As Intel moves to becoming a platform company, more emphasis has been put on understanding end-users' needs and integrating user requirements into the process of platform design. As described in [1], Intel has been using a usage model-driven process to define platform capabilities and ingredients. There are four major milestones in this process: 1) researching user needs, 2) usage models and use case development, 3) platform capabilities and ingredient mapping, and 4) platform verification and refinement through Proof of Concept (PoC) projects.

User needs. We typically use ethnographic or qualitative research to understand top user challenges, barriers, and needs for technology adoption. These research methods often involve in-depth interviews with end users in business settings or at their homes, direct observation of user activities and environments, or job shadowing and participant observation. In a business setting, we try to understand three tiers of needs to identify the best technology solutions, which lie at the intersection of business needs, IT needs, and end-user needs. Other sources of information on user needs include feedback from sales teams in the field, direct engagement with customers, ecosystem players (key software and hardware vendors, service providers, etc.) and independent market analysts (Gartner*, IDC*, etc.).

Usage models. A usage model is a representation of desired user experiences and user interactions with a computing system. Key components of a usage model include user personas, use conditions, storyboards, use cases, usage scenarios, and user task flows [2]. The creation of usage models is a structured process that requires expertise from different disciplines, including user research or human factors engineering, platform architecture design and business development, or product marketing.

Platform and ingredient capabilities. For high-priority use cases and usage scenarios, we focus on two areas of analysis: platform capabilities and platform ingredient requirements. For platform capabilities, we perform gap analysis on new platform capabilities and ingredients that are needed to enable the usage models or user experience; for platform ingredient requirements, we conduct workload analysis or predictions to identity performance bottlenecks and desired architecture. In order to prioritize different platform features, we typically conduct quantitative market research to understand market potential for different features.

Platform verification and refinement—Proof of Concept (PoC). After the platform features are defined and a prototype is developed, we engage selected customers or end users in different regions and industries to conduct PoC evaluations of targeted platforms, which consist of engagement with real end users, pilot solution deployment with the new platform features, and analysis of the pilot result against expected business outcome. The same use cases and usage scenarios used for platform definition are used to guide PoC projects. Through this type of deep customer verification processes, we gather further feedback for our platform designs and further refine our platforms before offering them to the general market. In this way, we develop end-to-end integration stories with end user buy-in before a platform is launched, so that the platform is better suited for adoption by both end users and the entire supply chain.

We use three cases studies in this paper to illustrate the usage model-driven platform design process. We discuss how usage models are defined and used to create new platform capabilities or to optimize platform performance.