Information Architecture

Use this when you need to evaluate an existing product's content organization (audit mode) or design an IA from scratch (create mode). Produces a scored assessment with restructuring recommendations, or a validated navigation tree.

Process

Step 1: Gather inputs

Ask the user to provide:

1. Mode — audit (assess existing IA) or create (build new IA)
2. Product description — what it is, who uses it, current state
3. User segments — who are the primary audiences? What tasks bring them here?
4. Top user tasks — the 5-10 things users come to do most often (e.g., "find a practice guide," "set up a new tool," "prep for standup")
5. Current navigation — sitemap, sidebar config, or URL structure (for audit mode)
6. Content inventory — list of pages/sections with content types (for audit mode; auto-generate if the agent can access the codebase)
7. Known pain points — anything the team already knows is broken or confusing (optional)

If auditing a codebase the agent can access, offer to auto-generate the content inventory and navigation tree from source files.

Step 2: Content inventory

Build or verify a complete inventory:

## Content Inventory — (Product name, date)

| # | Page/Section | Content Type | Parent | Depth | Status |
|---|-------------|-------------|--------|-------|--------|
| 1 | (Page name) | Guide / Template / Tool / Reference | (Parent section) | (1-5) | Active / Stub / Broken |

### Inventory Summary
- Total pages: (count)
- Content types: (breakdown)
- Max depth: (deepest nesting level)
- Stub/broken pages: (count and list)

Step 3: Navigation path analysis

For each top user task, trace the shortest path to completion:

### Navigation Path Analysis

| Task | Audience | Shortest Path | Clicks | Alternate Paths | Dead Ends |
|------|----------|---------------|--------|-----------------|-----------|
| (User task) | (Segment) | (Section → Subsection → Page) | (count) | (Other ways to get there) | (Paths that fail) |

### Path Issues
- **Duplicate paths:** (Same content reachable from multiple nav locations)
- **Orphan pages:** (Content not reachable from primary navigation)
- **Depth violations:** (Content beyond 3 clicks from entry)

### Screen Reader Navigation Audit
Evaluate how assistive technology users experience the IA:

| Element | Status | Issue |
|---------|--------|-------|
| **Heading hierarchy** -- do h1-h6 levels follow a logical, sequential order without skipping? | Pass / Fail | (Details) |
| **Landmark regions** -- are nav, main, aside, footer used to define page regions? | Pass / Fail | (Details) |
| **Skip navigation** -- can keyboard users bypass repeated navigation to reach main content? | Pass / Fail | (Details) |
| **Navigation labels** -- do nav elements have accessible names (aria-label) distinguishing them? (e.g., "Primary navigation" vs. "Footer navigation") | Pass / Fail | (Details) |
| **Page titles** -- do page titles uniquely identify each page for screen reader users switching tabs? | Pass / Fail | (Details) |
| **Breadcrumbs** -- if present, are they marked up as a nav landmark with aria-label="Breadcrumb"? | Pass / Fail / N/A | (Details) |

IA structure directly determines screen reader experience. A heading hierarchy that skips from h1 to h4 is as broken for screen reader users as a missing nav link is for sighted users.

Step 4: Taxonomy and labeling audit

Evaluate how content is named and grouped:

### Labeling Audit

| Label | Matches User Language? | Ambiguous? | Overlaps With | Recommendation |
|-------|----------------------|-----------|---------------|----------------|
| (Nav label) | Yes / No — users say "(term)" | Yes / No | (Other label) | Keep / Rename / Merge |

### Taxonomy Issues
- **Jargon labels:** (Internal terms users wouldn't search for)
- **Overlapping categories:** (Sections that compete for the same content)
- **Missing categories:** (Content types with no clear home)

Step 5: IA scorecard

Rate the architecture across five dimensions:

## IA Scorecard — (Product name, date)

| Dimension | Rating | Evidence | Impact |
|-----------|--------|----------|--------|
| **Findability** — Can users find content within 3 clicks? | (1-5) | (Summary) | (Who's affected) |
| **Labeling** — Do labels match user language and expectations? | (1-5) | (Summary) | (Who's affected) |
| **Navigation efficiency** — Are paths direct without dead ends? | (1-5) | (Summary) | (Who's affected) |
| **Content coverage** — Does every content type have a clear home? | (1-5) | (Summary) | (Who's affected) |
| **Scalability** — Can the IA absorb new content without restructuring? | (1-5) | (Summary) | (Who's affected) |

**Overall:** (average) / 5

### Scoring Guide
- 5 = Excellent — no issues found
- 4 = Good — minor issues, low impact
- 3 = Adequate — noticeable friction, workarounds exist
- 2 = Poor — users regularly fail tasks or get lost
- 1 = Broken — content is unfindable or architecture prevents core tasks

Step 6: Restructuring recommendations (audit) or navigation tree (create)

For audit mode — propose a restructured navigation:

### Proposed Navigation Tree

(Indented tree showing new structure)

### Migration Plan

| Current Location | New Location | Action | Breaking Changes |
|-----------------|-------------|--------|-----------------|
| (Old path) | (New path) | Move / Merge / Rename / Remove | (Redirects needed, links to update) |

### What Stays the Same
- (Sections that work well — protect these)

For create mode — build the navigation tree from the content types and user tasks, then validate:

### Navigation Tree

(Indented tree with content types mapped to locations)

### Validation Against User Tasks

| Task | Path in Proposed IA | Clicks | Passes? |
|------|-------------------|--------|---------|
| (Task) | (Navigation path) | (count) | Yes / No — (issue) |

Step 7: Review

Ask the user:

• Does the scorecard match your experience? Any ratings feel off?
• Are the top user tasks correct, or are we missing critical ones?
• Does the proposed structure make sense to someone new to the product?
• Any political or organizational constraints on restructuring? (Teams that own sections, naming conventions that are non-negotiable)
• What's the appetite for change — incremental or full restructure?

Related skills

• /ux-audit — broader UX assessment (IA is one dimension)
• /journey-map — maps user tasks that the IA should support
• /persona-create — defines the user segments that navigate differently
• /competitive-analysis — how competitors organize similar content
• /accessibility-audit — IA structure affects screen reader navigation

Portfolio example: B Lab (2020-2021)

Kate audited B Lab's global site navigation and ran a card sort exercise. The existing IA spread content across "About B Corps," "Certification," "B Economy," "For B Corps," and "COVID-19 Resource Center" -- with critical items like the B Corp Directory and SDG Action Manager buried deep. The card sort reorganized content into user-intuitive groups: "Boring Logistics" (Directory, Legal, Governance), "About Us/Our Why," "Certification" (requirements, fees, search), "Portfolio/Experience," and "Action Items/Learn More." A "???" category captured orphaned items (B Movement Builders, B Analytics, B Economy, B Impact Assessment) that didn't fit cleanly -- revealing naming and positioning confusion. Key finding: content was obscured behind unintuitive "B" language that made sense to insiders but not to companies.

Output location

Present the assessment as formatted text in the conversation. For audit mode, the restructuring plan can be filed as a planning inbox entry.