Designed and launched CNN's commenting feature
PRODUCT DESIGN • COMMENTING
OVERVIEW
I designed and launched CNN's commenting feature, which consistently drives engagement and registration. Its initial release in the Travel section resulted in a 40% uplift in session duration. The two main challenges in designing commenting were:
Defining what commenting would mean for CNN as a publisher; i.e. What is the ideal commenting system? Why does it matter to users? How will users discover commenting? What kind of damage could commenting do to the brand?
Uncovering constraints and designing across multiple teams' work: OpenWeb for the commenting module, Motif (WBD in-house engineering team) for display name moderation, Growth Engineering for identity and account management, and Editorial as commenting would be on articles they produced. I worked closely with these teams, along with my core-partners: Product, Analytics, and Content Design.
Commenting is now available across verticals, to all domestic desktop and mobile web audiences.
ROLE
Design Lead
TEAM
CNN Growth
TIMELINE
H2 2023
SKILLS
Research, Design Thinking, Diagramming, Wireframing, Prototyping
PROBLEM
Users want to feel empowered by their news experience, but do not have a way to react to or discuss the issues that they care about
Allowing users to express opinions and engage with topics that matter to them provides a more engaging experience that deepens their relationship with the CNN brand.
Goals
BUSINESS GOAL
Increase engagement
Measured through session duration, page views, and scroll depth
BUSINESS GOAL
Grow the registered user base
Commenting as a reason to create a CNN account
USER GOAL
Express opinions and react to topics that matter to them
Measured through comment submissions, reactions, and return rates
USER GOAL
Engage with a community of CNN readers and journalists
Commenting to feel included as part of a broader community of news-readers
Risks
There was no comments section on CNN as of 2023, when the Growth team began to explore it. Introducing commenting to a major news publisher comes with stakes. An Editorial Director shared three risks early in the process that shaped our design decisions to follow.
Previous comments sections devolved into hostile, unproductive bickering
Fully staffing a human moderation team was not feasible at scale
Comments on breaking news are particularly vulnerable to controversy and toxicity
Understanding this context was crucial; in the words of my Design Director, good communities are a fractal image of the larger product. A comments section that turned toxic would reflect on CNN’s journalism directly.
This reframed the design challenge for me. This project was not merely about building a place for users to talk, rather building the right conditions for healthy conversation, while making registration feel like a natural entry point.
Partner
Partnering with OpenWeb was an exciting opportunity. OpenWeb is a third-party platform that powers publisher commenting and community features, enabling media companies like CNN to foster on-site reader engagement at scale. Their AI-powered moderation and infrastructure provided a low-friction path for CNN to introduce commenting without building from scratch, freeing the team to focus on designing a healthy, high-value registration experience rather than the operational complexity of content moderation.
WORKFLOW
Defining the ideal commenting system, then simplifying
We took a phased approach to this project:
Defining the ideal: What should a healthy comment system look like for CNN? We worked through heuristics, competitive patterns, and editorial considerations before simplifying to an MVP.
Testing moderation: We ran a moderation test on a percentage of Travel articles to validate OpenWeb’s technology against CNN’s Editorial content.
Launching MVP in Travel: A lighter, less politically charged section gave us a lower-risk environment to observe user behavior, registration conversion, and comment quality.
Expanding iteratively: Based on performance, commenting rolled out to Science, Wellness, Entertainment, Health, Business, etc. as the whole platform was also moving onto our new CMS.
The project kicked off with several discussions around the question: what does a perfect commenting system look like? I worked with my PM and cross-functional partners to define a set of heuristics for healthy commenting on CNN.
User flow
The core experience looks like this at a high level:
The Pareto Principle states the reality of engagement: typically 20% of people interact with comments; 80% don’t. With this in mind, I began to design entry points into commenting that would invite inclined users to engage.
ENTRY POINTS
Discovery: How will users discover commenting?
Comments sections are typically buried at the bottom of an article. Getting users to notice and engage required intentional placement. I designed a set of in-article discovery touchpoints.
A comment icon with count at the top of the article: Glanceable, tells users how active the conversation is
A full-width inline promo mid-article: We tested placing this promo 5 or 10 paragraphs from the bottom to weigh scroll depth against active engagement
An anchor pill at the bottom: Mirrors the top treatment, bringing users directly to the comment section
I hypothesized that showing a high comment count would encourage participation. I worked with my Content Design partner to establish a rule that articles with fewer than 9 comments would suppress the count, to avoid the reverse effect.
Post-launch data validated the end-of-article placement as the strongest entry point. 72% of users entered through the end-of-article CTA, compared to 7% inline and 5% at the top. This made a lot of sense, as users who reached the bottom of CNN articles were already highly engaged, and only needed a simple nudge.
REGISTRATION
Sign up to post: Registration as part of the flow
Users needed to create an account to post a comment. Rather than treating registration as a wall, I designed it as a natural continuation of the commenting flow. Using contextual, motivated language makes clear why users are being asked to provide information.
DISPLAY NAMES
Display names: A tricky design problem
Display name moderation was a technically complex design problem. Users needed to set a public display name that would be reviewed by Motif’s software before going live.
