OPTIMIZING INFORMATION ARCHITECTURE (IA) FOR PRODUCT APPLICATION SOLUTION PAGE
OPTIMIZING INFORMATION ARCHITECTURE (IA) FOR PRODUCT APPLICATION SOLUTION PAGE
ABOUT COMPANY:
Waters Corporation is a global B2B eCommerce and SaaS leader in analytical instruments, softwares and services. It serves industries including pharmaceuticals, biotechnology, food science, environmental analysis and clinical research. Waters.com is its primary digital touchpoint and it is a complex, multi-segment website that houses thousands of products, application notes, support resources and educational content.
MY ROLE:
Led and managed end-to-end Tree Testing, Collaborated with stakeholders including Information Architect, Subject Matter Expert, Content Designer, Product Owner, Product Manager, etc, Defined research goals, objectives and success metrics aligned with product and business strategy, Designed the complete tree test study: navigation tree, task scenarios in collaboration with the stakeholders, participant screener, Configured the Treejack study on Optimal Workshop including correct destination mapping and participant messaging, Executed live study, monitored data collection, conducted in-depth quantitative analysis, prepared a final report with high priority IA recommendations and debriefed stakeholders over MS Teams.
PROJECT BACKGROUND:
Waters.com serves a scientifically sophisticated B2B audience who arrive with specific analytical needs, such as finding the right chromatography column for peptide analysis (an analysis to ensure the quality, efficacy and safety of peptide-based therapeutics and to discover novel biomarkers for disease diagnostics) or identifying which system to use for PFAS (Polyfluorinated Alkyl Substances linked to serious health risks) environmental testing. The core challenge: product content was organised by product category (Products -> Chromatography -> Columns), while previous research has shown that many users think in terms of their workflow or application field (Applications -> Biopharma and Pharma/Environmental) so the Information Architect wanted to understand if users would look for a particular product solution in the Applications area of our Website or in the Products area. Through this tree testing, the team was able to diagnose this IA gap with clean and unbiased navigation data. Insights revealed that dual mental models exist among users and the navigation on Waters.com must support them both. Although, for reliable results for tree testing the ideal sample size is 50-100 for statistical difference, this research need came in the middle of the sprint and was a high priority because insights needed to be presented by stakeholders in the business meeting with directors and senior directors next day so the team was looking for early signals and hence we closed it with 10 participants (the United States, the United Kingdom and Canada).
OBJECTIVES:
To evaluate whether users could successfully locate product and application-specific information within the current Waters.com navigation tree.
To measure success rate, directness and efficiency across three analytically distinct product-finding tasks.
To identify navigation failure points (where users go wrong and why) and determine whether users' mental models align more with a product-centric or application-centric navigation paradigm.
To uncover Information Architecture (IA) level improvements needed to support the Product Application Solution Page and related content discovery on Waters.com.
RESEARCH PROCESS:
1 - Stakeholder Alignment and Building Navigation Tree
I first aligned stakeholders, involving the Information Architect, Subject Matter Expert, Content Designer, Product Owner and Product Manager working on the project to understand the strategic intent behind the need for tree testing. The goal was to ensure that I provide high-value early signals, as time was a major constraint in this particular project. I had them talk about the background of the project, covering facts, opinions, guesses, etc. I further determined our target users and defined research goals and objectives and discussed success metrics.
I then mapped Waters.com navigation hierarchy into Treejack in the Optimal Workshop platform, replicating the site's existing structure without simplification. I showed the items in the specified order (not randomised) to accurately reflect the Waters.com navigation experience participants would encounter on the live website.
2 - Task Design (Writing Scenarios in User Language)
I moved forward and wrote 3 tasks in the natural language of target users and deliberately avoiding product category names to prevent navigation cueing and test true mental model alignment.
Text In Green Defines Correct Destinations
3 - Participant Screener, Messaging and Study Launch (Optimal Workshop)
I then set up a pre-study screener capturing role-based demographic data to ensure target audience alignment, representing the core Waters.com user base. Tasks were presented in random order to each participant to eliminate bias and ensure the validity of the results, along with task skipping, to understand where they were leaving the task.
4 - Data Collection, Analysis and Reporting
After launching the study, I waited till evening to collect as many responses as possible. I observed that 16 participants attempted the tasks, 6 abandoned the tasks and 10 completed all tasks. All 10 participants attempted 100% of tasks with 0% skip rate, signaling strong engagement with the task scenarios. I further did task-by-task analysis and prepared an engaging final report in which high-priority IA design recommendations were discussed. The stakeholders were debriefed over a Microsoft Teams call, focusing on how IA improvements could directly affect the strategic roadmap for Waters website.
