top of page

Anjanesh Sharma

INFORMATION ARCHITECTURE REDESIGN VIA CARD SORTING

Project Overview

Role

Lead UX Researcher

Timeline

3 week

Methodology
Hybrid Card Sorting, Tree Testing, First-Click Testing

Scope
87 participants across open and closed card sorting studies

Business Context

  • A brokerage platform was preparing to redesign its navigation architecture to accommodate new product offerings across Mutual Funds, Fixed Deposits, Insurance, and Investment Advisory services. The existing navigation had evolved organically over three years, resulting in a fragmented information architecture where similar functions lived in different sections.

  • User complaints centered on difficulty finding features particularly newer offerings. Support ticket volume for Where is feature? queries had increased 34% quarter-over-quarter. The product team needed to redesign the navigation structure before launching additional financial products, but lacked validated user mental models to guide the restructure.

  • The strategic question: How do users mentally organize financial products and trading functions—and does our current architecture match their expectations?

The Challenge

The challenge wasn't just reorganizing navigation—it was doing so for a platform serving three distinct user segments with different mental models:

  • Active Traders (F&O, Equity Intraday): Task-oriented, speed-focused, organized by execution type

  • Investors (Mutual Funds, Long-term Holdings): Goal-oriented, organized by financial product type

  • Beginners (Exploring products): Uncertainty-driven, organized by familiarity and perceived complexity

A navigation structure optimized for one segment could alienate the others. The research needed to surface shared mental models while identifying where segmentation was necessary.

Research Approach & Thought Process

I designed a three-phase mixed-methods study combining generative and evaluative card sorting techniques:

Phase 1: Open Card Sorting (Generative)
Sample: 32 participants (balanced across trader, investor, and beginner segments)
Method: Unmoderated remote card sorting via OptimalSort
Task: Participants organized 45 feature cards into groups they created and named themselves

Rationale: Open card sorting reveals users' natural mental models without researcher bias. By analyzing how participants grouped features and what category labels they chose, I could identify patterns in how different segments conceptualize the platform's offerings.

Phase 2: Closed Card Sorting (Validation)
Sample: 55 participants
Method: Participants sorted the same 45 features into pre-defined categories based on patterns from Phase 1
Task: Validate whether the emergent structure from open sorting held at scale

Rationale: Open card sorting can produce highly variable results. Closed card sorting with a larger sample validates whether the patterns are statistically significant or artifacts of small sample variation.

Phase 3: Tree Testing (Evaluative)
Sample: Same 55 participants from Phase 2
Method: Task-based findability testing using the proposed new architecture
Task: "Where would you go to set up a SIP for mutual funds?" (8 tasks total)

Rationale: Card sorting tells you how users categorize—not whether they can find things in that structure. Tree testing simulates real navigation tasks to validate that the proposed architecture actually improves discoverability.

Key Findings

Finding 1: Two Dominant Mental Models—Execution vs. Product

Analysis of open card sorting revealed two dominant organizational strategies:
Execution-Based Model (64% of Active Traders):
Features grouped by what you do—"Quick Trade," "Monitor Positions," "Analyze Markets," "Manage Funds"
Product-Based Model (73% of Investors):
Features grouped by financial product type—"Stocks," "Mutual Funds," "Insurance," "Fixed Deposits"
Insight: Active Traders think in verbs (actions they perform daily). Investors think in nouns (products they own). The existing architecture tried to blend both, satisfying neither.
One active trader during post-sort interview: "I don't care if it's a mutual fund or a stock—I care if I'm buying, selling, or watching it. That's how my brain works when I'm trading."

Finding 2: "Financial Goals" Resonated Across All Segments

Despite segmentation in primary models, 82% of participants across all segments created or sorted features into a category related to financial goals, planning, or advisory using labels like "Planning," "Goals," "Advisory," "Learn & Grow."

Insight: There was universal demand for a goal-oriented navigation path that didn't currently exist. Users wanted a section that wasn't organized by product or execution, but by life stage and financial objectives retirement planning, wealth building, emergency funds.

Finding 3: "Insurance" and "Advisory" Were Organizationally Orphaned

In open card sorting, Insurance and Advisory services had the highest placement variability they appeared in 7 different top-level categories across participants, with no consensus on where they "belonged."

