HOW DISQUS 2012 TURNS "TIME SPENT" INTO "TIME INVESTED" FOR PUBLISHERS

Introduction

The unveiling of Disqus 2012 this past June represented a major milestone for the five-year-old commenting platform.  With a 100% new codebase and totally revamped design, Disqus 2012 is a complete re-imagining of how we think online communities engage - and establishes a scalable, extensible foundation for those communities (and Disqus) to grow and develop.

Here at Disqus, we are extremely focused on delivering excellent user experiences.  While readers and commenters most readily come to mind when thinking about users of Disqus, we never forget that publishers are also one of our most vital user groups.  Based on a lot of great feedback we’ve received from our publisher partners - both free users and VIP customers - we identified a single, universal concern: time.

Whether you’re an individual blogger or a global news enterprise like CNN, all publishers want to spend less time moderating comments and fighting trolls (there are significant operational cost implications as well for a major media publisher).  They would much rather be spending this time connecting with their audience, engaging with the conversations happening in their comment threads, and building a community - after all, these are some of the key motivations for adopting Disqus in the first place.  Moreover, they want their audience to spend more time with their content - whether it is diving deeper into a heated discussion or simply spending more time on a publisher’s site overall and generating more impressions.

About the Disqus Footprint

At recent count, Disqus reaches an audience of 800 million unique viewers per month across 1.57 million websites.  This massive scale affords Disqus the ability to generate accurate interest graphs and powerful network effects - which, in turn, enable key features of the Disqus network through the application of these capabilities.

Disqus 2012 introduced significant changes on many levels, and one of the most fundamental improvements was a clear and focused design decision to leverage the depth and richness of our vast footprint.  We conducted the following study to test the many design decisions for Disqus 2012 that we hoped would result in better user experience and performance for visitors, commenters, publisher partners, and soon, content promoters.  At the end of July 2012, we compiled a list of all the properties that were currently using Disqus 2012.  This set includes a variety of sites (e.g. large publishers, small bloggers, beta testers) across many topical “verticals” (e.g. music, news, entertainment, tech). 

Where to Find the Time

Disqus 2012 was designed to leverage network effects in order to enable the self-policing of online communities.  It is common to hear about technical scalability when discussing SaaS products, but we understand that scalability extends beyond our system infrastructure.  Some major publishers employ or outsource entire teams of moderators to review posted comments and police their communities - this is not a scalable approach.  Instead, we designed Disqus 2012 to mimic real-life and gave participants more levers to influence the social dynamic of their communities. 

• I. Voting and Flagging
  
• II. Moderation Activity
  
• III. User Engagement
  

I. Voting and Flagging

By upvoting, downvoting, flagging, social-sharing, replying - and many other actions - every user is producing signals.  Disqus captures these signals, puts them through our advanced algorithms, and determines (in combination with a number of other inputs) which are the best quality comments and commenters to surface to the top of a discussion thread.  One of the chief aims of Disqus 2012 was to promote the usage of voting and flagging due to their critical role in allowing us to identify the best quality comments. 
Scope: Since Disqus 2012 was officially released in early/mid June of 2012, we wanted to cover a sufficiently wide period of time so that the before-and-after effects of Disqus 2012 would be more readily apparent.  We chose two two-week periods (May 13, 2012 to May 26, 2012 and July 15, 2012 to July 28, 2012) for this part of our study.

• Upvoting increased 15.57%.  Upvoting activity increased significantly compared to “like” activity (the analogy for upvoting in the previous version of Disqus).  This increase is attributed to the improved design of the UI and the more intuitive notion of upvoting and downvoting.
  
• Flagging decreased 79.56%.  This metric may seem a bit puzzling on its own, but is extremely significant when coupled with the fact that downvoting activity has grown to the same order of magnitude as upvoting.  In other words, users previously flagged low-quality comments - flagging is meant to be a method for users to direct truly inappropriate or abusive comments to moderators for review.  With Disqus 2012, users are downvoting low-quality comments instead, sending key signals across the network and minimizing the moderation-load by flagging only those comments that truly need to be flagged.
  
