Two Ideas For Improving 'Most Popular' Posts Lists

In an earlier post, I decried the dangers of “most popular” lists, which so many publishers use on their sites as a way of promoting clicks to content. These lists, whether curated editorially or algorithmically, have a significant impact on readership levels of individual posts. I know this for a fact -- whenever one of my posts makes it to one of the “top” lists, views, shares and recommendations all soar.

Although the details vary, the general approach to determining what content should be promoted goes something like this: “article X got clicked more times than article Y, so let’s feature it at the top of the list.” In my earlier rant, I pointed out that this leads to a sort of winner-take-all effect, in which a few articles get tons of coverage, while the vast majority of articles get very little traffic, suppressing discovery and promoting run-of-site content.



I recently realized that there is an even more pernicious problem associated with this sort of favoritism: once certain articles make it to the “most popular” widget, the single most important factor in driving clicks is the title. And with the proliferation of teaser titles (“Seven Reasons Why...” “You Won’t Believe What X Said...” “the Dirty Little Secret of...”), what publishers are unwittingly doing is promoting catchy titles, which may or may not reflect how interesting the content may be.

The success of companies like Outbrain and Taboola, coupled with the notorious increase in the salaciousness of the content they end up promoting, points to an evolutionary regression toward the least common denominators of smut and drivel. But while smut and drivel may make great click-bait, they won't necessarily increase engagement and loyalty.

There are two ideas I would like to share with publishers looking to create more lasting value from their content: (1) diversify; (2) eschew hill-climbing.

Diversify the metrics you use to decide what articles to promote, beyond the click. Once readers land on a page, see how far down the page they go. See how much time they spend on the page. See how many times they click on ads. See how many times they click on more content. See how often they post a comment. See how often they share the page on a social network. Use combinations of these metrics to create “top lists” with specific business objectives.

The kind of hill climbing to which I refer does not involve boots and blisters. Rather, it refers to the typical approach of increasing the rank
of an item that is doing well. But first let’s use climbing as a metaphor: Imagine you're in an unfamiliar landscape with hills and valleys. Your goal is to get to a high point in the landscape. It’s foggy, so you can only see a few steps in front of you. With each step, you always look for the next step that increases your elevation the most. Do this, and you will inevitably end up at the top of the nearest hill, no matter how insignificantly small it is.

Promoting articles that get more clicks over a short time span has exactly the same effect. You might have missed another article that, given a chance, might have eclipsed the popularity of all the other articles on your “most popular” list.

While entire doctoral theses have been written on different ways to achieve global optimization, there are some simple ways to reduce the risk of getting stuck on the pitcher’s mound of editorial greatness. For example, reserve a portion of the list for randomly selected articles – akin to letting yourself wonder around to see if there are steeper slopes to be climbed. Or keep a list of 20, 30 popular articles, and rotate them based in part on their rank, with some randomness thrown in for good measure.

By following these simple ideas, you might be able to encourage exploratory behavior that is good for your business as a whole, not just for the inventory it generates.

2 comments about "Two Ideas For Improving 'Most Popular' Posts Lists".
Check to receive email when comments are posted.
  1. Ari Rosenberg from Performance Pricing Holdings, LLC, May 21, 2015 at 7:33 p.m.

    Paolo, your concepts make too much common sense -- please complicate them and repost :) -- nice job as always.

  2. Paolo Gaudiano from Infomous, Inc., May 22, 2015 at 1:53 p.m.

    Thanks Ari, you are too nice! Hopefully my next one will be a bit less clear :-)

Next story loading loading..