Relevant stakeholders with requirements for quality attributes are identified (business, operations, security, compliance, users).
Quality Attribute Workshop
Prerequisite
What needs to be finished first
A rough architecture sketch and architecture goal are documented so Quality Attribute Scenarios can become concrete.
Preparation
What needs to be ready before start
Whiteboard or Miro board with columns (Stakeholder, Quality Attribute, Scenario, Priority); template for Quality Attribute Scenario (Source, Stimulus, Environment, Artifact, Response, Response Measure); list of established QA categories (Performance, Availability, Security, Usability, Modifiability, Testability, Interoperability).
One facilitator with QAW experience; architect; one representative per stakeholder group; scribe; decider with mandate for priorities.
Architecture context, planned or existing system sketch; known SLA/SLO; legal and compliance constraints; business goals; learnings from similar systems.
1 day (6-8 h) for a medium-sized system
Quadrant board for scenarios. Make definition of Quality Attributes visible. Print scenario template large. Rule: scenarios are measurable and concrete, no wish lists.
Core question
The one question this method answers
Which Quality Attributes with which concrete scenarios are architecture-critical for the system, and in which order must they be addressed?
Flow
Marker: Phase
| Step | Duration | Action | Hint |
|---|---|---|---|
1Phase 1: Business context and drivers | 45 min | Architect presents system context and architecture goals (max 30 min). Stakeholders add business drivers (growth, risks, compliance, cost). These become the source for Quality Attributes. | If the presentation takes 90 min, the architecture package is not ready yet. Pause workshop and complete Inception Canvas. |
2Phase 2: Quality Attribute brainstorm | 45 min | Prioritize Quality Attributes per stakeholder, for example with Quality Tree. Go through at least 7 categories, focus on 4-6 architecturally relevant ones. | If all 7 attributes are weighted high, they are not prioritized. Forced distribution (for example 2 High, 2 Medium, rest Low) forces decision. |
3Phase 3: Generate scenarios | 120 min | 3-5 scenarios per top attribute. Strict template: Source, Stimulus, Environment, Artifact, Response, Response Measure. Example: "At 5x load spike (Stimulus) during peak (Environment), checkout API (Artifact) responds with p95 below 800 ms (Response Measure)." | If Response Measure is missing, it is not a scenario but a hope. Measurability forces clarity. |
4Phase 4: Prioritization | 60 min | Assess scenarios against business priority and technical risk (for example High/Medium/Low x High/Medium/Low matrix). Mark top-10 scenarios as architecture-critical. | Decider must decide visibly. Consensus decision in this phase leads to dilution. If uncertain, prepare separately and run second session. |
5Phase 5: Followups and tactics | 60 min | Sketch first architectural tactics per top scenario (caching, replication, auth pattern, decoupling). Owner and deadline for detailed elaboration. Link to Architecture Decision Records. | QAW does not deliver design, but requirements plus first hints. ADRs arise separately, with scenarios as input. |
Artifact
What comes out at the end
QAW report with business drivers, Quality Tree, prioritized scenario list (according to template), evaluation matrix, tactics sketches and follow-up table with ADR references, owner and deadline.
One QAW report per architecture cycle with date. Scenarios as persistent requirements, tracked across releases. Changes to priorities with rationale and decider signature.
- Miro or FigJam with QAW template
- Confluence or Notion page with scenario table
- Markdown in architecture folder under docs/quality-attributes/
- ATAM tooling or SEI templates for formal workshops
quality-attribute-workshop-working-template.md
Compact working template for Quality Attribute Workshop with context, input, output artifacts, and next step.
Quality Attribute Workshop Working Template
Goal
A structured workshop for architecture-critical quality attributes and scenarios.
Context
When and for what do we use this method?
Input
Which data, observations, decisions, or materials are available?
Execution
Short notes along the runsheet.
Output artifacts
- Quality Scenarios:
- Priority List:
- Architecture Concerns:
Assumptions and open questions
- ...
Decision / Next step
Owner, date, and success signal.
Example output
Concrete filled scenario, fictional example
quality-attribute-workshop-beispiel.md
Concrete filled scenario, fictional example
Quality Attribute Workshop - Workspot Booking Platform, 2026-05-15
Business drivers: growth target 5x bookings in 12 months. Expansion across EU. SLA 99.9%.
Top Quality Attributes (prioritized)
- Availability (High)
- Performance (High)
- Security (High)
- Modifiability (Medium)
- Interoperability (Medium)
Architecture-critical scenarios (excerpt)
- QAS-01 (Performance): At 5x standard load (Stimulus) during Peak Friday 17-19 (Environment), the booking API (Artifact) responds with p95 below 600 ms (Response Measure).
- QAS-02 (Availability): On failure of one AZ (Stimulus) during live operation (Environment), the platform (Artifact) remains available with max 2 min recovery time (Response Measure).
- QAS-03 (Security): On brute-force attempt (Stimulus) against login (Artifact), the system blocks after 5 failed attempts per IP for 15 min (Response Measure).
- QAS-04 (Modifiability): Adding a new payment method (Stimulus) by backend team (Environment) is possible in maximum 3 person-days without changing other modules (Response Measure).
Tactics sketches
- QAS-01: edge caching, read replicas. ADR-014 in preparation. Owner: @lisa.
- QAS-02: multi-AZ deployment, lower health probe threshold. ADR-015. Owner: @marcus.
- QAS-03: rate limiter with IP-based blocking. ADR-016. Owner: @ben.
Decider: @julia (CTO). Next review: 2026-06-30.
Pitfalls
Recognize symptoms and steer against them
Wish lists instead of scenarios
Entries are "must be fast" or "must scale" without thresholds.
Enforce template strictly. All six fields per scenario. If Response Measure is missing, do not include it in the artifact.
All attributes high
Stakeholders prioritize nothing, everything is equally important.
Forced distribution. Maximum two attributes at High. Decider must decide ties. Consensus is not the goal.
Architect alone in the room
No business or operations stakeholders present, scenarios are technology-heavy.
Workshop only with at least one business, operations and security representative. Otherwise postpone or add conference call.
Tactics too early
Discussion jumps into solution design, scenarios are not completed.
Phase 5 only after prioritization is complete. Solution ideas go to parking lot, not mixed into scenarios.
No reference to ADRs
Scenarios remain in report, architecture decisions do not reference them.
Every ADR contains references to addressed scenarios. If inconsistent, update scenario or have ADR explain why it deviates.
Stop criteria
Done signals checkable in under a minute
Finished the runsheet?
Go to the profile for purpose, similar methods, and sources or continue to the next method in the catalog.