The constraints I uncovered were significant:
Moderation happened server-side and was not always instantaneous
Motif could not notify users when their name was approved or rejected
Specific rejection reasons could not be surfaced to users at launch (i.e. inappropriate language, pretending to be someone else)
Users still needed to be able to comment while their name was under review
The design challenge went from how do I guide users to successfully set a display name? To: how do I guide users to set a name that will be approved, communicate an opaque backend process clearly, and keep them in the flow rather than abandoning?
Screenshots of WIP flow and screens in flow
These approaches trade community identity for moderation simplicity. The auto-assigned display name solution preserves identity while solving the same problem.
My solutions:
Auto-assigned temporary display names: Users with no display name could comment immediately using a system-assigned name (i.e. CNN-user-xyz123), removing the barrier to commenting
Gentle, specific language: Copy that guided users toward approvable names without listing every rule
Clear pending state messaging: Communicating that their name was in review, that they could still comment in the meantime and where to manage it later
Display name management in Settings: Making it easy to update at any time, with states for pending, approved, and rejected
Distinct onboarding flow: Display name creation steps added specifically for users from commenting entry points, not globally
Pre-approving old names: Some users had display names attached to their account from other CNN services, which I advocated to be run through Motif prior to commenting launch
Display name as a foundation, not just a commenting feature: The language, the settings architecture, and the moderation states were all built to be extensible, so that when CNN introduces other community features, the foundation is already there
Early versions of the flow asked users to set a display name before commenting at all; feedback from team reviews made clear this was too much friction. In further iterations, I was able to come up with a solution that used auto-assigned names and optional name creation to meaningfully reduce abandonment risk.
KEY DECISIONS / TRADEOFFS
Posting comments is more desirable from a business standpoint than creating a display name
Early in the flow design, users were required to set a display name before they could post. Needing a public identity to participate in a public conversation seemed logical, but considering priority, it became clear that comments posted under temporary names were more valuable to the business than users who abandoned because name creation felt like too much work.
We shifted to auto-assigning temporary display names and letting users set permanent names in their own time. Removing the barrier and inviting investment later increases engagement, which is a key Growth goal.
Inline discovery placement required restraint
There was pressure early on to make the in-article commenting promo more eye-catching and more colorful. I pushed for restraint, working within the existing design system. CNN article pages are information-dense, and introducing a visually heavy promo risked feeling like an ad unit rather than a native feature.
The data ultimately supported this: users who were already engaged enough to reach the end of the article converted at the highest rate, without needing to be interrupted mid-read.
Designing around what we couldn’t control
Several constraints regarding display names were uncovered with no workarounds. Each of these created potential for confusion or abandonment. Rather than designing as if these constraints didn’t exist, I worked with my Content Design partner to craft messaging that was honest about the process and next steps without over-explaining.
SPOTLIGHT
Spotlight: Engage with CNN staff
One differentiator between commenting on CNN articles natively vs. on CNN articles on social media is the ability to engage with CNN staff. Anderson Cooper's All There Is podcast was one of the early teams to adopt commenting for interacting with listeners. The CNN Staff label and studio profile icon indicate that users are interacting with someone on the editorial team, and in this case, Cooper himself. There are many potential ways audiences can interact with CNN staff through comments, and I'm excited to see how this evolves.
SUCCESS METRICS
Measuring success
Our high-level business objectives were engagement, registration, and revenue. To measure against these, I worked with my Product and Analytics partners to define a layered set of metrics.
Engagement 👂
Comments posted, likes, replies, mentions
Scroll depth and commenting section views
Clicks across discovery touchpoints
Incremental impact 📈
Time on page per article
Monthly sessions per user
Page views per session
Impressions per session
Registration
New accounts created through the commenting flow
Conversion rate of users engaged with commenting prompts
Results
The initial launch in the Travel section drove a 40% uplift in session duration. Post-launch data confirmed that the end-of-article CTA was the highest-converting entry point at 72%, which informed how we prioritized discovery placements in subsequent rollouts.
Commenting expanded from Travel to Science, Wellness, Entertainment, Health, and Business, and is now live across all CNN verticals. It remains an active registration value prop and a continued area of iteration as moderation understanding deepens and user feedback accumulates.
TAKEAWAYS
Challenges and what I learned
1. Sometimes the honest answer is "this is broken"
During a design review nine weeks into the project, my design director looked at the display name flow and asked, at what point do we stand up and say this is a broken experience? I'd been designing around limitations such as moderation opacity, page refresh requirements, and rejection reasons we couldn't surface, and the flow had become one that asked users to do something and then gave them no meaningful feedback. His question was a wake-up call for me to think outside the box and try a different approach, auto-assigning a temporary name, letting users post immediately, and making the permanent name optional.
2. Involve Content Design early and keep them close
Some of the hardest design problems on this project were content design problems. How do you tell a user their display name is under review without making them feel like they did something wrong? How do you communicate a complicated backend process without over-explaining it? I worked closely with my Content Design partner to solve these questions about tone and trust, and her thoughtful questions helped push my design in the most intuitive direction.
3. Cross-functional coordination is a key design skill
This project had more moving parts than any I'd worked on: four external and internal teams, each with their own timelines, constraints, and definitions of done. I learned to treat alignment as part of my design process, not an afterthought. Regular check-ins, a centralized source of truth for decisions and open questions, and proactive communication when constraints changed all became key to my process.
Enter password to view case study
MM © 2026