Rapid testing and design for the Post's commenting community

The new comments experience had just launched. The redesign fixed critical UX issues around legibility and layout. However, early user feedback made it clear: people still struggled to find value, signal quality, and feel represented in the conversation.

My role was to work with the existing team to evolve the post-launch beta into a more expressive, trustworthy system by:

• Introducing a serious quality signal (rank voting)

• Clarifying emotional reactions

• Surfacing featured voices in context

• Reducing friction between reading and replying

• Introducing more intelligent sorting and discovery modes

Below is how we used research, concept testing, and iteration to move from “functional” to discoverable, expressive, and trustworthy.

1. Reactions & Rank Voting

Research: The problems we heard

• The existing “word reactions” were confusing; users treated “Thoughtful” as a like button, leading to over 90% of reactions being “Thoughtful” across the site.

• People wanted a way to say “this is helpful / not helpful” without writing a reply or relying on vague reaction labels.

• Several users felt that news deserves a more serious primary signal than emojis alone.

Concepts: What we tested

• Whether adding a primary up/down “value” vote on each comment would give readers a more straightforward, more serious way to signal quality than emojis alone.

• Re-thinking our word reactions to be secondary and simplified, using clearly labeled emojis so users could quickly understand them—or ignore them when not relevant.

Results

• Users overwhelmingly loved rank voting as an appropriate, low-friction way to engage, especially on serious stories.

• The full range of new, simplified emojis were all used; even users who disliked emojis felt they were now clearly secondary to the primary vote.

• The combined model of rank voting + clarified reactions led to hundreds of thousands of additional user actions, deepening engagement without sacrificing clarity.

2. Unified Reply Flow

Research: The problems we heard

• The previous model used two separate controls (one to show replies, one to write a reply), which confused many users.

• People weren’t sure how to “enter” a thread—some expected replies to open automatically, others assumed clicking would only expand content.

• The extra friction meant fewer users actually read or added replies, especially on mobile and long threads.

Concepts: What we tested

• A single reply button that both reveals existing replies and opens the reply input in one motion.

• A simplified vertical flow where parent comment → replies → reply box appear together, reducing cognitive load and navigation steps.

Results

• The new “one door into the conversation" limited choice allows users to engage more confidently, knowing what would happen.

• The simplified layout scaled better on mobile, making it easier to scan, read, and respond without jumping around.

• Since viewing replies and replying were combined into a single action, users were more likely to leave a comment.

3. VIP Badges & Priority Identity

Research: The problems we heard

• Readers often didn’t realize when authors, Post staff, or experts were participating in the comments.

• Without clear identity cues, the thread felt like “everyone is equal noise”, regardless of expertise or role.

• Users wanted to see author clarifications, expert context, and official sources first when trying to make sense of complex stories.

Concepts: What we tested

• A simplified summary card, generated with AI, for users to understand the sentiment of all comments in context.

• Badges and labels for verified participants in the thread.

• A priority ordering that gently elevates these VIP comments near the top of relevant views (e.g., Featured, Top), while still allowing all comments to remain accessible.

Results

• Users immediately gravitated toward clearly labeled author and expert comments.

• Transparently labeled badges were trusted and appreciated

• Threads that included visible VIP identity reinforced trust in both the story and the conversation.

4. Comment Sorting & Discovery

Research: The problems we heard

• A single default sort could not satisfy all use cases—some users wanted editorial guidance, others wanted community ranking, and power users wanted raw chronology.

• People had trouble finding their own comments and tracking ongoing replies across time.

• Users wanted control over whether they saw the newest or the oldest comments first, especially on long-running discussions.

Concepts: What we tested

• Leveraging existing AI / Algorithmic tools to eliminate "featured comments" above the feed.

A new tabbed framework for discovery:

• Featured – editorially elevated posts from authors, experts, and editors.

• Top – comments ordered by community rank voting + reactions.

• My comments – a personalized view of all your comments and replies.

• All – a complete chronological list of every comment for transparency.

• A consistent per-tab sort control allowing users to toggle between “Newest first” and “Oldest first” in any view.

Results

• Still under production and testing