Current WSG auditability¶
This inventory classifies all 197 success criteria across 71 current WSG guidelines by the evidence and judgment needed to assess them. It uses the published WSG draft, with the 16 October 2026 publication pinned in the guideline-review skill. These are project assessment classes, not W3C classifications, conformance results or a count of implemented checks.
The five classes build on Thibaud Colas’s assessment of WSG audit types. Each decision below applies to the current criterion wording. Current criterion IDs and links are the index; guideline summaries collect their constituent classes.
What the classes mean¶
| Primary class | Criteria | What is needed |
|---|---|---|
| Static analysis | 3 | Inspect code, declarations or configuration using structural rules; no interactive journey is needed. Publicly delivered code can be sufficient. |
| Automated | 6 | Run a repeatable runtime or HTTP procedure with explicit observable assertions and a declared scope. |
| Manual quantitative | 48 | Define a task, comparison or threshold, then interpret measurements. Scripts and agents can execute the procedure. |
| Manual qualitative | 50 | Interpret meaning, relevance, usability, architecture or adequacy using public-facing behavior or supplied material. |
| Consulting | 90 | Assess organizational decisions, governance, resourcing, specialist context or sustained practice. Published records may suffice; otherwise request owner evidence. |
Primary class means the evidence needed to judge the complete current criterion within the stated audit scope. It is not the easiest detectable failure. A required title can be detected automatically, for example, while deciding whether optional head elements improve usability needs interpretation. Supporting methods identify useful partial checks; they do not make the entire criterion automatic.
“Manual” and “consulting” describe the kind of assessment, not who must perform it. An agent can conduct measured tasks, review qualitative evidence and inspect project records. It must retain uncertainty where evidence is missing. A public policy can support a policy-disclosure criterion; it does not establish that an operational process happens. Conversely, a criterion asking only for a published commitment should not be expanded into an audit of its execution.
Evidence access is separate from audit method. Database query comparisons can be quantitative even when they require backend traces; source review can be qualitative; a bot opt-out test can be automated even when the bot is not publicly runnable. Consulting is used where organizational context or sustained practice is integral, not simply because a file is private.
A guideline can contain several classes. Its summary below counts each criterion once by its primary class. Supporting methods do not enter those counts. The implemented check reference separately records what this audit system can currently collect and judge; classification is not implementation or complete guideline coverage.
Using the inventory¶
- Select current criteria for the site, user tasks and audience. Read the linked WSG wording and supporting context before narrowing an expectation.
- Choose representative pages, states and records. Collect repeatable facts with deterministic tools, then use the stated quantitative or qualitative review route.
- Give reviewers the criterion, declared scope and evidence references. Separate observation, interpretation and missing evidence; do not treat agreement among models as new evidence.
- Report tested scope and unresolved questions. Request the smallest useful additional record, task or comparison instead of converting inaccessible evidence into a failure or a pass.
This follows STAR’s evaluation process: define scope, explore, sample, evaluate and report. Public-site auditing is the default; optional source and organizational records extend what can be substantiated.
Agent reference and maintenance¶
The sustainability-guideline-review skill packages criterion text in references/wsg.json and these classifications in references/assessment-crosswalk.json. The published skill catalog exposes both. Load the relevant guideline group, not the whole inventory into every worker. The classification reference records the WSG version and content digest; full WSG attribution and licensing remain with the text reference.
Validate the current inventory against its pinned WSG reference:
A WSG update requires reviewing the affected wording and classifications together. No historic assessment download is needed to validate this inventory. This is planning metadata: an audit outcome still requires a scoped test or review and its evidence.
User experience design¶
17 guidelines, 48 criteria. Static analysis: 1; Automated: 1; Manual quantitative: 12; Manual qualitative: 16; Consulting: 18.
2.1 Identify, assess, disclose, review, and mitigate sustainability impacts¶
Read guideline 2.1. Primary classes: Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Impact analysis | Consulting | Periodic impact assessment and demonstrated progress require organizational evidence. Review: Review dated assessments, material issues, public reporting and actions across at least two review periods. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| External impact | Consulting | Supply-chain and non-digital impacts cannot be established from browser behavior. Review: Inspect the impact boundary, logistics and infrastructure evidence, reduction plan and measured follow-through. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
2.2 Understand user requirements or constraints¶
Read guideline 2.2. Primary classes: Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Audience evaluation | Consulting | Identifying affected groups and involving them requires research records. Review: Review recruitment, participant coverage, findings and decisions; compare them with the service's public tasks. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
2.3 Integrate sustainability into every stage of the ideation process¶
Read guideline 2.3. Primary classes: Consulting (4).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Sustainable brand development | Consulting | Asset measurements do not establish approved branding requirements or recurring review. Review: Compare brand requirements, asset variants and review records; measure representative published assets. Supporting methods: manual quantitative, manual qualitative. Evidence: live site, published records, source/configuration, project records. |
| Wireframes and prototypes | Consulting | A finished interface does not reveal whether early prototypes informed development. Review: Inspect prototypes, tests and decisions showing which risks or unnecessary work were avoided. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Participation and testing | Consulting | Participation throughout design requires evidence of who influenced decisions. Review: Review participant selection, session records, feedback and resulting design changes. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Environmental ideation | Consulting | Environmental boundaries and non-human needs must be evidenced in ideation decisions. Review: Inspect impact assumptions, personas or equivalent methods, and decisions traceable to them. Supporting methods: manual qualitative. Evidence: published records, project records. |
2.4 Design efficient and streamlined user journeys¶
Read guideline 2.4. Primary classes: Manual quantitative (1), Manual qualitative (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Efficient paths | Manual quantitative | Task efficiency can be compared through a defined journey and measurable steps, with usability judgment. Review: Define representative tasks; record actions, elapsed time, choices and completion, then review unnecessary friction. Supporting methods: automated, manual qualitative. Evidence: live site. |
| Patterns for efficiency | Manual qualitative | Recognizable patterns and clear journeys depend on task and audience interpretation. Review: Complete representative tasks and compare interaction patterns, consistency and explanations with familiar conventions. Supporting methods: manual quantitative. Evidence: live site. |
| Content discovery | Manual qualitative | Finding useful content and evaluating navigation clarity requires contextual judgment. Review: Test representative information-finding tasks through navigation, search, sitemap and update channels; inspect freshness. Supporting methods: automated, manual quantitative. Evidence: live site, published records. |
2.5 Design to assist and not to distract¶
Read guideline 2.5. Primary classes: Manual qualitative (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Respect user attention | Manual qualitative | Meaningful control of attention is broader than detecting notification buttons. Review: Exercise information-delivery controls and defaults during tasks; assess interruption, choice and cognitive demands. Supporting methods: automated. Evidence: live site. |
| Distraction-free design | Manual qualitative | Deciding what is relevant or safely removable requires user-task and accessibility judgment. Review: Observe initial and interactive states; assess interruptions, optional assets and essential information before suggesting removal. Supporting methods: automated, manual quantitative. Evidence: live site. |
| Remove engagement traps | Manual qualitative | Intent and unnecessary engagement cannot be inferred from scroll mechanics alone. Review: Exercise feeds, pagination and browser controls; assess whether continuation or navigation restrictions serve the task. Supporting methods: automated. Evidence: live site. |
2.6 Avoid being manipulative or deceptive¶
Read guideline 2.6. Primary classes: Manual qualitative (4).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Deceptive design patterns | Manual qualitative | Pressure, misleading claims and choice asymmetry require interpretation of the complete flow. Review: Compare alternatives, defaults, disclosures and consequences through consent, registration, purchasing and cancellation tasks. Supporting methods: automated. Evidence: live site. |
| Advertisements | Manual qualitative | Disclosure clarity, disruption and meaningful user control require qualitative review alongside measurements. Review: Inspect advertisement labels and controls, exercise dismissal and measure loading cost in representative states. Supporting methods: automated, manual quantitative. Evidence: live site. |
