Stardog Documentation System
Restructured Stardog’s documentation around how users actually work replacing product-first organization with clear, task-based pathways.
Role
Project Manager, UX Research Lead
Timeline
10 months
Team
1 PM & Researcher
2 Leads
1 Engineer
4 Designers
Contributions
Research
User interviews
Survey co-design
Usability testing
Card sorting
Analysis
IA audit
Competitive analysis
SWOT analysis
Affinity mapping
Design
IA restructuring prototype
Search functionality prototype
00 THE PROBLEM
Documentation that scaled with the org, not the user
Internal employees, business stakeholders, and external clients all struggled to navigate the same docs.
As Stardog grew, its docs grew with it content added ad hoc, organized by internal product structure rather than user intent. What accumulated is thorough but assumes the reader already knows where things live. New users are left to navigate a system built around the company's logic, not their own.
01
Slows learning, content assumes users already know the system
02
Increases support load , half of customers email staff before checking docs
03
Blocks adoption, business users hit walls early, limiting Stardog's reach
01 THE SOLUTION
We rebuilt Stardog's docs from the ground up, replacing product-first architecture with role based entry points, task driven navigation, and a search experience that actually understands what users are looking for.
Homepage
01
Role selector on arrival, Business, Engineer, Data, New user so every persona has a clear starting point
02
Release notes moved to the homepage, previously buried, now the first thing returning users see
03
Quick links surfaced as one-click pills, replacing the wall of nav links with no visual hierarchy
Navigation
01
Nav restructured around user flow, Install, Configure, Query, Deploy not internal product architecture
02
Right nav expanded by default with bolded headings no more manually opening sections to orient
03
Shows only the categories under your current heading no clutter, no overload.
Glossary
01
Glossary lifted from buried support content to a top-level nav item accessible from anywhere in the docs
02
Inline term tooltips hover any defined term for a short definition without leaving the page
03
Every term backlinks to the pages where it appears making the glossary double as an index
Search
01
Popup search overlays the page so users scan results without losing their current context
02
Faceted filters by content type, product, task, and role mapped to how users frame what they need
03
Standardized result cards with breadcrumb, title, summary, and tags replacing inconsistent results
02 THE RESEARCH
How did we get here
Internal employees, business stakeholders, and external clients all struggled to navigate the same docs.
03 USER TESTING
Navigation
What worked
1
"Getting Started" and "Support" were navigated confidently across all four sessions.
2
The right nav redesign landed. Expanded by default, bolded headings, cleaner hierarchy.
What didn't worked
1
Out of 3 tests everyone failed 1 when it came to finding content that was nested in the navigation. Participants were still struggling to predict what lived inside a category before clicking.
2
The overall structure still resembled the previous navigation. The accordion was long, nesting ran deep, and participants felt it took too many steps to reach their destination.
3
Managing two simultaneous navigation systems felt tiring rather than efficient across multiple sessions.
What Changed
What Changed
1
Restructured navigation to reflect the order in which Stardog is actually used, with top-level categories organized around how Stardog is installed and used rather than internal product architecture.
2
Selecting a top-level category surfaces only that category's contents in a right-hand panel keeping the experience focused and reducing the depth of nesting users had to navigate through. Visual hierarchy within the panel was clearly marked with headings and indentation.
What didn't worked
3
Retained the right-side in-page navigation to help users move through longer pages with abundant content by allowing them to jump to headings.
4
Added a legend to help users identify different content markers across the documentation.
Search
What worked in Design A
1
Breadcrumbs were immediately noticed and praised. Participants said they surfaced higher-level topic context entirely absent from the current docs, making results easier to evaluate before clicking in.
2
The page preview panel was the decisive advantage. Participants could confirm relevance without leaving search context directly addressing the frustration with the current site's title-only results.
What worked in Design B
1
Descriptive tags on cards were positively received. Participants wanted content-type and topic labels carried forward into Design A's layout not abandoned along with the card format.
What didn't work in both
1
Content filter missed by every participant without moderator prompting. It was described as "disjointed" users never reached it naturally.
2
Role filter was consistently misunderstood. Participants couldn't model what it changed about results and worried it would restrict rather than refine especially for cross-functional users who span multiple roles.
What didn't work in Design B
3
Card-based results were harder to scan than expected. Each card showed too little content to judge relevance, forcing more clicks to evaluate results, slowing down goal-directed users who needed to scan quickly.
