Principles of Product Management: Information Architecture (Sitemap, Card Sorting, Tree Testing)

Information Architecture (Sitemap, Card Sorting, Tree Testing)

Information Architecture (IA): Sitemap, Card Sorting, Tree Testing

What This Is

Information Architecture (IA) is the structural design of shared information environments—how you organize, label, and navigate content so users can find what they need without thinking. Bad IA = frustrated users, high bounce rates, and failed product adoption. Good IA = intuitive flows, lower support costs, and higher conversion.
Example: When Airbnb redesigned its search filters (e.g., "Unique Stays" vs. "Entire Homes"), they used IA techniques to reduce drop-off by 12%—users could now find their ideal stay faster.

Key Terms & Frameworks

• Information Architecture (IA): The art/science of organizing content (labels, categories, navigation) to help users complete tasks efficiently. Think: "The Dewey Decimal System for your product."
• Sitemap: A hierarchical diagram of all pages/screens in a product, showing parent-child relationships. Example: Home → Product → Category → PDP (Product Detail Page).
• Card Sorting: A UX research method where users group labeled cards (representing content/features) into categories they find logical. Two types:
• Open: Users create their own category names (exploratory).
• Closed: Users sort into predefined categories (validation).
• Tree Testing: A usability test where users navigate a text-only version of your sitemap to find items. Measures "findability" without visual design bias.
• Taxonomy: The classification system (e.g., "Men’s → Shoes → Running" vs. "Shoes → Men’s → Running").
• Faceted Navigation: Filtering by multiple attributes (e.g., Amazon’s "Price," "Brand," "Rating"). Formula for complexity: Number of facets × Average options per facet = Cognitive load.
• Hick’s Law: Decision time increases with the number of choices. Formula: T = a + b log₂(n) where T = time, n = number of options, a/b = constants.*
• Miller’s Law: The average person can hold 7±2 items in working memory. Use this to limit menu items or steps.
• Mental Model: How users expect information to be organized (e.g., "Settings" under a gear icon). Mismatch = confusion.
• Wayfinding: How users navigate (e.g., breadcrumbs, search bars, CTAs). Example: Uber’s "Where to?" vs. Lyft’s "Pickup location."
• IA Heuristics (from Rosenfeld & Morville):
• Objects: Treat content as living things with lifecycles (e.g., "A blog post has a publish date, author, tags").
• Choices: Fewer options = faster decisions (see Hick’s Law).
• Disclosure: Show only what’s needed at each step (e.g., progressive disclosure in forms).
• Exemplars: Use clear examples (e.g., "Popular searches: iPhone 15, AirPods").
• Front Doors: Assume users enter anywhere (e.g., SEO-optimized landing pages).
• Multiple Classification: Let users find content in >1 way (e.g., search + browse).
• Focused Navigation: Navigation should answer: Where am I? What can I do here? Where can I go next?

Step-by-Step Process Flow

1. Define Goals & Scope

• Action: Align with business/user goals. Example: "Reduce support tickets about ‘Where is my order?’ by 30%."
• Output: A IA brief with:
• Target users (e.g., "First-time shoppers on mobile").
• Key tasks (e.g., "Find a gift under $50").
• Success metrics (e.g., "Task completion rate >80%").

2. Audit Existing IA (If Redesigning)

• Action: Map current sitemap (tools: Miro, Whimsical, or even Excel). Label:
• Orphans: Pages with no clear parent.
• Dead ends: Pages with no exit (e.g., no "Back to results").
• Duplicates: Same content in multiple places.
• Output: A gap analysis (e.g., "Users can’t find ‘Returns’ because it’s buried under ‘Account’").

3. Conduct Card Sorting (Open or Closed)

• Action:
• Recruit 5–8 users per segment (e.g., new vs. power users).
• Create 30–50 cards (e.g., "Track Order," "Gift Cards," "Size Guide").
• Run remotely (tools: OptimalSort, Miro) or in-person.
• Analyze with dendrograms (tree diagrams) or similarity matrices (shows how often items were grouped together).
• Output: Top 3–5 category names (e.g., "Account" vs. "Profile" vs. "Settings").

4. Build & Validate Sitemap

• Action:
• Draft a hierarchical sitemap (tools: Figma, Lucidchart).
• Tree test it (tools: Treejack, UserZoom):
  • Write 5–10 tasks (e.g., "Where would you find your order history?").
  • Recruit 20+ users (remote unmoderated works).
  • Measure:
  • Success rate (% who found the item).
  • Directness (% who took the shortest path).
  • Time on task.
• Iterate based on failure points (e.g., "Only 40% found ‘Returns’ under ‘Account’").
• Output: A validated sitemap with >70% success rate on key tasks.

5. Design Navigation & Wayfinding

• Action:
• Apply IA heuristics (e.g., "Disclosure" = hide advanced filters under "More").
• Test faceted navigation (e.g., "Filter by price, color, brand").
• Add wayfinding aids:
  • Breadcrumbs (e.g., "Home > Women > Shoes").
  • Search with autocomplete (e.g., "Showing results for ‘wireless earbuds’").
  • CTAs (e.g., "Shop Now" vs. "Learn More").
• Output: High-fidelity wireframes with labeled navigation.

