IBM Guardium Data Protection — Protect

Bob-Assisted D&UX Provisional Review · Security Policies → Policy Builder for Data → New

Estimated first-pass · Requires human validation
⚠️ Provisional assessment. This is a Bob-assisted first-pass D&UX review based on live/demo flow inspection, screenshots, browser evidence, and local IBM Experience Standards repositories. Scores and findings require validation by the designer and/or the formal D&UX review team. For further assistance, contact vinoy.daniel@ibm.com or Slack: @vinoy.daniel.

Review overview

Product IBM Guardium Data Protection
Flow reviewed Protect → Security Policies → Policy Builder for Data → New
Primary user Gate Keeper / Database Administrator
Review date 2026-07-12
Product version IBM Guardium V12.2.2
Standards source IBM Experience Standards (local repo) · D&UX scoring model (scores.mdx)
Review type Bob-assisted provisional first-pass
Screens reviewed 10 (login, dashboard, navigation, policy list, create policy, rules, create rule, criteria, action, cancel)

Universal Experience mapping

Universal Experience Relevance Themes assessed Estimated level
Get started (Primary) Gate Keeper / DBA creates first data security policy. Setup & onboarding. Setup & installation, Onboarding Minimal (Level 1)
Use (Secondary) Ongoing policy and rule configuration for productive data security management. Productive use, Capabilities Minimal–Good (Level 1–2)
Get help (Secondary) In-product support, contextual help, documentation access. Contextualized in-product support Minimal (Level 1)

Overall provisional assessment

Accessibility
Minimal
Level 1 · ~33/100
Estimated first-pass
Visual Design / Carbon
Minimal
Level 1 · ~27/100
Estimated first-pass
Get started
Minimal
Level 1 · ~33/100
Estimated first-pass
Use
Minimal–Good
Level 1–2 · ~50/100
Estimated first-pass
Get help
Minimal
Level 1 · ~33/100
Estimated first-pass
Overall selected flow
Minimal (Level 1)
~35/100 · Grade: C- (Estimated first-pass)
Estimated first-pass · Human validation required

Numeric scores are estimated using the D&UX formula: (points scored / points possible) × 100. Levels follow the IBM Experience Standards 4-level scale. Not official D&UX scores.

Screenshot-by-screenshot observations

IBM Guardium login page showing username and password fields with Login button
E01 · Login
Clean login page. Labels properly associated. IBM Plex font present. Custom blue gradient button — not Carbon Button. No Carbon theme in use.
Post-login dashboard with system notifications panel showing warnings and informational messages
E02 · Post-login dashboard
4 system notifications appear on login. Mixed severity (warning, info) without clear visual priority hierarchy. No clear 'dismiss all' for non-critical notifications. Notification panel blocks main content.
Protect navigation dropdown showing Database Intrusion Detection, Security Policies, and Uncover Threat Vectors menu items
E03 · Protect navigation
Dark flyout navigation. Not a Carbon Shell navigation pattern. Sub-items use custom expand/collapse tree. Navigation accessible via treegrid with keyboard (Ctrl+arrow).
Security Policies list showing 74 policies in a data table with icon-only toolbar and Filter control
E04 · Security Policies list
74 policies listed. Icon-only toolbar (F01). No empty-state guidance (F07). Table uses radio selectors, not checkboxes — single selection only. 'Include templates' toggles template visibility.
Create New Policy form showing Type radio buttons, required Name field, Category combobox, Roles button, Show advanced options link, and Rules section below
E05 · Create New Policy form
Policy type radio buttons: all have value='on' (F02). Required Name field (*). No inline help for policy type selection (F07). 'Show advanced options' hides Log flat, Rules on flat, Selective audit trail, Audit pattern fields.
Create New Policy with Rules section expanded showing empty rule table with columns for Order, Rule type, Rule name, Tags, Criteria, Actions, Continue to next rule
E06 · Rules section
Rules table visible. Custom Expand/Collapse pattern (F06). Empty state: 'No items to display'. Rules toolbar uses same icon-only pattern (F01). Import button for importing from existing policies is a useful productivity feature.
Create New Rule form showing Rule type dropdown set to Access, required Rule name field, Category and Classification inputs, Severity dropdown, Tags control, and collapsed Rule criteria and Rule action sections
E07 · Create New Rule
Rule type dropdown (Access by default). Required: Rule name (*). Category and Classification fields have ? tooltip help. Severity defaults to 'Information'. Rule criteria and Rule action both collapsed. OK disabled until rule action added.
Rule criteria section expanded showing Session level criteria, SQL criteria, and Other criteria groups each with parameter name dropdown, operator, and parameter value field
E08 · Rule criteria
Three criteria groups: Session level, SQL, Other. Each has parameter name → operator → value triplet. Operator/value disabled until parameter selected. Triplets lack programmatic grouping (F08). Visual grouping by section heading only.

Accessibility findings

F03 No validation error message when required field is empty
High WCAG 3.3.1 · 3.3.3

Screen: Create New Rule · Evidence: E08

When the required Rule name field is empty and user clicks OK, the button is silently disabled. No inline error message appears, no field highlighting, no indication of what is needed. Only the greyed-out OK button signals failure. Violates WCAG 3.3.1 — error must be identified and described in text.

Recommendation: Display inline error message on field blur or submit attempt: "Rule name is required." Use Carbon's form field error pattern with error icon and red border.

F04 Page title does not update on navigation
Medium WCAG 2.4.2

Screen: All screens · Evidence: E03, E04

document.title remains "IBM Guardium (cm01)" throughout all navigation. Screen reader users and users relying on browser history cannot determine their current location within the product.

