AI-Powered System That Transformed Comment Moderation from Chaos to Clarity

Introduction: The Challenge of Moderation at Scale

Online discussions are the heart of digital news platforms, fostering engagement and driving user interaction. However, managing thousands of comments daily—especially during high-traffic news events—can be overwhelming. Commod is a comment management system designed to help comment managers efficiently moderate discussions, approve or reject comments, and ensure high-quality, advertiser-friendly conversations. To design this system effectively, we followed the Double Diamond process, ensuring a structured approach to problem-solving through

The Goal

  • Streamline moderation workflows

  • Reduce inefficiencies in navigating and managing comments

  • Leverage AI assistance to make the process faster and more intuitive

Business Goal

Commod enhances comment moderation efficiency to keep online discussions engaging, brand-safe, and advertiser-friendly. By leveraging AI-powered moderation, the system:

  • Ensures high-volume comment management for active user engagement.

  • Creates a safe, well-moderated space that attracts advertisers.

  • Increases ad revenue by improving content quality and user experience.

    The Result? A seamless balance between user interaction, platform credibility, and business growth.

How the Double Diamond Process Helped Build a Better System

The Double Diamond Process provided a structured, user-centered approach to solving the challenges of large-scale comment moderation. Following this framework ensured that every design decision was backed by research, iteration, and validation.

Discover – Understanding the Problem

Understanding the Pain Points through Research and Stakeholder Interviews

To understand comment managers' real challenges, we conducted 5 in-depth interviews with moderators from different news departments. Here are key discoveries:

  • System inefficiencies slow down the process: Batch loading requires frequent refreshing, and navigating between sections is unintuitive.

  • Lack of filtering tools: No duplicate comment detection, making it time-consuming to reject repetitive spam.

  • Unclear channel labels: When moderating comments, managers don’t always know which news section they are reviewing.

  • High-traffic news events amplify frustrations: The system struggles to handle the volume, increasing stress and reducing productivity.

After gathering the information, we centralized it using various UX research methods to uncover key insights

  • Affinity Mapping helped me identify patterns in user frustrations, while Empathy Mapping allowed me to step into their shoes and understand their emotions.

    Key Findings from Affinity Mapping:

    • Navigation confusion due to unclear channel labels.

    • Batch comment loading slowed productivity.

    • Spam handling was tedious without duplicate filters.

    • AI moderation needed better transparency.

User Journey Mapping: Mapped out how comment managers interact with the system in different scenarios.

Define – Refining the Core Challenges

Once we deeply understood the problem space, we prioritized the most impactful pain points affecting both users and the business.

Pain Points Identified:

  • Batch Loading of Comments → Slows down workflow due to constant page refreshes.

  • Unclear Channel Labels → Leads to confusion and wasted time during moderation.

  • Frequent System Disconnections → Causes lost progress and frustration.

Most impactful pain points for users & businesses:

  • Implementing continuous scrolling instead of batch loading.

  • Fixing channel labeling to display the correct news section.

  • Improving system stability to reduce disconnects.


HMW (How Might We) questions helped turn user frustrations into actionable design challenges. By framing problems as opportunities, we were able to focus on solutions that directly addressed key pain points, such as improving navigation, enhancing AI assistance and more. This approach ensured that every feature added to the system was user-driven, solving real issues with clear, purpose-built solutions.

Develop – Ideation & Prototyping

Using insights from the Define phase, we began generating solutions. we used Crazy 8s brainstorming, mind-mapping, and user flow diagrams to explore different ways to improve the system.
Key Features Introduced:

  • Continuous scrolling for seamless comment loading.

  • AI-powered comment filtering for spam detection and tone analysis.

Performance dashboard for comment managers to track approvals, rejections, and efficiency.


Crazy 8s: Rapid Ideation

Using Crazy 8s to explore quick solutions for AI moderation, smart navigation, and real-time performance tracking. This method helped generate diverse ideas that refined how comments were flagged, filtered, and approved.

Mind Mapping: Structuring Solutions

Mapped out ideas, grouping them by user needs. This clarified feature dependencies and ensured a cohesive system design that prioritized efficiency.

User Flow: Streamlining Moderation

Detailed user flows helped define smooth interactions with the most command tasks the users do on a daily basis.

Wireframing & Sketching

Before moving into high-fidelity designs, we created low-fidelity sketches to validate the user flow. A clear, structured user flow ensured comment managers could quickly and intuitively navigate the system.

Deliver – Measuring Success & Final Design

With the system refined, we focused on measuring success through KPIs and ensuring the final design met both user and business needs.

Key Performance Indicators (KPIs) to Measure Success

To track the system’s impact, we defined KPIs across engagement, usability, and business goals:

User Engagement KPIs

  • Number of Comments Moderated per Hour → Tracks efficiency.

  •  Time Spent Moderating Each Comment → Ensures speed without sacrificing quality.

Usability KPIs

  • Error Frequency → Measures system reliability.

  • System Downtime Rate → Tracks performance stability.

Business Impact KPIs

  • Moderation Completion Rate → Ensures timely comment approval.

  • Advertiser Retention Rate → Proves the value of maintaining a clean comment section.

Applying visual design

Once we understood how the systems work and learned and organized all the data from my research we started working on the visual aesthetics of the main page and the main flow of the system. Below are the finalized designs for the home page, comments by sections, by article, user managing page, user info, and the AI page where the user can see what the AI does and help with uncertainty comments

The Home Page: a central hub for the user to access all of the insights needed

The main page of the system. Here the user can see all the data he needs. He can see data about his performance, the website, and the AI Assistant.

Moderating pending, approved, and rejected comments with full article context

The comment manager can read all the comments that are pending, approved, or rejected for the News section. The user can read the comment and he has a context by having a link to the full article.

Enable the user to scan comments on a single topic with complete focus

The user can decide to read all the comments for a specific article. Here he has more data thanks to the AI Assistant and he has more context by having more information about the article without opening the full article in a different tab.

Managing User Activity and Comment History

In these pages, the comment manager can access the information the website has about the registered users. He can see how active they are and can read all the comments a user has written.

AI-Assisted moderation with smart insights

On this page, the user can read all the comments the AI reviewed and help with comments the AI wasn’t sure about. The AI highlights all the text that seems suspicious and gives a short insight to help the comment manager work faster.

Results: Transforming Moderation with AI & UX Enhancements

How the design impacted users & the business

For Comment Managers:

  • Reduced time spent per comment by eliminating unnecessary page refreshes.

  • Improved navigation clarity with better labeling and filtering.

  • Enabled faster approvals & rejections with AI-assisted moderation.

For the Business:

  • Increased engagement by ensuring more timely comment approvals.

  • Enhanced advertiser confidence with better comment quality control.

  • Boosted moderator productivity, allowing teams to handle more comments in less time.

Final Thoughts & Next Steps

Building Commod was all about simplifying moderation workflows while ensuring scalability for high-traffic events. By leveraging user research, AI integration, and UX best practices, the new system makes comment moderation faster, smarter, and more intuitive.

Next Steps:

  • Gather post-launch user feedback to refine the system further.

  • Continue improving AI accuracy for better comment filtering.

  • Explore mobile optimization to enable moderation on the go.

Conclusion: Designing for Efficiency & Engagement

Commod showcases how UX-driven decision-making can transform a complex system into an intuitive, high-performing tool. Through research, planning, and design iteration, we ensured that the final product not only enhances user efficiency but also aligns with business goals to increase engagement and revenue.