider as they impact multiple teams. Platform services should be supported with the same (if not more) rigor and attention as external customer-facing applications. Services offered by platform teams should have documented SLAs that are monitored and reported on for their availability and performance. Platform teams should be accountable for their SLAs and be prepared to triage production incidents alongside stream-aligned teams.

Measuring value

While cost savings are often used to justify platform teams, cost alone should not be the focus. Advocate for the platform as a catalyst for innovation, not a cost center. The value of a platform should be measured by developer velocity gains, platform adoption rates, and direct business impacts. Examples of measures of developer velocity include the number of deployments in a period, the number of automated processes, commit to production time, and incident resolution time. Platform adoption measures include the number of teams using the platform, active initiatives on the platform, and usage metrics of its tools. Business impact measures include time to market, cost savings, and improvements in the quality and reliability of products. Choose metrics that capture a holistic impact.

Closing thoughts

By offering standardized tools, infrastructure, and services, platform teams streamline development processes. Stream-aligned teams can focus on coding and innovation without the burden of managing intricate details or navigating diverse technologies. This reduction in cognitive load enhances productivity, accelerates development cycles, and empowers teams to deliver high-quality solutions efficiently. In essence, platform teams can transform raw resources into customized platforms that cater to specific needs and accelerate workflows across the organization.

Additional Reading

Team Topologies – Learn

Martin Fowler on Team Topologies

Business Capability Centric

Shifting to a product-based model and positioning people for success

How to Set Up a Platform That Effectively Supports Your Development Teams

About the authors:

Nurani Parasuraman is part of the Customer Solutions team in AWS. He is passionate about helping enterprises succeed and realize significant benefits from cloud adoption by driving basic migration to large-scale cloud transformation across people, processes, and technology. Prior to joining AWS, he held multiple senior leadership positions and led technology delivery and transformation in financial services, retail, telecommunications, media, and manufacturing. He has an MBA in Finance and a BS in Mechanical Engineering.

Bryan Callahan is a Senior Customer Practice Manager with the AWS Professional Services team. He is focused on helping enterprise manufacturing customers transform legacy tools, technology, and processes into modern, data-driven solutions to increase efficiency and value to end customers. Prior to joining AWS, he led numerous modernization initiatives and delivery teams in the public sector and lead commercial development efforts in healthcare, media, and financial services. He has a BS in Economics

Ryan Lindeman leads a team of AWS Solution Architects that accelerate OT/IT innovation in Automotive and Manufacturing. Before joining AWS, Ryan built AI/ML security products at Microsoft and more recently ran a multi-cloud managed service for 100s of global customers at Deloitte. He is an AWS certified Solution Architect with a BS in Information Systems and Graduate Degree in Business Analytics