| Analytics and tracking | Manual qualitative | Consent behavior is observable, but deciding whether tracking is necessary needs purpose evidence. Review: Trace requests before and after consent, compare declared purposes, and request justification for unexplained tracking. Supporting methods: automated, consulting. Evidence: live site, published records, project records. |
| Search Engine Optimization | Manual qualitative | User-serving content and manipulation cannot be decided from metadata presence alone. Review: Review representative content, duplicate pages and accessibility-related metadata for usefulness and misleading optimization. Supporting methods: static analysis, automated. Evidence: live site. |
2.7 Make deliverables understandable and reusable¶
Read guideline 2.7. Primary classes: Manual qualitative (2), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Deliverables reusability | Manual qualitative | Reusable formats and practical reuse need examination of the delivered material. Review: Inspect representative deliverables for open formats, exportability, licensing and usable structure. Supporting methods: static analysis. Evidence: published records, source/configuration. |
| Deliverables documentation | Consulting | Maintained documentation and replacement processes require evidence across releases. Review: Inspect technical guidance, ownership, change history and examples of retiring or replacing obsolete information. Supporting methods: manual qualitative. Evidence: published records, source/configuration, project records. |
| Deliverables readability | Manual qualitative | Access and understandable source documentation can be reviewed, but quality is interpretive. Review: Inspect source access, comments and contributor guidance against a representative maintenance or reuse task. Supporting methods: static analysis. Evidence: published records, source/configuration. |
2.8 Use a design system for interface consistency¶
Read guideline 2.8. Primary classes: Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Design system | Consulting | Visual consistency does not establish component governance, ownership and maintenance. Review: Inspect component reuse and page payloads alongside system ownership, versions, standards and documentation history. Supporting methods: static analysis, automated, manual quantitative, manual qualitative. Evidence: live site, published records, source/configuration, project records. |
2.9 Optimize media to reduce resource use¶
Read guideline 2.9. Primary classes: Manual quantitative (3), Manual qualitative (1), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Need for media | Manual qualitative | Whether media aids understanding is a content and user-task judgment. Review: Review each sampled media item's purpose and an alternative without it; preserve meaningful and accessible content. Supporting methods: automated, manual quantitative. Evidence: live site. |
| Optimized media | Manual quantitative | Format, size and rendering comparisons are measurable but need acceptable-quality and device constraints. Review: Compare delivered dimensions, compression, formats and native playback across representative devices and user needs. Supporting methods: automated, manual qualitative. Evidence: live site. |
| Lazy loading | Manual quantitative | Choosing critical media and demonstrating deferred transfer needs a scoped interaction experiment. Review: Record above-fold requests, then interaction and scroll requests; verify critical media remains available without premature heavy downloads. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site. |
| User-controlled media | Manual quantitative | Playback, resolution and low-data choices need a repeatable interaction checklist with understandable labels. Review: Exercise initial playback, controls and lower-bandwidth alternatives; verify length, format and data-use disclosures against assets. Supporting methods: automated, manual qualitative. Evidence: live site. |
| Media management and use | Consulting | A media lifecycle policy and its application require project records. Review: Review compression, performance, retention and deletion requirements plus examples of their use. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
2.10 Ensure animation is proportionate and easy to control¶
Read guideline 2.10. Primary classes: Automated (1), Manual quantitative (2), Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Need for animation | Manual qualitative | Animation's contribution to understanding or task completion is interpretive. Review: Compare animated and reduced-motion states for meaning, task support and unnecessary distraction. Supporting methods: automated, manual quantitative. Evidence: live site. |
| Avoiding overburdening | Manual quantitative | Animation load and replay frequency can be measured against a scoped performance and attention assessment. Review: Count concurrent animations and replays, profile rendering, and test constrained devices and motion preferences. Supporting methods: automated, manual qualitative. Evidence: live site. |
| Control animation | Automated | Start, pause, stop and other controls can be exercised through explicit state assertions. Review: Trigger representative animations and verify the required controls actually change or stop their state. Supporting methods: manual qualitative. Evidence: live site. |
| Efficient animation | Manual quantitative | Comparing rendering cost needs controlled measurements and acceptable-output constraints. Review: Profile equivalent native and custom animation variants for main-thread work, layout, frame delivery and visual function. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site, source/configuration. |
2.11 Use optimized web typography¶
Read guideline 2.11. Primary classes: Static analysis (1), Manual quantitative (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Pre-installed typefaces | Static analysis | Font declarations and referenced assets identify reliance on pre-installed fonts. Review: Inspect font-family and font-face declarations; confirm a cold visit does not fetch avoidable custom fonts. Supporting methods: automated. Evidence: live site, source/configuration. |
| Web font optimization | Manual quantitative | Required axes, font complexity, fallbacks and loading behavior need measured comparison. Review: Inventory font files and ranges, compare alternatives, and test fallback readability during delayed and failed loading. Supporting methods: automated, static analysis. Evidence: live site, source/configuration. |
| Web font subsetting | Manual quantitative | Safe subsetting depends on declared language coverage and dynamic-input requirements. Review: Compare font coverage with supported scripts and content; test missing glyphs, dynamic input and incremental delivery where applicable. Supporting methods: static analysis, automated, consulting. Evidence: live site, published records, source/configuration, project records. |
2.12 Avoid unwanted notifications¶
Read guideline 2.12. Primary classes: Manual qualitative (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Need for notification | Manual qualitative | Notification necessity and preference alignment cannot be judged from permission state alone. Review: Inspect opt-in flows and delivered messages against their stated purpose; use delivery records where the public flow is insufficient. Supporting methods: automated, consulting. Evidence: live site, published records, project records. |
| Notification settings | Manual qualitative | Visible, accessible and easy controls need task-based interpretation as well as state checks. Review: Exercise notification defaults, settings, unsubscribe, logout and account closure with a representative account. Supporting methods: automated. Evidence: live site. |
2.13 Reduce the impact of downloadable and physical documents¶
Read guideline 2.13. Primary classes: Manual quantitative (3), Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Printed documents | Manual quantitative | Print layouts and page counts are measurable, while unnecessary paper depends on workflow context. Review: Print representative real content, compare page counts and usability, and review digital alternatives through the same task. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site. |
| Optimized documents | Manual quantitative | Document size and format comparisons need accessibility and suitability checks. Review: Measure downloadable files, compare compressed or HTML alternatives and verify reading and assistive-technology access. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site. |
| Optimized delivery | Manual quantitative | Reuse and generation behavior require controlled requests plus server traces; private evidence does not change the quantitative method. Review: Inspect generation and cache configuration with processing traces; test reuse and cookie-free delivery where relevant. Supporting methods: automated, static analysis. Evidence: live site, source/configuration, project records. |
| Labels and choice | Manual qualitative | Pre-download summaries and suitable language/format choices require content interpretation. Review: Inspect document links before activation and verify names, formats, sizes, summaries and available alternatives. Supporting methods: automated. Evidence: live site. |
2.14 Involve users early in the project¶
Read guideline 2.14. Primary classes: Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Early user testing | Consulting | Representative-user validation requires research and prototyping records. Review: Review testing processes, participants, real-world conditions, findings and resulting feature changes. Supporting methods: manual qualitative. Evidence: published records, project records. |
2.15 Audit and test for bugs or issues requiring resolution¶
Read guideline 2.15. Primary classes: Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Ongoing evaluation | Consulting | One audit cannot establish an ongoing multidisciplinary evaluation program. Review: Inspect dated UX, code, performance, accessibility and security reviews with issue resolution records. Supporting methods: static analysis, automated, manual quantitative. Evidence: published records, source/configuration, project records. |
| Automated tests | Consulting | A test file does not establish critical coverage or consistent execution across releases. Review: Inspect suite scope, CI configuration and recent run histories; trace representative regressions to tests. Supporting methods: automated, static analysis. Evidence: source/configuration, project records. |
