How Ranker doubled its viewability rates in 15 months

The improved viewability has also contributed to a 64 percent increase in Ranker’s desktop CPMs and an 87 percent increase in its mobile CPMs over the past 12 months. While several factors — having faster-loading pages, using different exchanges and market conditions outside Ranker’s control — affected the increased ad rates, the improvement in viewability was the largest single factor, Benson said.

In May 2016, Ranker averaged 43 percent viewability on desktop and 31 percent viewability on mobile. Mobile viewability was worse than desktop because Ranker’s slideshows had worse viewability on mobile. Mobile screens aren’t big enough to fully display the slideshow’s image and an ad when a user landed on the page. Unless the user scrolled, mobile slideshows were a drag on viewability.

Among the various steps Ranker took to increase viewability, eliminating slideshows had the biggest impact. Getting rid of the format increased mobile viewability by 11 percentage points and brought it more in line with Ranker’s desktop viewability, Benson said.

According an Integral Ad Science survey, about 70 percent of advertisers transact on viewability. So increasing mobile viewability was imperative for Ranker since about 90 percent of its 25 million unique visitors came from mobile last month, according to comScore. Ranker gets 92 percent of its ad revenue from programmatic.

To further boost viewability, Ranker adopted lazy-loading ads, which only show an ad once it is viewable. This step increased viewability by 7 percentage points, Benson said.

Lazy-loading ads helped improve Ranker’s page-load speeds. Ranker also cut off slow exchanges and restricted the amount of code it would accept on ad creative. These factors together contributed to a 60 percent improvement in how long it takes Ranker to render content on mobile devices, from 1.2 seconds to 0.5 seconds.

While pushing for viewability has driven ad revenue for Ranker, Benson acknowledged the risk of unintended consequences. Viewability mandates have led to more ad clutter and reductions in how long ads are in view.

Ranker’s viewability quest hasn’t led to more ad clutter since it only allows one sticky ad per user session, and the ad units it cut to boost page speed were among its most intrusive. But time spent per session has declined about 10 percent, from 3.9 minutes to 3.5 minutes, since the slideshows it eliminated used to keep users on page longer.

Cutting out slideshows and slow-loading ad units also contributed to a drop in the number ad impressions per session by 36 percent for desktop and 41 percent for mobile. But the increases in CPMs offset this, Benson said.

“We feel the end result is better for user experience, but it wasn’t a clean, easy process,” Benson said.