Recommendation: Update document.title dynamically on navigation events: e.g., "Security Policies — IBM Guardium", "Create New Policy — IBM Guardium".

F02 Radio button values are all 'on' — semantic ambiguity
Medium WCAG 4.1.2

Screen: Create New Policy form · Evidence: E05

DOM inspection: all three policy type radio buttons share value="on". Labels correctly associated, but non-unique values break standard radio semantics for AT users who depend on value announcements.

Recommendation: Set unique value attributes: value="data-security", value="session-level", value="advanced-session".

F08 Rule criteria triplets lack programmatic grouping
Medium WCAG 1.3.1

Screen: Create New Rule → Rule criteria · Evidence: E08

Each criteria row (parameter name, operator, value) is a logical triplet but not wrapped in a fieldset/legend or ARIA group role. Screen reader users cannot determine which value field belongs to which parameter name.

Recommendation: Wrap each row in role="group" with aria-label like "Session criteria row 1: parameter, operator, value".

Visual Design / Carbon findings

F05 Legacy DOJO framework — no Carbon React/Web Components in use
High Carbon components: Level 0

Screen: All screens · Evidence: E04, E05, E06

Guardium uses a legacy DOJO framework (dijit_form, idx_form element IDs). Carbon React / Web Components are not in use. Carbon 2x grid, type tokens, color tokens, and component patterns are not implemented. Carbon components and patterns both score Level 0.

Working well: Internal consistency within the DOJO system. IBM Plex font appears present. Blue/neutral color approach is IBM-aligned.

Recommendation: Strategic migration to Carbon React or Web Components. Tactical: align palette and spacing to IBM Design Language within current framework. Requires Carbon visual design reviewer for full assessment.

F01 Icon-only toolbar buttons — no persistent visible text labels
Medium Visual Design: UI Icons

Screen: Security Policies list · Evidence: E04

Primary action toolbar (New, Edit, Copy, Remove, Refresh, Comment, Download) uses icon-only buttons without persistent text. ARIA labels are present in the AT tree. However, visually no text labels appear. Carbon standard requires icons not appear without text except in global header or icon buttons with tooltips.

Recommendation: Add text labels to primary actions (New, Edit) or ensure Carbon-compliant tooltips are consistently attached.

F06 Progressive disclosure uses text links, not Carbon Accordion pattern
Low Carbon patterns

Screen: Create New Policy / Create New Rule · Evidence: E06

Form sections use plain 'Expand'/'Collapse' text links instead of Carbon Accordion. Keyboard accessible, but the pattern does not align with Carbon interaction affordances.

Recommendation: Migrate to Carbon Accordion when re-platforming to Carbon React.

Universal Experience findings

F07 No contextual onboarding or first-use guidance for policy creation
High Get started: Onboarding · Use: Productive use

Screen: Security Policies list / Create New Policy · Evidence: E04, E05

For the Gate Keeper / DBA persona, creating a data security policy is a high-stakes task requiring prior knowledge. The flow offers no guidance on policy types, rule types, or how to leverage templates. No empty-state guidance, no recommended starting path, no inline explanations.

Working well: 74 policies available including compliance templates. Help icon present. Progressive disclosure reduces initial cognitive load.

Recommendation: Add empty-state guidance: "Start with a template to configure your first data security policy." Add tooltip/helper text for policy types and rule types. Design a short guided path for new Gate Keeper users.

Main risks

  1. Accessibility legal risk: WCAG 3.3.1 (F03) and 2.4.2 (F04) are Level AA failures that could expose IBM to procurement compliance issues with government clients who require accessible products.
  2. Carbon compliance debt: The legacy DOJO framework creates significant and growing technical debt relative to IBM's Carbon Design System mandate. Scores L0 on Carbon components and patterns.
  3. New user activation risk: Without onboarding guidance (F07), new Gate Keeper / DBA users face a steep learning curve, increasing time-to-value and risk of policy misconfiguration.
  4. Form submission reliability: Silently disabled OK buttons without error feedback (F03) may cause users to assume they have saved when they have not, or to refresh and lose work.

Recommended next steps

Immediate (sprint-ready)

Near-term (backlog)

Strategic

Items requiring human validation

ItemWhySuggested approach
Formal color contrast measurement Agent-browser cannot run automated accessibility scan IBM Equal Access Checker / axe / Deque WAVE
Keyboard trap testing in nested dialogs (Policy → Rule) Cannot be verified without interactive keyboard testing Manual tester navigating with Tab/Shift+Tab
Screen reader behavior (NVDA, JAWS, VoiceOver) AT announcements cannot be verified by agent-browser Accessibility specialist with AT
Full Carbon visual design review Requires component-level design review by Carbon expert Carbon reviewer auditing each component against measure-success criteria
Help panel content quality Content was not reviewed in depth within this session Content reviewer assessing help panel text quality and completeness
Persona-based usability — Gate Keeper / DBA Usability findings require user research with target persona Usability study with 3–5 Gate Keeper / DBA participants

Agent-browser: strengths and limitations

✓ Strengths

  • Navigated full flow via accessibility tree (treegrid, tablist, ARIA refs)
  • DOM inspection surfaced the radio value="on" semantic defect
  • Screenshot evidence captured efficiently at each key state
  • Verified page title, heading structure, ARIA labels via browser
  • Identified silent validation failure (F03) through interaction testing
  • Read local IBM Experience Standards repositories as source of truth

✗ Limitations

  • Cannot formally measure color contrast ratios
  • Cannot operate an actual screen reader
  • Cannot verify motion/animation timing or behavior
  • Cannot assess subjective usability or task completion times
  • Cannot fully test keyboard traps without manual interaction
  • Demo environment may not reflect production exactly