| Performance testing | Consulting | Per-release monitoring and corrective action require longitudinal project evidence. Review: Review laboratory and real-user measurements across releases, bottleneck investigations and verified fixes. Supporting methods: automated, manual quantitative. Evidence: published records, source/configuration, project records. |
2.16 Validate usability through testing and real-world usage¶
Read guideline 2.16. Primary classes: Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Usage changes | Consulting | Feature feedback, adoption and churn are normally project-specific usage evidence. Review: Review scoped analytics and feedback with decisions showing how they informed releases. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Usability testing | Consulting | Repeated user research and measured follow-through require participant and delivery evidence. Review: Trace usability findings and real-user metrics to changes, then inspect post-release validation. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
2.17 Provide cross-platform compatibility support¶
Read guideline 2.17. Primary classes: Manual quantitative (1), Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Compatibility documentation | Consulting | Public support lists can be read, but regular review requires versioned maintenance evidence. Review: Review supported platforms and versions alongside dated updates and the process for keeping them current. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Maintaining compatibility | Consulting | Avoiding planned obsolescence requires support decisions and security/operational tradeoffs over time. Review: Compare supported releases, compatibility tests, security constraints and user-facing update explanations. Supporting methods: automated, manual quantitative, manual qualitative. Evidence: live site, published records, source/configuration, project records. |
| User constraints | Manual quantitative | A declared browser, device and network matrix makes compatibility checks repeatable. Review: Run key tasks across representative constrained connections, browsers and older devices; record failures and limits of emulation. Supporting methods: automated. Evidence: live site. |
Web development¶
16 guidelines, 45 criteria. Static analysis: 2; Automated: 4; Manual quantitative: 16; Manual qualitative: 14; Consulting: 9.
3.1 Set goals based on performance and energy impact¶
Read guideline 3.1. Primary classes: Manual quantitative (1), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Performance goals | Consulting | Choosing project-specific limits requires declared goals and ownership rather than observed page size alone. Review: Inspect resource and performance targets, rationale, owners and the thresholds used to judge delivery. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Energy intensity | Manual quantitative | Lower-impact choices require matched resource measurements and a user-value comparison. Review: Compare equivalent components under controlled conditions; retain energy-model limits and the value of intensive features. Supporting methods: automated, manual qualitative. Evidence: live site, source/configuration. |
3.2 Minify and remove unused code¶
Read guideline 3.2. Primary classes: Automated (1), Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Minified code | Automated | Delivered code and data can be checked for minification and functional regressions within the sample. Review: Inspect response bodies for avoidable minification overhead and rerun representative functionality after transformation. Supporting methods: static analysis. Evidence: live site. |
| Remove redundancy | Manual quantitative | Non-execution in one visit is not proof that code is unused; coverage needs a scoped task matrix. Review: Collect coverage across initial and interactive states, inspect build/source reachability and verify candidate removal with regressions. Supporting methods: automated, static analysis. Evidence: live site, source/configuration. |
3.3 Modularize bandwidth-heavy components¶
Read guideline 3.3. Primary classes: Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Code splitting | Manual quantitative | Loading boundaries and splitting overhead require comparison across task and cache states. Review: Trace module downloads during tasks, compare cold and repeat visits, and assess excessive requests or duplicated chunks. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site, source/configuration. |
3.4 Avoid redundancy and duplication in code¶
Read guideline 3.4. Primary classes: Manual qualitative (2), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Remove or simplify | Manual qualitative | Safe structural simplification requires qualitative source review, supported by duplication and performance measurements. Review: Inspect duplicated logic, shared abstractions, tests and before/after performance and maintainability evidence. Supporting methods: static analysis, manual quantitative. Evidence: source/configuration, project records. |
| Iteration over recreation | Consulting | Whether rebuilding is justified depends on decision history, effort and projected benefits. Review: Compare reuse and replacement options using user value, performance, scalability and maintenance evidence. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Organize code arrangement | Manual qualitative | The balance between reuse, complexity and maintainability requires source-level interpretation rather than duplicate counts alone. Review: Inspect representative CSS/JavaScript refactors, dependencies, tests and performance rather than duplicate counts alone. Supporting methods: static analysis, manual quantitative. Evidence: source/configuration, project records. |
3.5 Treat third parties the same as first parties¶
Read guideline 3.5. Primary classes: Automated (1), Manual quantitative (1), Manual qualitative (1), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Assess and reduce | Consulting | Supplier standards, dependency risks and early assessments require procurement and governance evidence. Review: Inventory third parties, measure delivery costs, and review necessity, contractual standards and exit or portability options. Supporting methods: automated, manual quantitative, manual qualitative. Evidence: live site, published records, project records. |
| Third-party implementation | Automated | Interaction-gated third-party requests and available alternatives can be asserted in a browser. Review: Record requests before interaction and after activation; verify a direct or lightweight alternative where offered. Supporting methods: manual qualitative. Evidence: live site. |
| Self-hosting | Manual qualitative | Practical self-hosting requires interpreting infrastructure control, requirements and feasible delivery alternatives. Review: Inventory asset origins and review ownership, hosting constraints, reliability and relevant self-hosting alternatives. Supporting methods: automated, manual quantitative. Evidence: live site, published records, source/configuration, project records. |
| Third-party preferences | Manual quantitative | Working opt-out controls require a defined before/after interaction and request comparison. Review: Exercise each non-essential third-party control and verify its effect on requests and functionality. Supporting methods: automated, manual qualitative. Evidence: live site. |
3.6 Ensure code follows good semantic practices¶
Read guideline 3.6. Primary classes: Static analysis (1), Manual qualitative (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Semantic code | Static analysis | Markup validity and standard conformance have deterministic structural checks. Review: Validate representative delivered and generated markup; distinguish syntax failures from broader accessibility judgments. Supporting methods: automated. Evidence: live site, source/configuration. |
| Optional features | Manual qualitative | Omitting syntax safely depends on readability, accessibility and rendering tradeoffs. Review: Compare candidate omissions with validator, browser and accessibility results; retain syntax that improves clarity or compatibility. Supporting methods: static analysis, automated, manual quantitative. Evidence: live site, source/configuration. |
| Non-standard code | Manual qualitative | Exceptions for compatibility or accessibility need justification beyond deprecation detection. Review: Inventory obsolete or proprietary features and polyfills, then assess supported-browser needs and removal evidence. Supporting methods: static analysis, automated. Evidence: live site, source/configuration. |
| Custom code | Manual qualitative | Whether a native feature meets the requirement needs functional and maintainability judgment. Review: Compare custom components with platform alternatives for required behavior, accessibility, reuse and resource use. Supporting methods: static analysis, automated, manual quantitative. Evidence: live site, source/configuration. |
3.7 Defer the loading of non-critical resources¶
Read guideline 3.7. Primary classes: Manual quantitative (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Asynchronous code | Manual quantitative | Criticality and render effects require timed observations, not only async/defer attributes. Review: Record the loading waterfall and visual states under constrained networks; verify deferred assets do not break essential rendering. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site. |
| Optimized loading | Manual quantitative | The value of priority hints depends on measured competition and loading order. Review: Compare equivalent loads with and without hints; inspect request priorities, critical rendering and wasted preload transfer. Supporting methods: automated, static analysis. Evidence: live site. |
3.8 Structure metadata for machine readability¶
Read guideline 3.8. Primary classes: Manual qualitative (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Required elements | Manual qualitative | Title presence is static, but useful optional head elements require context. Review: Validate head structure and review whether titles and optional elements improve the sampled task without unnecessary work. Supporting methods: static analysis, automated. Evidence: live site. |
| Meta tags | Manual qualitative | Metadata must be relevant and meaningful, not merely present or syntactically valid. Review: Inspect browser and sharing metadata against page content and intended use; flag missing, misleading or redundant values. Supporting methods: static analysis. Evidence: live site. |
