“Every page is a homepage.” As side-door, social, and mobile traffic skyrocket, homepages are no longer the work horses they once were. The load now rests on the article page to keep users engaged and moving through the site – something the rise of mobile and one-hit readership has made essential. How? Better recirculation.
There is little if any research I’ve been able to find about content recirculation. I haven’t found any papers, case studies, or analytics breakdowns. It doesn’t seem like something web based content platforms have thought to or been willing to share. Maybe I just don’t know where to look. At any rate, I hope this changes soon. In the meantime I will break down some thoughts from my own personal experience to make do.
Overall, analyzing the nature, length, and frequency of a websites’ content and value proposition is the best way to create a recirculation strategy. Barring personalization, curatorial recirculation concepts are generic across all web content platforms (latest, related, featured, editor’s picks, most commented, most read, most shared, most popular, trending, etc.). Every model has a slightly different offering, and it’s important to identify which best serve the sites’ content and audience. The challenge lies in choosing which models are contextually effective and what UI form they should take. To shed some light on the nuances, I’ll outline “the big three.” Keep in mind these are all in context of detail pages.
So, what’s new? Content on the web is ever flowing, so this model is a standard for websites that publish lots of content (daily at the very least). It’s most valuable to frequent visitors whose interests span the global offering of the site.
Good example: Bleeding Green Nation
The website is dedicated to an NFL sports team. Users (in this case fans) are highly likely to be interested in all news/articles and eager for new updates. More globally, the sports universe is in industry, which means timely reporting is essential to professionals. This can apply to any publication which covers an industry (WSJ for example).
Not so good example: Listverse
This website creates list articles for entertainment. The number of subjects is massive, so people are likely not interested in all of its content. Fun/interest pieces like this don’t relate to a professional industry so timely updates are less crucial.
Note: As I will cover in related, latest can be very effective on a large site if it stays category/topic specific. Of course in order to be useful, there needs to be sufficient fresh content on a topic by topic basis. At a certain scale, the latest becomes useless because it inherently lacks curation; imagine a latest videos module in a right rail on Youtube – it would be full of garbage.
Also standard for a website that publishes with high frequency, popular is most effective for users who have little time to browse, are looking for a quick hit, and trust the communities’ taste. These headlines tend to be inherently clickable, so it can be a positive by winning over the user who didn’t intend to stay.
Good example: Listverse
I earlier pointed out this sites’ range of topics is massive, and its content is meant for entertainment/time killing, so popular helps users ignore the overload and find the ‘best’ content quickly. The authority of popularity instills trust and lowers the effort needed for decision making.
Note: Most shared, most commented, most emailed, and most read are variations of most popular. They are more specific in their algorithms and clearer in their language. All are effective in creating a sense of community and highlighting active user participation. Though not used as recirc, the most dramatic I’ve seen is “going viral” which displays concurrent users on a given piece of content (see Distractify).
Another variation is “trending.” Popular and most shared are most likely calculated by traffic quantity over a given period (ex. 3 days, depending on frequency of publishing). Trending is meant to indicate that a piece of content is gaining momentum, e.g. its hourly traffic/share rate is higher than the previous hour. Call it “upward trending”. Once this is no longer true, it “downward trends” and may be retired to “popular.” Mashable’s trend graph is a nice example.
One negative to mention: there is something to be said for a community’s ability to influence curation. The Huffington Post perhaps sees itself as an integral, liberal journalism enterprise, however they appear otherwise when the following land in Most Popular when I checked at the time this post was written:
- NSFW PHOTOS: Meet The People Of Ultra
- World’s Dumbest Facebook Post Could Land Woman In Jail
- 16 Alarming Airline Secrets You’ll Wish You Never Heard
- 17 Easy Ways To Prank Your Friends
Tempted? Fifth in line is an article about premium increases looming for Obamacare, but that link is off site. I’m not calling HuffPo a bad example, but as a caveat, if a platform publishes click bait that clashes with perceived brand integrity, this will happen.
A staple of all content detail pages, related works on the assumption that a user is topic focused and will want to consume more of the same. There are many variations of related: “you may also like”, “recommended for you”, “read next”, etc. Sidenote on the variations: a recent analysis showed the words “you” and “your” seem to accompany the most viral headlines. Language that favors the reader seems to pay off. (Oh yeah, don’t forget “NSFW”).
Good example: Tech Crunch
Though the UI/interactions are a bit clunky, the concept of the left-hand tag stack is great. It surfaces related content from all the tags/topics from the story. I think it’s important to indicate why something is related or what it’s related to, e.g. “more on xyz.”
Good example: Refinery29
At the bottom of this article about Katy Perry is an article grid labeled “More celebs and influencers stories”. This works nicely because it’s very clear to the user how the listing below relates to what they just read.
Note: Related can also pull different content types such as “related videos” attached to an article. I believe that these are less effective as content substance is more important to users than the form it takes. However, if these different related content types can be embedded in-line, they can be enriching. See this example on Bustle (scroll to bottom).
Some sites will have multiple business products such as shopping, a directory, events, etc. that they would like to surface and promote through editorial. This is a great way to create awareness of the sites’ offerings and demonstrate value by tying them in. The dissolve surfaces a directory utility and review product. Intothegloss has a neat UI for showing related beauty products mentioned in an article (click the image fixed to the left).
Some news sites have a neat editorial mechanism to package content around topics or themes, sometimes chronologically. This method, in essence, turns a breaking or long tail story into a connected and developing content group.
Good example: SB Nation
Within a package, or “story stream”, SB Nation uses design and presentation to illustrate a narrative. However, the page linked is a listing of that story stream and the in-article recirculation method is not as strong.
Note: I would like to see more of this kind of topic based curation/circulation happen automatically by blending related and latest in to one module. Rather than editors having to create a new package entity, they could add a “grouping tag” i.e. “DeSean Jackson Release” to a breaking story and any subsequent stories/updates after that. Any article with a grouping tag would spit out a module depicted below. It’s a simple workflow for editors with great benefits to users.
Despite the lack of conversation about recirculation models, we’re seeing lots of innovation in this space across all content platforms. The titular quote derived from our work with Quartz, which boldly removed the homepage altogether. Moving forward, we need to start testing, conducting and sharing research on what works and what doesn’t work. Although every situation is different, we could then start to define best practices for a flattening web in transition.
The Weekly Wonk – Cool NYT style top feed with a smooth AJAX pageload.
PCAH – Scroll after articles goes into immersive grid feed. I would advise against the infinite load. These has been overdone and in its worst form becomes mashable. The concept of “endlessness” or “bottomlessness” has never been a comforting idea to humans. It is disorienting, laborsome, and mentally taxing. See my article about decision fatigue and UX.
Time – The now popular infinite scroll with mini-feed. Time does a nice job with pulling headlines out on the righthand side in addition to the mini-feed. I have a suspicion that the concept of “right rail blindness” may be applying to these mini-feed based websites as well, albeit on the left.
Ableton – Note that this has the left and right arrows “next” and “previous”. I do not like these, though they are quite common. The arrows provide no headlines or images to create intrigue or motivation for a user to interact with them. Additionally, the order is not explained to the user. Is it related, reverse chronological?
Business Insider – Great use of intuitive behavior with this one. After a user is finished or bored with an article, their most common behavior is to scroll up to the top (sorry, concrete evidence at the moment). When doing so on this article, a shelf drops down to show a slider of featured, trending, and recent articles.
Medium – An entrancing ajax pageload animation loads the next article in from the bottom, or by hovering over the icon on the bottom left.