• About 1 in every 4 votes is a downvote.  When we first announced the voting concept for Disqus 2012, we heard concerns that downvoting was another means for abuse and that trolls would “spam-downvote” the quality comments and compromise discussions.  Firstly, a single account can only downvote once per comment - so this limits the influence of any individual user.  However, our analysis shows overall that downvotes are not being used for abuse and that the vast majority of votes are positive.  This implies that users in aggregate are using downvoting judiciously and meaningfully.
  

II. Moderation Activity

One of the key purposes in algorithmically determining the quality of conversations is to reduce moderation needs and enable true manageability - as a community grows, it should be able to police itself given the right tools. The idea here is to free up that time that would have been spent moderating, so that publishers and authors can spend more time focusing on the things that matter most: developing quality content and experiences for their audience.
Scope: These are two key metrics that we regularly monitor to determine the impact of Disqus 2012 on moderator activity. For this analysis, we compare between the official launch of Disqus 2012 and the most recent data available at the time of this report’s writing (August 2012).

• Moderator Actions decreased 25.22%. We observed that the weekly average of moderator actions for sites using Disqus 2012 has decreased dramatically. This is huge, and equips moderators with more time to participate in conversation, or focus on other areas of their site’s social presence.
  
• Moderator Comments increased 7.05%. One of the things we like to see is moderators and authors who are active in their communities. With Disqus cleaning up the spam and lessening the workload that comes with moderating a community, we are observing higher levels of engagement by site “officials” as they more frequently wade into conversations, address questions head-on from their readers, and keep the discussion alive.
  

III. User Engagement

The previous version of Disqus redefined user engagement in the world of comments; Disqus 2012 is designed to extend this engagement beyond the comment thread on a single page.  Disqus 2012 includes a number of features in support of this mission, including: 1) the Community tab which shows the trending, most popular discussions currently underway on a site; 2) the My Disqus tab which shows personalized notifications (e.g. when someone responds to one of your comments); 3) real-time alerts which show new comments and responses that are rolling in, and much more. 
Scope: For this analysis, we looked at the month of July 2012 - the first full calendar month since the official launch of Disqus 2012.  We compare metrics between sites on Disqus 2012 and the rest of the Disqus network (sites not on Disqus 2012).

• Average Visit Duration is 11.54% higher. Compared to those not experiencing Disqus 2012, the users on Disqus 2012 sites spend much more time on-site. By bringing the very best quality comments to the top of the heap, users can continue on past the article into the thought-provoking or challenging views of fellow readers engaging in meaningful discourse.  This increase in time spent/visit between the previous version of Disqus and Disqus 2012 are over and above the increases we see on average when sites switch from another commenting solution to Disqus.
  
• Average Pages per Visit is 27.95% higher. Not only are users spending more time on a publisher site, but they are also consuming more content. The previous version of Disqus already demonstrated a 13% increase in pages/visit compared to white-label, non-networked commenting solutions. The new Disqus 2012 increases this effect substantially further and would have an even more pronounced impact for publishers switching from one of these other solutions. Whether drawn in by the level of discussion happening within a given article or referred over through the Community tab of Disqus 2012, users are finding more interesting content and staying on-site for it.

Summary

Disqus 2012 has only been live for a few months, but is already showing remarkable results. Our research shows that Disqus 2012 is performing as designed and is delivering increased value to publishers by turning the time they once spent on moderation into time invested into their communities. Moreover, users are enjoying the new experience offered by Disqus 2012 as shown by the increased time spent engaging with content and others in the discussion.  It is also important to note that this research focused on a very specific aspect of Disqus 2012; there is much more in the new version of Disqus, including: the Discovery module which recommends organic and promoted content, improved social sharing mechanisms for thread and comment sharing, as well as an upcoming revamp of the My Disqus tab for an improved user experience.