| Structured data | Manual qualitative | Schema validity is automatable, while faithful interpretation and reusable meaning need review. Review: Parse structured data, compare entities and properties with visible content, and test intended consumer use. Supporting methods: static analysis, automated. Evidence: live site. |
3.9 Use media queries that support sustainability goals¶
Read guideline 3.9. Primary classes: Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Media and preference queries | Manual quantitative | Preference support needs a repeatable state matrix and a judgment of where each preference benefits users. Review: Exercise supported color, contrast, motion, data, transparency, print and scripting states; compare behavior and resource use. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site. |
3.10 Ensure layouts work for different devices and requirements¶
Read guideline 3.10. Primary classes: Manual quantitative (2), Manual qualitative (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Device-adaptable | Manual quantitative | Cross-device behavior requires representative viewport, capability and task measurements. Review: Run essential tasks across device sizes and missing-feature states, recording overflow, blocked actions and functional loss. Supporting methods: automated, manual qualitative. Evidence: live site. |
| Progressive enhancement | Manual qualitative | Core functionality and acceptable enhancement need task-based interpretation. Review: Complete key tasks with scripts and styles separately unavailable; distinguish lost decoration from blocked core service. Supporting methods: automated, static analysis. Evidence: live site. |
| Carbon-aware design | Manual quantitative | Carbon-triggered variants need controlled signals and measured effects with user choice preserved. Review: Exercise high/low-carbon or load states, compare resource use, and verify explanations and restoration controls. Supporting methods: automated, manual qualitative, consulting. Evidence: live site, published records, source/configuration, project records. |
| Alternative browsing | Manual qualitative | Useful non-visual and indirect interaction depends on task equivalence and accessibility. Review: Test representative tasks with assistive technologies, reader modes and relevant alternative inputs; record functional gaps. Supporting methods: automated, manual quantitative. Evidence: live site. |
3.11 Use sustainable JavaScript and APIs¶
Read guideline 3.11. Primary classes: Manual quantitative (2), Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Sustainable JavaScript | Manual quantitative | Unnecessary calculations, DOM work and requests need task-scoped profiling. Review: Profile interactions and idle states; compare required output with repeated computation, DOM changes and network activity. Supporting methods: automated, static analysis. Evidence: live site, source/configuration. |
| Energy-relevant APIs | Manual quantitative | An API's presence does not prove that it reduces energy or improves efficiency. Review: Compare matched scenarios using relevant APIs, such as visibility-driven suspension, against resource use and functionality. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site, source/configuration. |
| API requests | Manual qualitative | Request necessity and minimal payloads depend on the task and data contract; client and server traces provide supporting measurements. Review: Inspect client and server traces, payload fields and purposes; measure redundant calls and over-fetching in representative tasks. Supporting methods: automated, manual quantitative, static analysis. Evidence: live site, source/configuration, project records. |
3.12 Use dependencies sparingly and maintain them¶
Read guideline 3.12. Primary classes: Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Dependency management | Consulting | Regular removal and maintenance require a dependency inventory and project history. Review: Review manifests, usage, removal decisions and dated dependency reviews rather than treating loaded bundles as the complete inventory. Supporting methods: static analysis, automated. Evidence: source/configuration, project records. |
| Dependency necessity | Consulting | Necessity, security assurance and maintained inventories require project decisions and records. Review: Compare package/module costs and alternatives with security constraints, usage, inventory and review evidence. Supporting methods: static analysis, automated, manual quantitative, manual qualitative. Evidence: live site, source/configuration, project records. |
| Dependency updates | Consulting | Current versions are detectable, but regular maintenance needs complete inventory and review history. Review: Compare lockfiles and deployed versions with supported releases, then inspect update cadence and justified exceptions. Supporting methods: static analysis, automated. Evidence: source/configuration, project records. |
3.13 Include expected and beneficial files¶
Read guideline 3.13. Primary classes: Automated (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Expected files | Automated | Named standard-file presence, response and basic syntax can be checked deterministically. Review: Request the expected files at their defined locations, validate responses and record applicability or missing standard support. Supporting methods: static analysis. Evidence: live site. |
| Beneficial files | Automated | Discovery and syntax of named standard files are repeatable public checks. Review: Fetch applicable beneficial files, validate recognized formats and distinguish presence from truth of any declarations. Supporting methods: manual qualitative. Evidence: live site. |
3.14 Use the most efficient solution for your service¶
Read guideline 3.14. Primary classes: Manual quantitative (2), Manual qualitative (2), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Appropriate Implementation | Manual qualitative | Balancing implementation effort, runtime impact and user value is an architectural judgment supported by measured costs. Review: Compare implementation options using build/runtime measurements, maintenance effort, dependencies and user value. Supporting methods: manual quantitative. Evidence: published records, source/configuration, project records. |
| Optimized methodology | Manual qualitative | Choosing native, existing or custom solutions requires interpreting performance, maintenance and security tradeoffs. Review: Review native, existing and custom options against measured performance, tooling cost, support and trusted security components. Supporting methods: static analysis, manual quantitative. Evidence: published records, source/configuration, project records. |
| Static over dynamic | Manual quantitative | Static versus dynamic delivery can be compared quantitatively against declared update and functionality requirements. Review: Inspect rendering architecture, update needs and server measurements; compare a feasible static or cached alternative. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site, source/configuration, project records. |
| Extensions and plugins | Consulting | Regular compatibility, efficiency and accessibility review requires plugin inventory and maintenance history. Review: Inspect installed extensions and themes, review records and representative regressions or performance comparisons. Supporting methods: static analysis, automated, manual quantitative, manual qualitative. Evidence: live site, source/configuration, project records. |
| Interface components | Manual quantitative | Component overhead can be measured, while avoidable complexity requires a functionally equivalent comparison. Review: Profile component bundles, rendering and interactions; compare simpler alternatives that preserve required behavior. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site, source/configuration. |
3.15 Use the latest stable language version¶
Read guideline 3.15. Primary classes: Static analysis (1), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Versioning | Static analysis | Declared language and framework versions can be compared mechanically with stable releases. Review: Inspect authoritative runtime/build version declarations and compare them with upstream releases; verify deployed identity separately. Supporting methods: automated, consulting. Evidence: source/configuration, project records. |
| Language choice | Consulting | Language suitability depends on team capability, adoption costs and task-specific requirements. Review: Review language alternatives, workload measurements, maintenance ecosystem, costs and effects on team well-being. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
3.16 Reduce the number and complexity of database queries¶
Read guideline 3.16. Primary classes: Manual quantitative (3), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Efficient database querying | Manual quantitative | Query plans and workload traces support repeatable comparisons of retrieval, filtering and database work. Review: Inspect query traces and plans for key workloads; compare filtering, field selection and query counts at the database layer. Supporting methods: static analysis. Evidence: source/configuration, project records. |
| Reuse of retrieved data | Manual quantitative | Repeated retrieval is measurable within a declared request or process; source and traces expose the relevant work. Review: Trace representative requests or jobs and identify repeated retrieval of the same data before validating reuse changes. Supporting methods: static analysis. Evidence: source/configuration, project records. |
| Data minimization in queries | Manual quantitative | Returned fields and rows can be compared with declared data needs using source and workload traces. Review: Compare query projections and returned rows with application use; verify reduced retrieval preserves required output. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
| Data lifecycle controls | Consulting | Retention, de-identification and purpose limitation require data governance and execution records. Review: Inspect data purposes, TTL policies, identifiers and deletion evidence across the relevant backend stores. Supporting methods: static analysis, manual quantitative, manual qualitative. Evidence: published records, project records. |
Hosting, infrastructure and systems¶
12 guidelines, 34 criteria. Manual quantitative: 17; Manual qualitative: 4; Consulting: 13.
