CMMI 評価のための SCAMPI 方法論の詳細

• Ensure meticulous evidence collection, organization, and traceability for every PI. Conduct thorough readiness checks and pre-appraisal activities.
• Use an experienced SEI-Authorized Lead Appraiser with relevant domain experience.
• Prepare participants for formal interviews. Ensure strong, visible sponsor commitment and resource allocation.

• Focus on honesty and identifying genuine improvement areas, not just “passing.” Use it as a learning and team-building opportunity.
• Prioritize findings based on business impact and feasibility. Involve a good mix of practitioners and managers to get diverse perspectives.
• Can be used to “pilot” process changes.

• Keep the scope extremely narrow and well-defined. Focus on actionable feedback and quick wins.
• Use internal experts familiar with the specific process. Quick turnaround from initiation to feedback is paramount.
• Good for verifying specific process changes or addressing a particular concern.

• Technicities/Controls/Quality/Process Points: characterized by ad-hoc, inconsistent processes. Success depends on individual heroics rather than institutionalized processes. Few stable processes exist. Inspection focuses on verifying that work *is* being performed and results produced, even if chaotically, as a baseline. No specific “ML1 processes” are “inspected for compliance” beyond confirming activity. Failure to demonstrate ML2 PAs results in ML1.
• Contracts/Legals: little to no systematic approach; handled reactively.

• Class B: would observe and document the ad-hoc nature of processes, lack of defined project management, inconsistencies, and reliance on individuals. Highlights these as fundamental gaps and risks, providing a baseline for recommending basic project management disciplines (ML2).
• Class C: might investigate a single, highly problematic, and uncontrolled process that is symptomatic of ML1 (e.g., “why are critical defects repeatedly missed in product X testing?”) to understand its specific failures and recommend immediate, localized containment or rudimentary controls.

• Technicities: documented project plans (including estimates, resources, schedules, risks), managed requirements, version-controlled work products, defined measurement repository, documented quality assurance procedures, supplier agreements.
• Controls: monitoring project performance against plans, tracking progress, corrective actions based on deviations, configuration audits, PPQA audits ensuring compliance with project-defined processes, formal reviews with management and stakeholders.
• Quality: adherence to project-specific defined processes and standards, objective 検証 of process compliance (PPQA). Work products are controlled.
• Process Points: established policies for managing projects, documented project-specific processes, provision of resources (funding, people, tools), assigned responsibilities, training for project personnel on their roles and processes. Stakeholder involvement is planned and managed.
• Contracts/Legals (SAM PA): formal agreements with suppliers, monitoring supplier performance against those agreements, managing changes to subcontracts.

Evidence of project-level process discipline is mandatory for an ML2 rating.

• Technicities: examines existence and practical use of project plans, requirements management approaches, version control, measurement activities, QA activities, and supplier management.
• Controls: how projects generally monitor progress, manage changes, involve stakeholders, and ensure quality. Focus on identifying inconsistencies or areas lacking control.
• Quality: general adherence to planned project processes, effectiveness of QA reviews, identification of common quality issues at the project level.
• Process Points: understanding if projects are planned and managed according to some documented processes. Are basic project management disciplines evident? Identifies gaps in policies, resources, or training that hinder project management.
• Contracts/Legals (SAM PA): reviews how supplier relationships are managed, if agreements exist, and if performance is tracked.

Aims to provide a gap analysis against ML2 PAs for internal improvement or Class A readiness.

• Technicities/Controls: adherence to a specific documented procedure, use of a specific tool (e.g., CM tool), execution of a specific review (e.g., PPQA audit for one artifact).
• Quality: effectiveness of a specific control or quality check within the narrow scope (e.g., are defects found by PPQA reviews being addressed?).
• Process Points: is there a documented process for the specific item under review for that project? Are people aware of it?

Provides quick, targeted feedback on a limited aspect of ML2.

• Technicities: documented organizational standard processes, tailoring guidelines, Process Asset Library (PAL) containing templates, checklists, lessons learned, process descriptions. Documented engineering processes (RD, TS, PI, VER, VAL). Risk management plans and status. Decision analysis records for key decisions.
• Controls: organizational oversight of process definition and improvement. Collection and analysis of process-related experiences and measurements to improve organizational assets. Projects use their defined process (tailored from organizational standards) and integrated project management practices. Configuration management of organizational process assets.
• Quality: consistency in process execution across similar projects due to tailoring from common standards. Quality of organizational process assets themselves. Effectiveness of verification and validation processes.
• Process Points: established organizational infrastructure for defining, maintaining, and improving standard processes (OPF, OPD). Organizational training programs aligned with standard processes (OT). Mechanisms for organizational learning and sharing best practices. Proactive risk management (RSKM). Formal decision-making processes (DAR).
• Contracts/Legals: indirectly, organizational standard processes might include guidelines for supplier selection or contract reviews, building upon SAM at ML2.

