Agile Practice Guide

A collaborative effort between the Project Management Institute (PMI) and Agile Alliance, this guide provides practical guidance for project leaders and teams seeking to adopt agile approaches. It positions itself as a bridge between traditional, plan-heavy project management and agile methods, arguing that rapid technological change, customer demand for immediate value, and competitive pressure from faster-moving organizations require a shift toward agile thinking. The guide limits its scope to project- and team-level implementation, covering popular agile approaches and mapping them to PMI's established project management framework. It does not prescribe step-by-step instructions or endorse any single methodology.

The guide opens by distinguishing two types of project work. Definable work involves clear procedures with low uncertainty, such as manufacturing a car after the design is complete. High-uncertainty work involves exploration and novel problem solving, as encountered by software engineers, product designers, doctors, and lawyers. As automation handles more definable work, teams increasingly face high-uncertainty projects marked by rapid change, complexity, and risk. Traditional predictive approaches, which attempt to lock down requirements upfront, struggle under these conditions. Agile approaches were created to explore feasibility in short cycles and adapt quickly based on feedback.

The guide traces agile's formal origins to the 2001 Manifesto for Agile Software Development. The Manifesto establishes four core values: individuals and interactions over processes and tools; working software over comprehensive documentation; customer collaboration over contract negotiation; and responding to change over following a plan. Twelve supporting principles emphasize early and continuous delivery, welcoming changing requirements, self-organizing teams, sustainable pace, technical excellence, and regular reflection. The guide clarifies that "agile" is a blanket term for any approach fulfilling these values, including Scrum, Kanban, eXtreme Programming (XP), Crystal, and others. It also positions agile and the Kanban Method, a workflow management approach inspired by lean manufacturing, as descendants of lean thinking, a philosophy centered on delivering value, minimizing waste, and continuously improving.

A central argument connects uncertainty to the selection of a project life cycle, the overall approach governing how a project progresses from start to finish. Using an adapted version of the Stacey Complexity Model, the guide argues that as uncertainty in requirements and technology increases, so does the risk of rework. Delivering work in small increments lets teams verify results, understand customer requirements faster, and adapt through short feedback loops and frequent replanning.

The guide defines four life cycle types along a continuum. Predictive life cycles are sequential approaches where bulk planning occurs upfront and execution happens in a single pass; they succeed when requirements are firm but do not deliver business value until the end. Iterative life cycles improve products through successive prototypes and stakeholder feedback, optimizing for learning. Incremental life cycles deliver finished subsets of the product at regular intervals, optimizing for speed. Agile life cycles combine iterative and incremental characteristics. Within agile, the guide distinguishes iteration-based approaches, where teams work in fixed-length timeboxes (predetermined periods of equal duration), from flow-based approaches, where teams pull features from a backlog (a prioritized list of pending work items) based on capacity and manage work-in-progress limits (caps on how many items may be active simultaneously).

The guide introduces hybrid life cycles as combinations tailored to project needs. Examples include a pharmaceutical company that used agile drug trials but retained a sequential U.S. Food and Drug Administration approval phase, and a government credit insurance project where most development was agile but premium rate calculations were coded predictively. Hybrid approaches also serve as transition strategies for organizations moving gradually toward agile. Teams rarely limit themselves to one framework, often blending Scrum's structure with Kanban's visual workflow management and XP's engineering practices.

Turning to implementation, the guide argues that adopting agile requires servant leadership, the practice of leading through service to the team. Servant leaders establish purpose first, then focus on people, and finally on process. Key responsibilities include facilitating collaboration, removing organizational impediments, and supporting team self-management by working with management, product owners (who guide product direction and prioritize work by business value), and other stakeholders. The guide acknowledges that many agile frameworks do not address the project manager role but argues that project managers add significant value by shifting from coordinating to coaching.

Effective agile teams, the guide argues, range from three to nine members, are ideally colocated and fully dedicated, and are cross-functional, meaning they collectively possess all skills needed to deliver working product. It describes three common roles: cross-functional team members, the product owner, and the team facilitator or servant leader. The guide advocates for "T-shaped" team members who combine deep expertise in one area with broad competence across other skills to reduce bottlenecks. It warns against multitasking, citing productivity losses of 20 to 40 percent from task switching.

For delivery practices, the guide identifies retrospectives as the single most important agile practice: Teams examine qualitative and quantitative data to find root causes of problems and develop action plans. Other core practices include backlog preparation, where the product owner maintains an ordered list of work in story form (short, user-centered descriptions of desired functionality); daily standups, timeboxed at 15 minutes, where team members surface impediments; and demonstrations of working product at least every two weeks. Technical practices include continuous integration, testing at all levels, test-driven development, and spikes (timeboxed research experiments).

The guide critiques traditional measurement approaches such as percent-done estimates, introducing the "watermelon project" concept: a project that appears on track externally but is actually in trouble internally. Agile's regular delivery of finished work creates empirical data enabling better forecasting. Specific tools include burndown and burnup charts tracking work over time, velocity (the total story points, or relative estimates of work size, completed in an iteration) for capacity planning, and flow-based metrics such as cycle time and lead time that measure how quickly items move from request to completion.

On organizational considerations, the guide argues that culture, structure, and policies shape project outcomes, and that creating a psychologically safe environment is the most important cultural prerequisite for adopting new methods. It presents a framework for assessing organizational priorities along scales between competing values such as exploration versus execution and flexibility versus predictability. For procurement, it advocates shared-risk-reward contracts over adversarial arrangements, describing techniques such as multi-tiered structures, fixed-price microdeliverables, and team augmentation embedding supplier staff within the customer organization.

Addressing scaling, the guide notes that frameworks such as the Scaled Agile Framework (SAFe), Large Scale Scrum (LeSS), and Disciplined Agile have emerged for multi-team programs, but it cautions against scaling if single-team implementation is not yet working. It argues that an agile project management office (PMO) should be value-driven, invitation-oriented, and multidisciplinary. The guide recommends treating organizational evolution toward agility as an agile project itself, creating a backlog of changes prioritized by value, testing each as a time-limited experiment, and tracking progress on kanban boards (visual boards displaying work stages and item flow).

The guide concludes by affirming that agile adoption has grown dramatically since 2001 and applies across industries. It invites reader feedback, framing the guide itself as subject to iteration and improvement. Annexes map agile practices to the PMBOK® Guide (PMI's foundational project management standard), connect the Manifesto to the guide's sections, and provide overviews of major frameworks. Appendices offer tailoring guidance, cautioning that modifying approaches requires mastery of fundamentals first, and present an agile suitability filter tool that plots project attributes on a radar chart to indicate whether agile, predictive, or hybrid approaches are most appropriate.