4.1 Use sustainable hosting¶
Read guideline 4.1. Primary classes: Consulting (6).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Provider selection | Consulting | Transparent provider claims alone do not establish sustainability-led selection decisions. Review: Compare provider disclosures and metrics with procurement criteria, service scope and selection rationale. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Efficient operation | Consulting | Infrastructure efficiency over time requires operational monitoring and actions. Review: Review utilization, energy/water indicators and capacity changes across representative periods and workloads. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Equipment longevity | Consulting | Hardware life extension and procurement practices require asset and maintenance evidence. Review: Inspect hardware age, utilization, patching, repairs and supplier longevity commitments. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Low-carbon electricity | Consulting | Hosting electricity claims need location, supply and operational boundaries beyond a green label. Review: Review grid/location data, on-site generation and backup supply with hosting decisions and reporting methodology. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Remaining emissions | Consulting | Valid market-based claims depend on certificates, retirement and time/location matching. Review: Inspect electricity volumes and certificates, matching rules, retirement evidence and double-counting controls. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Domain names | Consulting | Registrar disclosures can be public, but their use in registration decisions needs project evidence. Review: Check registrar/registry impact information and review how missing or available evidence informed domain choices. Supporting methods: manual qualitative. Evidence: published records, project records. |
4.2 Optimize caching and support offline access¶
Read guideline 4.2. Primary classes: Manual quantitative (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Client-side caching | Manual quantitative | Correct reuse requires initial, repeat, update and authorization-state experiments. Review: Compare cold/warm requests, changed content and logout states; check reuse without stale content or unauthorized access. Supporting methods: automated, static analysis. Evidence: live site. |
| Server-side caching | Manual quantitative | Cache benefits and correctness can be compared through controlled workload, update and authorization tests with server evidence. Review: Inspect caching layers and processing evidence, then test updates and authorization transitions for correctness. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site, source/configuration, project records. |
| Offline access | Manual quantitative | Meaningful essential functionality must be defined before offline behavior can be tested. Review: Complete agreed essential tasks after network loss, including recovery and stale-data handling. Supporting methods: automated, manual qualitative. Evidence: live site. |
4.3 Reduce data transfer with compression¶
Read guideline 4.3. Primary classes: Manual quantitative (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Server-side compression | Manual quantitative | Compression savings and processing tradeoffs require comparison rather than a header-only check. Review: Compare compressed and uncompressed representative file types for transfer, loading and processing, avoiding already-compressed assets. Supporting methods: automated, static analysis. Evidence: live site. |
| Media compression | Manual quantitative | Useful compression requires size comparisons while preserving acceptable media quality. Review: Inspect representative asset formats and compression opportunities; compare variants for quality, transfer and compatibility. Supporting methods: automated, static analysis, manual qualitative. Evidence: live site. |
4.4 Setup necessary error pages and redirection links¶
Read guideline 4.4. Primary classes: Manual quantitative (1), Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Error pages | Manual qualitative | Helpful recovery and safe disclosure require interpretation of error states. Review: Trigger representative errors, assess explanations and recovery paths, and inspect exposure of sensitive or internal details. Supporting methods: automated. Evidence: live site. |
| Redirection | Manual quantitative | Chains and loops are measurable, but necessity, safety and proportionality need review. Review: Crawl and exercise relevant links, validate destinations, measure chains and review intentional redirects against user and maintenance value. Supporting methods: automated, manual qualitative. Evidence: live site. |
4.5 Avoid maintaining unnecessary virtualized environments or containers¶
Read guideline 4.5. Primary classes: Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Unused environments | Manual quantitative | Utilization and workload measurements can identify redundant capacity once required isolation and security constraints are defined. Review: Inspect environment inventory, workload/utilization records and security constraints before proposing shutdown or consolidation. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
4.6 Use automation wisely¶
Read guideline 4.6. Primary classes: Manual qualitative (1), Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Task automation | Manual qualitative | Appropriate automation and avoidable complexity need workflow interpretation, with run costs and source supporting the judgment. Review: Review CI/CD definitions and run histories; compare duplicated work, maintenance effort and platform impact. Supporting methods: static analysis, manual quantitative. Evidence: source/configuration, project records. |
| Necessary tasks | Consulting | Task necessity and ongoing review depend on the purpose and execution history of automation. Review: Inspect triggers, frequency and run results, then trace each job to a required outcome and remove proven duplication. Supporting methods: static analysis, manual quantitative. Evidence: source/configuration, project records. |
| Automated scaling | Consulting | Capacity optimization depends on infrastructure utilization, workload priorities and operating constraints. Review: Review demand and allocation histories, autoscaling policies and controlled load tests for overprovisioning and contention. Supporting methods: manual quantitative. Evidence: source/configuration, project records. |
| Suspicious activity management | Consulting | Effective bot restrictions need activity logs and decisions about legitimate access. Review: Inspect crawler policies, control outcomes and logs; test representative authorized access without assuming all automation is harmful. Supporting methods: automated, manual quantitative, manual qualitative. Evidence: published records, project records. |
4.7 Define the frequency of data refreshes¶
Read guideline 4.7. Primary classes: Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Refresh frequency | Manual quantitative | Refresh schedules can be compared quantitatively against declared freshness needs and resource costs. Review: Compare freshness requirements, refresh schedules and observed polling/cache behavior with measured resource use. Supporting methods: automated, manual qualitative. Evidence: live site, published records, project records. |
4.8 Back up critical data at routine intervals¶
Read guideline 4.8. Primary classes: Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Data backups | Manual quantitative | Defined backup, restore, retention and access tests can assess security and incremental behavior using authorized operational evidence. Review: Review backup scope, storage/access settings, retention and restore tests, including user-identifiable data handling. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
4.9 Assess the impact and requirements of data processing¶
Read guideline 4.9. Primary classes: Manual quantitative (5).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Carbon shifting | Manual quantitative | Controlled scheduling and placement comparisons can measure carbon-aware behavior within explicit residency, performance and security constraints. Review: Compare job timing/location and grid signals with measured energy, latency, security and data-residency requirements. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
| Protocol efficiency | Manual quantitative | Protocol support is detectable, but selecting efficient secure options requires scoped comparisons. Review: Inventory negotiated protocols and legacy fallbacks; compare transfer performance against compatibility and security requirements. Supporting methods: automated, manual qualitative. Evidence: live site, published records, source/configuration. |
| Event-driven architecture | Manual quantitative | Equivalent server workloads allow quantitative comparison of event-driven and synchronous architectures. Review: Compare event-driven and synchronous designs with workload traces, processing, power, performance and operational complexity. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
| Data processing efficiency | Manual quantitative | Equivalent workloads and output requirements allow measured comparisons of processing placement and redundant work. Review: Trace processing stages and compare feasible placements under equivalent output, workload and security conditions. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
| Efficient location | Manual quantitative | Application data flows and traces support comparisons of processing placement and serialization overhead. Review: Inspect data flows and measure transformation or serialization overhead near and far from the source. Supporting methods: static analysis, manual qualitative. Evidence: source/configuration, project records. |