Evidence of organizational standardization, tailoring, and process asset deployment is mandatory for an ML3 rating.

• Technicities: is there a documented set of standard processes? How are they tailored? Is there a PAL, is it used and useful? How are risks and decisions typically handled across projects? Are engineering processes defined and followed?
• Controls: how does the organization ensure its standard processes are used and improved? Is there evidence of integrated project management based on tailored organizational processes?
• Quality: consistency of process application. Are the organizational assets helpful? Are VER/VAL activities consistently applied?
• Process Points: are there mechanisms for organizational learning (OPF, OPD)? Is relevant training provided (OT)? Is risk management systematically approached (RSKM)?

Assesses the maturity of organizational process infrastructure and consistency of its use.

• Technicities/Controls: adoption and practical application of a specific organizational standard process element by a project. Usefulness of a specific PAL asset.
• Quality: fif-for-purpose of a specific organizational asset or tailoring guideline in a specific project context. Effectiveness of a specific engineering practice (e.g., peer review process defined organizationally).

Checks health, adoption, or effectiveness of a very specific part of the ML3 infrastructure.

• Technicities: statistically established process performance baselines (PPBs) and models (PPMs) for selected subprocesses. Use of statistical process control (SPC) techniques. quantitative objectives for quality and process performance set for projects, based on organizational objectives and PPBs. Data collection and analysis capabilities for quantitative management.
• Controls: projects select critical subprocesses for quantitative management. Monitoring and control of these subprocesses using statistical methods to understand and manage variation. Prediction of project performance based on quantitative data.
• Quality: subprocesses are statistically stable and capable. Reduced process variation. Objective data used to manage quality and performance.
• Process Points: organizational infrastructure (OPP) for defining process performance objectives, selecting subprocesses, collecting and analyzing quantitative data, and deploying statistical techniques. Projects use these organizational assets for Quantitative Project Management (QPM).
• Contracts/Legals: quantitative performance data might inform supplier selection or contract performance clauses in advanced scenarios.

Deep dive into data validity, statistical methods, criteria for subprocess selection, and how quantitative insights drive project and process management is mandatory for an ML4 rating.

• Technicities: are metrics collected that could form the basis for PPBs? Is there any use of statistical analysis for process understanding? Are any subprocesses identified as critical for performance? Is data used to set performance goals?
• Controls: how is performance data used for decision-making, even if not fully statistical? Are there attempts to predict outcomes?
• Quality: are there attempts to measure process capability or understand sources of variation, even qualitatively?
• Process Points: availability of historical data, analytical skills within the organization, understanding of basic statistical concepts and process variation. Are there organizational objectives that could be cascaded to projects quantitatively

Identifies opportunities for, and readiness towards, implementing quantitative management.

• Technicities/Controls: data collection practices for a specific metric; initial attempts at charting, control limits, or statistical analysis for a focused area. Feasibility of setting quantitative objectives for a small part of a project.

Assesses feasibility or early results of a very limited quantitative effort for a narrow scope.

• Technicities: use of causal analysis techniques (e.g., root cause analysis, Pareto, fishbone) to identify and address root causes of defects and problems. Proactive identification, evaluation, piloting, and deployment of incremental and innovative process improvements based on quantitative understanding of their impact.
• Controls: organizational mechanisms for continuous improvement. Organizational performance management aligned with business objectives, using quantitative data to select and prioritize improvements. Tracking and evaluation of deployed improvements to ensure they meet objectives and are institutionalized.
• Quality: measurable improvement in process performance, quality, and business outcomes, driven by analysis and innovation. Prevention of defect recurrence through addressing root causes. Culture of innovation and empowerment.
• Process Points: organizational infrastructure for selecting and piloting improvements (OPM), deploying changes, and performing Causal Analysis and Resolution (CAR). Empowered teams for improvement. Feedback loops from CAR and OPM into organizational processes and assets.
• Contracts/Legals: highly mature organizations might collaborate with suppliers on joint process optimization initiatives based on shared quantitative goals.

Evidence of a sustained, data-driven continuous improvement culture and tangible performance benefits is mandatory for an ML5 rating.

• Technicities: how are lessons learned used to prevent future problems? Are there any formal causal analysis activities? Are innovative ideas solicited, evaluated, and explored? How are improvements deployed?
• Controls: how are process improvements proposed, prioritized, and deployed? Is their impact assessed?
• Quality: is there a focus on preventing problems and proactively optimizing performance beyond current targets?
• Process Points: existence of a culture for improvement, mechanisms for sharing innovations and lessons learned. Is there top management support for optimization efforts?

Assesses potential for or early stages of establishing an optimizing culture and systematic improvement mechanisms.

• Technicities/Controls: the process and outcome of a specific CAR effort (e.g., did they identify a valid root cause for a specific defect type?). Results and learnings from a piloted innovation (e.g., a new tool or technique).

Evaluates a single instance of an improvement, problem-solving activity, or innovation effort.