What Changed
What changed
1
Combined the best of both designs. Design A's list and page-preview layout kept as the foundation, with Design B's descriptive tags added to each result card.
2
Popular searches added as data-driven chips in the search popup, giving users fast access to common queries without needing to type.
What changed
3
Filters redesigned around the constraints of Stardog's keyword-based search API. Rather than role-based filtering, a faceted filter panel surfaces granular options across content type, product, task, and technology giving users precise control over what they're searching within.
4
Descriptive tags on each result card mirror the filter categories, making content type visible at a glance and reinforcing what the filters do.
4
Breadcrumbs on each result now reflect subheading-level indexing so search surfaces specific sections within a page, and clicking a result jumps directly to that portion of the content.
Homepage
What worked
1
Visual overhaul universally praised color, typography, and layout improved perceived trust immediately, before users even began navigating.
What didn't worked
1
The persona/role selector was invisible to all three participants none discovered it without prompting. Weak labeling and nav-bar placement hid the page's key feature
2
The cards were appreciated in concept but the specific tasks shown did not always match what users actually do
What Changed
What changed
1
Relabeled and surfaced the persona selector directly on the page with clearer language ("I am a…" / "View as…").
2
Replaced the role based tags to tags indicating the type of content on the page.
What changed
3
Replaced "Start Your Journey" with task-based links mapped to the most common user actions in plain language. Audited card content per persona.
4
Add "release notes" terminology to the "What's New in Stardog" section so users searching for that phrase can find it
Glossary
Internal employees, business stakeholders, and external clients all struggled to navigate the same docs.
What worked
1
NA
What didn't worked
1
Visual styling caused confusion code, glossary, and link colors weren't clearly differentiated from each other.
2
Glossary terms weren't obviously interactive hover interaction was missed by most, leaving terms like SNARL and IRI unexplained mid-task.
2
More context on the words was needed in the glossary.
What Changed
What changed
1
Replaced hover-to-define with click-to-expand for glossary terms more accessible and universally discoverable.
2
Add a legend/key for highlighted text on content pages so users immediately understand what each colour/annotation type means
What changed
3
Added a drop down to show other pages using the same glossary term.
Created an end to end design system
Future steps after hand off
We provided Stardog with a few recommendation on how to implement some of the changes we proposed that can only be implemented at an organization level.
Extend search with AI
Layer Stardog's existing Voicebox feature into search to enablenatural-language queries. Users explore the docsconversationally, surfacing content keyword search alonecannot reach.
Refine the tag legend
Once the tag system is populated, refine the legend categoriesbased on the content types actually present. Keeps the taggingvocabulary meaningful and prevents sprawl.
Jobs-to-be-done as carousel
If four featured pages per card feels constraining, convert Jobsto Be Done into a carousel. Expands pages surfaced per userjourney without cluttering the homepage layout.
Standardize image sizing
Images across the docs vary in size, disrupting visual rhythm.Standardize dimensions, or constrain to a set of approvedaspect ratios, for a more consistent professional feel.
Intended Impact
As this project is in the process of deployment here are the metrics for its intended impact.
Support & Efficiency
Projected to reduce documentation-related support tickets by 20–35% by fixing the findability failure that caused users to email staff before checking docs
Estimated to decrease time-to-answer for new users by ~40% by replacing product-first navigation with task-based pathways
Designed to reduce steps needed to reach key content by 50%+, based on usability testing that showed the old accordion nav required too many clicks to reach a destination
Onboarding & Adoption
Role-based entry points designed to reduce new user drop-off in the first session for all 3 user types (Business, Engineer, Data)
Expected to improve time-to-first-successful-task by 2–3x over the baseline where users couldn't orient without knowing Stardog's internal product logic
Targets a 100% reduction in the need for prior product knowledge just to navigate the docs
Search & Discoverability
Faceted search filters (content type, product, task, role) designed to reduce failed search sessions by an estimated 30–40%
Search redesign intended to expand usability from power users only to all 3 user segments, including those who don't yet know exact terminology
Retention & Self-Service
Designed to increase self-service success rate by an estimated 25–40%, reducing dependency on direct staff support
Intended to recover users who previously bypassed docs entirely estimated at a meaningful portion of the 59 survey respondents who flagged search and navigation as broken
Check out more of my work
2025 • B2C • Recruitment Tool
Hack4Impact Recruitment Portal
Transformed Hack4Impact’s recruitment system into a single, transparent platform that supports applicants, reviewers, and directors at scale.