4.10 Use Content Delivery Networks (CDNs) when beneficial¶
Read guideline 4.10. Primary classes: Manual quantitative (2), Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Global CDNs | Manual quantitative | CDN benefit can be compared for a declared audience and workload using delivery and provider-impact measurements. Review: Compare delivery measurements by audience region with hosting/CDN footprints, caching behavior and local-hosting alternatives. Supporting methods: automated, manual qualitative. Evidence: live site, published records, project records. |
| Sustainability commitment | Manual qualitative | Credibility of published CDN commitments can be assessed from attributed evidence. Review: Compare scoped provider targets, methods, verification and progress; separate commitments from demonstrated reductions. Supporting methods: consulting. Evidence: published records. |
| Inappropriate resources | Manual quantitative | Dynamic-resource suitability requires measured caching behavior and update characteristics. Review: Compare CDN hits, freshness, CORS and authorization behavior for changing resources; inspect configuration where public traces are insufficient. Supporting methods: automated, consulting. Evidence: live site, source/configuration, project records. |
4.11 Ensure infrastructure fits project requirements¶
Read guideline 4.11. Primary classes: Manual quantitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Lowest requirements | Manual quantitative | Capacity can be compared with demand once service, recovery, security and resilience requirements are explicit. Review: Compare allocation with workload peaks, recovery/security obligations and tested headroom rather than average utilization alone. Supporting methods: manual qualitative. Evidence: published records, project records. |
4.12 Store data according to the needs of your users¶
Read guideline 4.12. Primary classes: Manual qualitative (1), Consulting (4).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Remove unused data | Consulting | Obsolete-data removal requires inventory, usage and retention evidence. Review: Inspect data-age/access records, justified retention and completed deletion or archiving actions. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Lifecycle management | Consulting | Storage tiers and lifecycle rules require operational policy and configuration evidence. Review: Review data classification, TTL/archive rules and access frequency against storage placement and executed transitions. Supporting methods: static analysis, manual quantitative. Evidence: published records, project records. |
| Efficient storage | Consulting | Lasting value and safe removal depend on operational, analytical and legal purposes. Review: Compare stored datasets, duplicates and compression with use, regeneration cost and retention requirements. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Optimized logging | Consulting | Logging scope, backups, privacy and scheduling depend on operational records. Review: Review log purpose, volume, retention, access, rotation and processing schedules; trace monitoring to useful corrective action. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Asset downloads | Manual qualitative | Useful downloadable alternatives depend on the asset and repeat-use task. Review: Identify large durable assets, test download and offline reuse, and assess whether the alternative reduces repeated retrieval. Supporting methods: automated, manual quantitative. Evidence: live site. |
Business strategy and product management¶
26 guidelines, 70 criteria. Automated: 1; Manual quantitative: 3; Manual qualitative: 16; Consulting: 50.
5.1 Have an ethical and sustainable product strategy¶
Read guideline 5.1. Primary classes: Manual qualitative (2), Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Public documents | Manual qualitative | Published policy clarity, accessibility and versioning can be assessed from public documents. Review: Discover the relevant policy set, inspect language and accessible formats, and compare versions and declared scope. Supporting methods: automated. Evidence: published records. |
| Achievements and compliance | Manual qualitative | The criterion asks for useful public disclosure, which can be reviewed directly. Review: Inspect the dedicated sustainability section for relevant achievements, evidence, updates and clear scope. Supporting methods: automated. Evidence: published records. |
| Governance over time | Consulting | Implementation and monitoring over time require operational records, not a policy alone. Review: Trace goals through dated decisions, measurements, accountability and corrective actions across review periods. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Technology legislation | Consulting | Legal compliance, advocacy and going beyond minimum obligations need organizational and specialist evidence. Review: Review applicable obligations, compliance evidence, advocacy and data-minimization decisions with qualified input where needed. Supporting methods: manual qualitative. Evidence: published records, project records. |
5.2 Assign a sustainability advocate¶
Read guideline 5.2. Primary classes: Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Advocate for sustainability | Consulting | A named advocate does not establish expertise, authority, time or resources. Review: Inspect the assigned role, responsibilities, digital expertise, allocation and evidence of active work. Supporting methods: manual qualitative. Evidence: published records, project records. |
5.3 Inform, raise awareness, and train for sustainability¶
Read guideline 5.3. Primary classes: Consulting (5).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Inform and aware | Consulting | Making materials available does not prove delivery to all relevant participants. Review: Review onboarding and workshops, intended audiences, delivery records and coverage gaps. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Routine training | Consulting | Regular active skills development requires participation and recurrence evidence. Review: Inspect training schedules, attendance, refreshed content and how learning supports project sustainability work. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Active participation | Consulting | Encouragement and organizational support require evidence of actual opportunities and resources. Review: Review participation channels, shared initiatives, provided resources and examples of supported action. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Training materials | Consulting | Suitable training content is reviewable, but delivery and maintenance require records. Review: Inspect materials against current policies, version history and evidence of delivery to relevant audiences. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Incentivize progress | Consulting | Time allocation and incentives require evidence of organizational practice. Review: Review recognition, allocated learning time, benefits and progress tracking without assuming an announcement proves delivery. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.4 Communicate the environmental impact of user choices¶
Read guideline 5.4. Primary classes: Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Communication of impact | Manual qualitative | Useful, understandable environmental feedback and meaningful choices require content judgment. Review: Exercise relevant decisions, inspect impact explanations and verify the assumptions behind displayed comparisons. Supporting methods: manual quantitative. Evidence: live site. |
5.5 Calculate the environmental impact¶
Read guideline 5.5. Primary classes: Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Life-cycle assessment | Consulting | A complete LCA needs a defined functional unit and lifecycle inventory. Review: Inspect boundaries, inventory, methods, factors, exclusions and evidence supporting lifecycle impact claims. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Competitor impact | Consulting | Comparability and justified additional impact depend on service value and aligned assessment boundaries. Review: Compare alternatives using equivalent functional units, lifecycle scopes and user value, recording data gaps. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Tooling impact | Consulting | Pipeline-wide tooling impact needs development and third-party usage evidence. Review: Inventory build, development and external tools; assess measured or estimated lifecycle impacts with declared boundaries. Supporting methods: manual quantitative. Evidence: source/configuration, project records. |
5.6 Define clear organizational sustainability goals and metrics¶
Read guideline 5.6. Primary classes: Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Web sustainability goals | Manual qualitative | Published goals, measurable indicators and reported progress can be reviewed directly. Review: Inspect targets, baselines, timeframes and progress reports for specificity and consistency of measurement. Supporting methods: manual quantitative. Evidence: published records. |
5.7 Validate web sustainability efforts through external verification¶
Read guideline 5.7. Primary classes: Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Obtaining verification | Consulting | Independent assurance requires scope, independence and verification evidence. Review: Inspect assurance statements, verifier independence, methods, exclusions and the claims actually verified. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Maintaining verification | Consulting | Reassessment and claim updates require a dated assurance and evidence history. Review: Compare successive verifications and new evidence with the claims currently published. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.8 Support mandatory disclosures and reporting¶
