00
Duration
2 months
Skills
UX Research
UX Design
Collaborator
2 UX Designers
3 Writers
Outcome
Internal Review Platform
Confidentiality Note
This case study has been adapted for portfolio purposes. Product names, customer information, proprietary interfaces, internal metrics, and implementation details have been omitted or modified. The underlying design process and my contribution remain representative of the project.
01
Project Overview
Challenge
Documentation reviews are often manual, inconsistent, and dependent on individual reviewer expertise. As documentation libraries grow, maintaining quality while reducing review effort becomes increasingly difficult.
Proposed Solution
Design a centralized content validation platform that guides writers through a structured review workflow, surfacing actionable recommendations across three core quality dimensions:
Searchability – Can users easily find this content?
Contextuality – Does the content provide sufficient context to complete a task?
Conciseness – Is the information clear, focused, and free of unnecessary complexity?
02
Background
Unclear terminology
Missing contextual information
Lengthy explanations
Poor searchability identified late
The Problem
Maintain consistent quality across hundreds of documentation pages.
Identify discoverability issues before publication.
Reduce the time spent on repetitive editorial reviews.
Detect unnecessarily verbose or difficult-to-scan content.
03
Secondary Research

Findability & Information Scent
Existing research on technical documentation shows that finding the right information is itself a significant part of the user experience. Gao, Gao, and Yu found relationships between users' perceived findability, task complexity, and search behaviour when studying technical documentation in help centres.
Information-foraging research provides a useful explanation for this behaviour: users rely on cues such as labels, headings, titles, and surrounding context to decide whether they are moving closer to their goal. Strong information scent makes the next step easier to predict; weak or ambiguous labels increase uncertainty and can lead users down unproductive paths.
Contextuality & Relevance
Documentation is information used within a particular task, not information consumed in isolation. Research into information quality distinguishes between information that is intrinsically accurate and information that is relevant and appropriate within the user's context.
Strimling's research on documentation quality similarly identifies relevance, accessibility, ease of understanding, and other reader-oriented dimensions as important components of documentation quality. His work argues for feedback that is both meaningful to readers and actionable for writers.
Research on developer documentation reinforces this relationship: usability problems can stem from inadequate structure, insufficient explanation, and poor contextual organization not simply incorrect information.


Conciseness & Comprehension
Research on technical communication cautions against treating readability as a simple numerical score. Readability formulas can capture surface characteristics of text but may not adequately represent how different readers actually understand technical information. User testing and comprehension therefore remain important complements to automated measures.
At the same time, research on documentation quality identifies conciseness and ease of understanding as important reader-oriented qualities. Strimling specifically describes concise documentation as information that is brief without becoming incomplete or overwhelming.
Studies of developer documentation have also identified lack of conciseness, poor explanation, weak structure, and legibility as contributors to poor usability and increased mental fatigue.
Design Opportunity
The literature also exposed a gap between evaluating content quality and helping authors improve it.
A useful validation system therefore needed to go beyond a single score or readability metric. Feedback needed to identify the specific issue, connect it to the source content, explain why it matters, and help the writer decide what to change.
Primary Research
Research Methodology
I used semi-structured interviews and workflow analysis to understand how documentation moves from creation to review and publication. The research focused on their existing behaviour

Interview Focus
The interviews explored four areas:
Current Workflow
How do writers create, review, and validate documentation?
Quality evaluation
What makes a page feel complete, useful, and ready for publication?
Feedback & tools
What makes feedback useful, trustworthy, and easy to act upon?
Pain points
Which issues are difficult to identify manually? Where does review become repetitive or time-consuming?
04
Empathy Maps



05
Prototype 1 - Custom GPT
Searchability

Contextuality

Conciseness

What we learnt
The prototype validated that the scoring matrix could translate content-quality criteria into measurable scores, giving writers a clear view of how their documentation performed across the defined dimensions. The GPT prototypes could also identify issues and provide AI-generated suggestions for improvement.
However, we identified an important limitation: because the feedback was delivered primarily as text-based output, it was difficult for users to intuitively understand where an issue occurred within the documentation. This made the experience less visual and less actionable than intended. The prototype also showed that the framework worked particularly well for already-published content, where searchability signals such as search ranking and page metadata were readily available. However, the core goal of the product was to help writers improve content before publication. For unpublished content, these search-based signals were unavailable, highlighting the need to rethink how Searchability and the overall evaluation experience could be assessed earlier in the content creation process.
Prototype 2 - Chrome Plugin
Searchability - Scoring

Conciseness - Scoring

What we learnt
The visual prototype established a clearer direction for the evaluation experience: writers could now see where problems occurred, understand how they affected the score, and identify where to make improvements. However, the Searchability dimension still relied heavily on signals available only for published content, such as search-result position and existing search behaviour.
This highlighted a fundamental gap in the original scoring model. To support the tool’s primary use case, evaluating unpublished documentation before it goes live. Searchability needed to be assessed without relying on live search performance. We therefore redesigned the Searchability matrix around static, content-level indicators that could be evaluated before publication, including user-style titles, synonym and exact-term coverage, common-question coverage, upfront summaries, and scan-friendly headings. The new matrix shifted Searchability from measuring how discoverable a published page already is to evaluating how well the content is structured for future discoverability.
06
From Evaluation Framework to Product Experience

Searchability
Contextuality


Conciseness
07
From Evaluation Framework to Product Experience
see also