In closed card sorting validation, when forced to place them in predefined categories, agreement scores were below 40% indicating participants were guessing rather than following a clear mental model.

Insight: These newer product offerings didn't fit users' existing mental models of the platform. Forcing them into existing navigation would create ongoing findability issues.

Finding 4: Tree Testing Revealed "Funds & Payments" Caused Confusion

The proposed architecture initially placed "Add Funds," "Withdraw Funds," and "Payment History" under a top-level category called "Funds & Payments."

Tree testing showed 58% of participants went to "Portfolio" or "Account" when asked "Where would you check your payment history?" not "Funds & Payments."

Insight: The label "Funds" was ambiguous—users interpreted it as "Mutual Funds" (a product) rather than "Add Money" (a transaction). The category needed to be renamed to "Wallet" or "Money" to eliminate ambiguity.

The Strategic Reframe

The most important output of this research was changing the strategic conversation.

OLD FRAME

Reorganize navigation to fit all new products into the existing structure.

NEW FRAME

Design a dual-pathway navigation system that serves execution-based and product-based mental models simultaneously, with a third pathway for goal-oriented discovery.

This reframe had architectural consequences. Instead of choosing one mental model and forcing all users into it, the solution required parallel navigation structures that let users enter based on their primary use case.

Strategic Recommendations

Based on card sorting analysis, tree testing validation, and cross-segment synthesis, I recommended a three-pathway navigation architecture:

01  QUICK ACCESS (for Active Traders)
Execution-based navigation: "Trade," "Watchlist," "Orders," "Portfolio," "Markets"
Rationale: Optimized for speed and muscle memory—traders shouldn't have to navigate hierarchies for daily tasks.

02  PRODUCTS (for Investors)
Product-based navigation: "Stocks," "Mutual Funds," "Fixed Deposits," "Insurance," "Gold"
Rationale: Investors manage by asset class—they want to see all holdings and actions for a specific product in one place.

03  GOALS & PLANNING (for All Segments)
Goal-oriented navigation: "Retirement Planning," "Wealth Building," "Tax Saving," "Advisory"
Rationale: Addresses the universal need for guidance and planning—positions Insurance and Advisory in a contextually appropriate section.

04  PERSISTENT UTILITIES
Non-categorized, always-visible: "Wallet," "Profile," "Reports," "Support"
Rationale: Functions that don't fit execution or product mental models—need to be persistently accessible regardless of entry point.

Impact & Outcomes

Quantitative Validation:

  • Tree testing task success rate improved from 67% (baseline) to 89% (proposed architecture)

  • Time-to-find for Insurance and Advisory features reduced by 41%

  • "Wallet" rename eliminated the 58% misplacement rate observed with "Funds & Payments"

Strategic Outcomes:

  • Three-pathway navigation architecture was adopted and implemented in Q4 2023 redesign

  • Product and design teams shifted from debating "where should Insurance go?" to "which pathway does this serve?"

  • The card sorting taxonomy was documented as the reference mental model for all future feature placement decisions

Organizational Change:

  • Established a validation protocol: any new feature or product must undergo tree testing before navigation placement is finalized

  • Card sorting study was cited as the model for future IA research across the platform

Research Reflection: What Made This Study Work

Sequencing was critical.

Running open card sorting first (generative) before closed card sorting (validation) before tree testing (evaluation) created a coherent argument. Each phase built on the last, moving from "what are users' mental models?" to "do these patterns hold at scale?" to "can users actually find things in this structure?"

Segmentation mattered.

If I had aggregated results across all participants without segmenting by user type, I would have concluded "there is no consensus" which would have been true but useless. By segmenting and discovering that Active Traders had one model and Investors had another, I could design for both.

Tree testing was the validation moment.

Card sorting tells you how people think tree testing tells you if they can act. The "Funds & Payments" label scored well in card sorting (people agreed it belonged as a category) but failed in tree testing (people couldn't find their payment history there). Without tree testing, we would have shipped a categorically logical but behaviorally broken structure.

The discipline of triangulating across three methods—open sorting, closed sorting, tree testing—is what made the final recommendation defensible. No single method would have been enough.

bottom of page