Read guideline 5.8. Primary classes: Manual quantitative (1), Manual qualitative (2), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Policies and practices | Manual qualitative | The criterion asks for published disclosure policies aligned with recognized standards. Review: Review public reporting commitments, applicable standards, coverage and clarity of disclosed procedures. Supporting methods: consulting. Evidence: published records. |
| Impact reports | Manual qualitative | Annual publication and comparisons can be assessed through publicly available report series. Review: Locate annual reports, check reporting periods and compare progress claims with prior reports and goals. Supporting methods: automated, manual quantitative. Evidence: published records. |
| Disclosure reporting | Consulting | Transparent reporting practices require evidence of how figures are produced and maintained. Review: Inspect reporting controls, data sources, boundaries, review responsibilities and publication history. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Impact reduction | Manual quantitative | Reduction claims need quantitative reconciliation and qualitative checks for misleading boundaries. Review: Recalculate disclosed reductions, compare baselines and scopes, and investigate exclusions, double counting and selective presentation. Supporting methods: manual qualitative, consulting. Evidence: published records, project records. |
5.9 Create one or more impact business models¶
Read guideline 5.9. Primary classes: Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Theory of change | Consulting | Revenue-impact relationships and iterative improvement need business-model evidence. Review: Review revenue streams, impact pathways, measurement methods and decisions demonstrating improvement over time. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.10 Follow a product management and maintenance strategy¶
Read guideline 5.10. Primary classes: Manual qualitative (1), Consulting (4).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Management and maintenance | Manual qualitative | The required management approach can be assessed as a maintained document. Review: Inspect product maintenance documentation for clear responsibilities, lifecycle scope and revision history. Supporting methods: consulting. Evidence: published records. |
| Planning strategy | Consulting | Plans must be implemented and maintained across products and services. Review: Review maintenance/security plans, resourcing and records showing sustainability priorities used in delivery. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Resourcing products | Consulting | Staffing, budgets and sustained maintenance cannot be inferred from a working site. Review: Inspect resourcing commitments, technical-debt work and continuity of support across the product lifecycle. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Resource measurement | Consulting | Maintenance-integrated measurement and decision incentives need longitudinal project evidence. Review: Review resource metrics, baseline changes and maintenance decisions; assess whether incentives encourage useful outcomes. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Failure indicators | Consulting | Documented risk indicators also need active monitoring and corrective action. Review: Inspect KFI definitions, thresholds, monitoring records and examples of risks influencing decisions. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.11 Implement continuous improvement procedures¶
Read guideline 5.11. Primary classes: Consulting (4).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Continuous improvement | Consulting | Embedded improvement requires evidence connecting decisions to sustainability outcomes. Review: Trace changes to environmental and operational measurements, explanations and review decisions. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Retrospectives conducted | Consulting | Regular research, debt reduction and learning require team process evidence. Review: Inspect retrospectives, allocation of time and resulting research, maintenance or delivery changes. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Iterative consideration | Consulting | A sustained improvement record and necessary analytics require history and project objectives. Review: Review iteration decisions, analytics purposes and resource/performance consequences against user needs and stated goals. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Functionality decisions | Consulting | Lifecycle decisions about features need usage, impact and maintenance evidence. Review: Trace representative retain/change/remove decisions to user value, resource impact, usage and ownership records. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.12 Document updates and evolutions¶
Read guideline 5.12. Primary classes: Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Feature changes | Manual qualitative | Clear structured change communication can be reviewed from releases and documentation. Review: Compare release notes or change logs with sampled feature changes for consistency, versioning and practical consequences. Supporting methods: automated. Evidence: published records. |
5.13 Evaluate if a digital product or service is necessary¶
Read guideline 5.13. Primary classes: Manual qualitative (2), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Sustainable Development Goals | Manual qualitative | Published SDG alignment needs a reasoned comparison with the service and specific targets. Review: Inspect the sustainability statement and assess whether named goals and targets have a plausible, scoped relationship to the service. Supporting methods: consulting. Evidence: published records. |
| Creation evaluation | Consulting | Desirability, feasibility and viability decisions require research and business context. Review: Review alternatives, user demand, delivery constraints and the decision to create or retain the service. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Obstacle consideration | Manual qualitative | Barriers and justified safeguards require task-based accessibility and fairness judgment. Review: Exercise representative access paths under user constraints and distinguish unnecessary barriers from necessary protections. Supporting methods: automated, manual quantitative. Evidence: live site. |
5.14 Provide a supplier standards of practice document¶
Read guideline 5.14. Primary classes: Manual qualitative (1), Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Vetting potential partners | Consulting | Sustainability-led procurement requires criteria, due diligence and applied decisions. Review: Review supplier selection and management records, lifecycle considerations and responses to identified risks. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Collaborative measurement | Consulting | Collaborative impact tracking requires supplier participation and shared evidence. Review: Inspect agreed indicators, responsibilities, measurements and actions involving affected parties. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Informative partner promotion | Manual qualitative | Public partnership disclosure and its claimed collective impact are directly reviewable. Review: Inspect partnership descriptions, scope and supporting impact information; separate promotion from demonstrated outcomes. Supporting methods: consulting. Evidence: published records. |
5.15 Share economic benefits¶
Read guideline 5.15. Primary classes: Manual qualitative (1), Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Living wage | Manual qualitative | This criterion requires a public commitment, not proof of payroll compliance. Review: Check the public commitment's scope across employees, contractors and other affected parties; record any exclusions. Supporting methods: consulting. Evidence: published records. |
| Incentivization | Consulting | Embedded incentives need policy and evidence of their application. Review: Review incentive criteria, affected parties and implemented arrangements against the stated impact goals. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Employee benefits | Consulting | Actual benefits and their relation to resources require employment and organizational evidence. Review: Inspect benefit eligibility, provision and resourcing; distinguish published promises from delivered benefits. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.16 Share decision-making power with affected parties¶
Read guideline 5.16. Primary classes: Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Decision-making | Consulting | Meaningful participation, authority and human appeals require governance and process evidence. Review: Review decision rights and participation records; exercise public automated-decision controls and inspect human-review procedures. Supporting methods: manual qualitative. Evidence: live site, published records, project records. |
5.17 Use Diversity, Equity, Justice, Inclusion (DEJI) practices¶
Read guideline 5.17. Primary classes: Manual quantitative (1), Manual qualitative (1), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| DEJI practices | Manual qualitative | Published commitments and explanations can be judged for scope and relevance. Review: Inspect DEJI commitments and explanations of how they address exclusion and sustained access; distinguish explanation from verified outcomes. Supporting methods: consulting. Evidence: published records. |
| DEJI training | Consulting | Providing training requires evidence of relevant content and actual delivery. Review: Review DEJI training coverage, audiences, participation and treatment of digital exclusion and algorithmic harms. Supporting methods: manual qualitative. Evidence: published records, project records. |