Note: The images above represent an example of analysis for Task 1, in which it was found that the majority mental model for peptide column discovery is product-centric (Products -> Chromatography -> Columns, 50% direct success). However, 30% took a dual path - Applications -> Biopharma and Pharma and Products -> Chromatography -> Biopharma Columns and Consumables, confirming that dual-path navigation is not just tolerated but actively expected by a significant user segment. The single failure (Events -> Seminars) could be an isolated outlier from misreading the top-level navigation. Similarly, the analysis for Task 2 and Task 3 was carried out.
KEY INSIGHTS:
Dual mental models are real and must both be supported across all three tasks, as participants demonstrated two distinct navigation paradigms: product-first and application-first, meaning the IA must genuinely support both entry points for the end users. Task 3 showed Applications -> Environmental as the most common correct destination, suggesting application-first is actually the primary model for environmental analysis tasks.
Task 3's dramatic drop (70% success, 60% directness, 5/10 score) reveals a structural gap for cross-platform analytical workflows. Waters system-level products sit in entirely separate navigation branches with no application-level bridge that unifies them for environmental or multi-platform workflows like PFAS analysis.
For both Task 1 and Task 2, the most common correct destination was Products -> Chromatography -> Columns. 90% directness on both Task 1 and Task 2 indicates that when participants found the correct destination, they did so without backtracking in the vast majority of cases. The core Chromatography Column discovery journey does not suffer from confusing hierarchies or dead ends and the challenge is only when the task context requires cross-branch navigation.
One participant in Task 2, while navigating, ended at Support -> Waters Knowledge Base for a product-finding task. The qualitative feedback received was that ''following product page links usually winds up getting dumped into knowledge base with no obvious thought that appropriate filters were applied''. This could be a content governance issue, not just an IA problem.
RECOMMENDATIONS:
Implement Bidirectional Cross-Linking Between Products and Applications: Eliminate the hard boundary between the Products and Applications branches. Every Product page could surface its relevant Applications prominently and every Application page should surface relevant product types directly. The Product Application Solution Page should serve as the canonical bridge, meaning accessible from both branches.
Enhance the 'Columns' Node as a Contextual Discovery Hub: As observed, Products -> Chromatography -> Columns was the top destination in both column tasks, so rather than a flat product list, it could function as a contextual hub meaning add application-based filtering (Peptide Analysis, Biopharma, Environmental, etc), surface curated application-column pairings and cross-link to Biopharma Columns and Consumables.
Create a Unified PFAS Analysis and Multi-System Application Hub: Create dedicated Application Solution Pages for major multi-platform workflows (PFAS analysis, food safety, etc) that could aggregate recommended systems across all product categories and be discoverable from Applications -> Environmental and from both Systems nodes under Chromatography and Mass Spectrometry.
Resolve Support/Knowledge Base Leakage for Product Journeys: When surfacing knowledge base or support links from within product or application pages, apply contextual filtering so users arrive at pre-filtered relevant results. Clearly label knowledge base links as "Technical Support." Consider a contextual "Resources" sub-section within product pages aggregating Application Notes, Knowledge Base Articles and How-To Videos without redirecting to Support.
RESEARCH IMPACT:
My research findings provided direct evidence for the team to prioritize cross-linking between Products and Applications, which is a site-wide change with conversion possibilities.
Task 3 PFAS findings directly informed the business case for creating dedicated Application Solution Pages for multi-platform analytical workflows, a new content type for Waters.com.
The dual mental model finding became a foundational reference for the redesign of the top-level navigation architecture and mega-menu structure.
KEY LEARNINGS FOR ME:
Tree testing is a powerful diagnostic quantitative method and demands precision design (Writing tasks in the user's natural language (not product labels) avoided response bias and tested true mental models).
Multiple correct destinations reveal mental model diversity (Configuring multiple correct destinations per task was one of the most valuable methodological decisions. It allowed me to distinguish between navigation style (product-first vs. application-first)).
IA research has the greatest impact when it speaks the language of business outcomes (The research impact became a catalyst for organizational change and not just a design input).
Thank you for your time
To discuss more about this project, you can reach out to me at shoryasaxena96@gmail.com or +91 9784088400