6. Launch & Monitor

• Action:
• A/B test IA changes (e.g., "Menu A: 5 top-level items" vs. "Menu B: 7 items").
• Track behavioral metrics:
  • Findability: % of users who use search vs. browse.
  • Engagement: Time on page, clicks per session.
  • Conversion: % who complete key tasks (e.g., "Add to cart").
• Iterate based on rage clicks (users clicking repeatedly in frustration) or heatmaps.
• Output: Data-driven IA improvements (e.g., "Move ‘Returns’ to top nav → 25% fewer support tickets").

Common Mistakes

PM Interview / Practical Insights

1. "How would you improve the IA of [X product]?"

• Trap: Jumping to solutions (e.g., "Add a search bar!") without research.
• Answer:
• Start with goals (e.g., "Reduce bounce rate on mobile").
• Audit current IA (e.g., "Users can’t find ‘Live Chat’").
• Run card sorting to uncover mental models.
• Tree test proposed changes.
• A/B test the new IA (e.g., "Top nav vs. hamburger menu").
• Pro Tip: Mention trade-offs (e.g., "More top-level items = faster access but higher cognitive load").

2. "How do you measure IA success?"

• Trap: Only citing "user satisfaction" (too vague).
• Answer: Track behavioral + attitudinal metrics:
• Behavioral:
  • Task completion rate (e.g., "% who found ‘Returns’").
  • Time on task (e.g., "Avg. 12s to find ‘Contact Us’").
  • Search vs. browse ratio (e.g., "60% use search → IA may be failing").
• Attitudinal:
  • SUS (System Usability Scale) score.
  • NPS or CSAT for "ease of finding things."
• Pro Tip: Use session recordings to spot rage clicks or dead ends.

3. "When would you use open vs. closed card sorting?"

• Trap: Saying "Always use open" (not always practical).
• Answer:
• Open card sorting: Early-stage, exploratory (e.g., "We don’t know how users categorize ‘Sustainability’").
• Closed card sorting: Validating a proposed IA (e.g., "Does ‘Gift Cards’ belong under ‘Account’ or ‘Shop’?").
• Pro Tip: Combine both: Start with open to generate categories, then closed to validate.

4. "How do you handle stakeholders who want to add more menu items?"

• Trap: Saying "No" without data.
• Answer:
• Show the trade-off: "Adding ‘Affiliate Program’ to top nav may reduce clicks on ‘Deals’ by 15% (Hick’s Law)."
• Propose alternatives:
  • Progressive disclosure (e.g., "More" dropdown).
  • Faceted navigation (e.g., filter by "Affiliate" in search).
• A/B test the change (e.g., "Menu A: 5 items" vs. "Menu B: 6 items").
• Pro Tip: Use RICE scoring to prioritize IA changes (e.g., "Impact: High (reduces support tickets), Effort: Medium (requires dev work)").

Quick Check Questions

1. Your e-commerce team wants to add a "Deals" tab to the top nav, but tree testing shows users already struggle to find "Sale" items. How do you decide?

Answer: Run an A/B test comparing: - Option A: Add "Deals" to top nav (risk: more clutter).
- Option B: Merge "Sale" and "Deals" into one tab (risk: less discoverability).
Explanation: Data > opinions—test which option improves task completion rate for finding discounts.

2. A stakeholder insists on using the term "My Wallet" instead of "Payments" in the app nav. How do you validate which is better?

Answer: Run a closed card sort with both labels and measure: - Success rate (% who place "My Wallet" vs. "Payments" in the expected category).
- Time on task (which label is faster to recognize?).
Explanation: Card sorting reveals users’ mental models—don’t assume your label is intuitive.

3. Your tree test shows 80% of users find "Settings" under "Account," but 20% look under "Help." What do you do?

Answer: Implement multiple classification: - Keep "Settings" under "Account" (primary path).
- Add a link to "Settings" in "Help" (secondary path).
Explanation: IA should accommodate different mental models—don’t force users into one path.

Last-Minute Cram Sheet

1. IA = organizing content so users find it without thinking. ⚠️ Not just "where buttons go."
2. Card sorting: Open (exploratory) vs. closed (validation). Recruit 5–8 users per segment.
3. Tree testing: Measures "findability" in a text-only sitemap. Aim for >70% success rate.
4. Sitemap: Hierarchical diagram of all pages. Label orphans, dead ends, duplicates.
5. Hick’s Law: More choices = slower decisions. Formula: T = a + b log₂(n).
6. Miller’s Law: Limit menu items to 5–7 (7±2 rule).
7. Faceted navigation: Filter by multiple attributes (e.g., Amazon’s "Price," "Brand").
8. Wayfinding aids: Breadcrumbs, search, CTAs. Example: "You are here: Home > Women > Shoes."
9. IA heuristics: Objects, choices, disclosure, exemplars, front doors, multiple classification, focused navigation.
10. Measure IA success: Task completion rate, time on task, search vs. browse ratio, SUS score. ⚠️ Not just "user satisfaction."