| DEJI improvements | Manual quantitative | Published progress requires comparable measurements across hiring, leadership and operations. Review: Inspect indicators, denominators, periods and progress claims; request underlying records where aggregates are insufficient. Supporting methods: manual qualitative, consulting. Evidence: published records, project records. |
5.18 Promote responsible data practices¶
Read guideline 5.18. Primary classes: Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Data practices | Consulting | Public policy review cannot establish the required data protection and governance practices. Review: Compare accessible policy documents with data flows, controls, applicable standards and evidence of stronger protections where needed. Supporting methods: manual qualitative. Evidence: live site, published records, project records. |
| Data ownership | Consulting | Visible data controls do not prove retention and deletion across stores, backups and caches. Review: Exercise access/export/opt-out flows, then trace purpose, retention and deletion through relevant systems and records. Supporting methods: automated, manual quantitative, manual qualitative. Evidence: live site, published records, project records. |
| Outdated content | Consulting | An archive or schedule does not establish executed expiration and deletion across content and data. Review: Inspect published schedules, automated rules and completed review/archive/deletion examples; test lightweight archive discovery. Supporting methods: automated, manual quantitative, manual qualitative. Evidence: live site, published records, project records. |
5.19 Establish responsible practices around AI and emerging or disruptive technologies¶
Read guideline 5.19. Primary classes: Automated (1), Consulting (4).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| AI and data collection | Consulting | Dataset necessity, sourcing, validation and fairness require model/data governance evidence. Review: Inspect dataset inventories, provenance, screening, minimization, evaluation and retention for the stated task. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Business adaptation | Consulting | Support for employees affected by technological change requires workplace evidence. Review: Review consultation, training, transition resources and records of how affected staff were supported. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Environmental responsibilities | Consulting | Technology necessity and lifecycle impact require model, infrastructure and business context. Review: Review alternatives, deployment/optimization measurements and first/third-party lifecycle accounting with stated uncertainty. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Automated tooling | Automated | A controlled test can assert bot identity and opt-out behavior; access to the operated tool is separate from test determinism. Review: Inspect crawler/tool identities and configuration, then test opt-out handling and user-agent declarations in controlled scenarios. Supporting methods: static analysis. Evidence: source/configuration, project records. |
| Quantum resilience | Consulting | Cryptographic migration depends on data lifetime, sensitivity and dependency assessment. Review: Review cryptographic inventories, threat/lifetime assumptions and migration planning against applicable NIST guidance. Supporting methods: static analysis, manual qualitative. Evidence: published records, project records. |
5.20 Adopt responsible financial practices¶
Read guideline 5.20. Primary classes: Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Fuel divestment | Consulting | Actual divestment and financial relationships require financial evidence. Review: Inspect banking, sponsorship and investment policies plus records demonstrating movement away from fossil-fuel exposure. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Responsible finance | Consulting | Budgeting and financing for long-term responsibility require financial and maintenance context. Review: Review funding horizons, maintenance allocations and environmental criteria in financial decisions. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.21 Adopt organizational philanthropy practices¶
Read guideline 5.21. Primary classes: Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Philanthropy policy | Consulting | A giving policy alone does not establish aligned philanthropic partnerships. Review: Inspect the policy, partner selection and evidence of active partnerships supporting stated objectives. Supporting methods: manual qualitative. Evidence: published records, project records. |
| Voluntary work | Consulting | Support for voluntary work needs evidence of time, resources or completed activity. Review: Review supported programs, allocation and participation records and benefits to the intended organizations. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.22 Plan for a digital product or service's care and end-of-life¶
Read guideline 5.22. Primary classes: Manual qualitative (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| End-of-life care | Manual qualitative | The criterion asks for clear documented end-of-life guidance that can be inspected directly. Review: Review disposal, archive and deletion instructions for scope, responsibilities and practical usability; do not infer that retirement occurred. Supporting methods: consulting. Evidence: published records. |
5.23 Repair, reuse, refurbish, recycle or reduce e-waste¶
Read guideline 5.23. Primary classes: Manual qualitative (1), Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| E-waste management | Consulting | Responsible hardware treatment requires asset disposition and service-provider evidence. Review: Trace retired assets through reuse, repair, redeployment or certified disposal and inspect provider policies. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| E-waste policy | Manual qualitative | Repair and recycling commitments can be assessed from maintained policy documents. Review: Inspect repair/recycling policy scope, responsibilities and support for repair; distinguish policy adequacy from execution. Supporting methods: consulting. Evidence: published records. |
| Refurbishment strategy | Consulting | Prioritizing existing hardware requires evidence of procurement and redeployment decisions. Review: Compare equipment requests with repair/refurbishment options and completed reuse or purchase decisions. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Right to repair | Consulting | Repairability, parts pricing and user capability require product and supply evidence beyond web copy. Review: Inspect repair instructions, parts availability/prices and restrictions; seek practical repair evidence for the product scope. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.24 Define performance, environmental, and human budgets¶
Read guideline 5.24. Primary classes: Manual quantitative (1), Consulting (3).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Environmental budget | Consulting | Lifecycle budget criteria must be actively applied and communicated. Review: Inspect lifecycle limits, affected audiences and decisions or alerts showing those limits influenced delivery. Supporting methods: manual quantitative. Evidence: published records, project records. |
| Performance budget | Consulting | A measured small page or budget file does not establish active enforcement and reduction. Review: Inspect thresholds and CI/monitoring results, then trace exceedances and measured changes to actions. Supporting methods: static analysis, automated, manual quantitative. Evidence: published records, source/configuration, project records. |
| Human budget | Consulting | Sustainable engineering budgets require staffing, workload and worker well-being evidence. Review: Review effort metrics, planning, workload limits and employee feedback rather than inferring sustainability from delivery speed. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Measurable improvements | Manual quantitative | Baseline comparisons and improvement claims can follow an explicit quantitative reconciliation. Review: Check equivalent scopes and methods, recalculate claimed changes and inspect evidence for attribution and uncertainty. Supporting methods: automated, consulting. Evidence: published records, project records. |
5.25 Use and contribute to open source¶
Read guideline 5.25. Primary classes: Consulting (2).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Open source governance | Consulting | Sustainable selection, approval and release practices require governance and applied decisions. Review: Review open-source policies, approvals, releases and maintenance responsibilities for reuse and long-term support. Supporting methods: static analysis, manual qualitative. Evidence: published records, project records. |
| Collaboration and contribution | Consulting | Regular community contributions require evidence across time and forms of support. Review: Review code, time and financial contributions, their cadence and their connection to shared maintenance or reuse. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
5.26 Create a business continuity and disaster recovery plan¶
Read guideline 5.26. Primary classes: Manual qualitative (1), Consulting (1).
| Current criterion | Primary class | Why and how to review |
|---|---|---|
| Plan of action | Consulting | Maintained and tested incident response requires exercise and incident records. Review: Review plans, tests, residual risks and actual incidents, then trace lessons to operational or retirement decisions. Supporting methods: manual quantitative, manual qualitative. Evidence: published records, project records. |
| Audience awareness | Manual qualitative | Clear incident and risk communication can be assessed from public status and notification records. Review: Inspect representative incident communications for affected services, availability/data risks, updates and practical user guidance. Supporting methods: consulting. Evidence: published records. |