Research-driven UX improvements
Structuring for Impact: How Redesigning Knowledge Base Articles Improved Engagement and Reduced Support Load
How data-informed UX writing turned a single overloaded article into a high-performing ecosystem of user-centric guides
Context & focus
One of the most underleveraged levers in UX is the content layer—especially in self-service support environments. At Ozon, we used UX writing as a tool to influence core performance metrics, conducting a structured A/B test to validate the redesign of materials explaining Template Campaigns—a tool that allows sellers to automate promotion settings and save time.
Our goal: demonstrate how editorial structure, clarity, and design directly affect KPIs like:
  • Read-through rate (content completion)
  • Contact rate (support deflection)
  • Customer Satisfaction Index (CSI)
This case breaks down what we changed, how we measured it, and what others can replicate in their own product content ecosystems.
Problem
The original article on Template Campaigns was:
  • Overly long and monolithic
  • Hidden behind login walls (accessible only from the seller dashboard)
  • Dense in information, but poor in structure and visuals
The impact on users was immediate and measurable:
  • 6.86% read-through—most users didn’t finish the article
  • 10.44% contact rate—indicating frequent confusion
  • CSI of just 0.044—low content satisfaction
There was a clear misalignment between the content’s format and the cognitive needs of users.

Challenge
To solve this, we focused on three core goals:
  1. Redesign the content architecture to reduce cognitive load, improve scanning, and make complex processes easier to follow.
  2. Validate changes through A/B testing, comparing the old and new formats across key metrics.
  3. Address high-friction topics, such as campaign setup and management, which had a history of triggering support tickets.
A major constraint: the new version lived on a public platform (seller-edu.ozon.ru) and received lower traffic, requiring careful normalization and interpretation of results.
Outcome: from monolith to modular content
Between November 5 and December 12, 2024, we tested two versions:
Old format—one long article behind a login wall
New format—a modular set of public-facing articles with:
  • Clearer segmentation by task
  • Simplified language and tone
  • Annotated screenshots and infographics
  • Cleaner layout and visual hierarchy
Users were split between the two formats, and we tracked their interaction using Knowledge Base analytics.
Results
1
Read-through rate: +537%
The restructured article “What is Template Campaign” achieved a 43.79% read-through rate, compared to 6.86% for the original. All new articles outperformed the original in engagement.
2
Support contact rate: fewer questions, less cost
Contact rate dropped from 10.44% to 9.06% for the main article—but what matters more is this:
  • Old version: 1,729 support requests (from 16.56k views)
  • New version: 174 support requests (from 1.92k views)
That’s a 10× drop in absolute tickets, showing how even a 1% improvement in percentage can have a major business impact.
Some articles on more complex subtopics still showed higher contact rates:
  • Managing Template Campaign: 41.47%
  • Launching and Setting Up: 19.82%
These now serve as diagnostic points for deeper refinement.
3
Customer satisfaction index: +245%
The new version earned a CSI of 0.152, up from 0.044 — making it the top-rated article in the set.
Key takeaways for UX teams
Structure before style
Don’t try to “write your way out” of a broken structure. Breaking a single monolithic article into goal-specific, task-based micro-articles radically improved engagement and comprehension.

Measure what matters
The real win wasn’t just better percentages—it was hundreds of support requests avoided, reducing strain on support teams and improving user trust.

Small UX changes = big business results
A few structural decisions: headings, hierarchy, language, layout led to measurable uplift across all KPIs.

Content reveals complexity
High contact rates on specific subtopics gave us data-driven insights into what concepts still confuse users, turning content into a diagnostic tool for UX research.

Next steps (and what you can steal)
  1. Expand the modular content approach to other product areas.
  2. Create internal tooling to spot articles with low CSI or high contact rate.
  3. Layer in interactivity—like collapsible sections, decision trees, or short videos—for topics with persistently high complexity.
  4. Systematize testing—A/B testing isn’t just for features. Test tone, visuals, formats, or even article titles.
Final Thought
Content isn’t the polish at the end—it’s part of the core product experience. When you treat it that way, structure it, test it, measure it, and you unlock one of the fastest, cheapest, and most scalable ways to improve UX and business outcomes.

If you're running a knowledge base and still using 2,000-word mega-articles to explain key features, ask yourself: are you supporting users—or overwhelming them?