Video ad demand is hot. So hot, in fact, that standard ad units are being converted into video ad units, giving birth to the widespread phenomenon of in-banner video. The revenue gains can be dramatic as compared to a traditional ad unit. But at what cost? Is the user experience penalty worth it? One of the most difficult questions a publisher must answer is how to make the tradeoff between ad revenue and the user experience. In this article we share how we evaluated in-banner video, and why we ultimately decided to ban it.
WARNING: For anyone who cares about website performance, the video you are about to see will make you cringe. The video is a recent screen capture of our site SpanishDict, taken while an in-banner video ad is playing on the right side of the page. During the video, we attempted to continuously scroll up and down. The pauses you see in the video are not because we stopped scrolling. The stuttering, halting, and jumpiness are a classic case of jank.
According to the website JankFree.org:
“Jank is any stuttering, juddering or just plain halting that users see when a site or app isn't keeping up with the refresh rate. Jank is the result of frames taking too long for a browser to make, and it negatively impacts your users and how they experience your site or app.”
Jank doesn’t just appear on old computers. The computer used to record this video was a top of the line Macbook pro, running the latest version of Chrome, hooked up to a 300 megabits per second Internet connection. The experience only goes downhill from there.
Jank annoys users. We could see in our analytics in PubNation that Ad Speed had become the number one complaint from users on our site:
The comments from users drove home the point. Here’s a taste of some of the feedback we received:
“The site tends to freeze a lot. It actually makes my entire web browser much much slower whenever that happens and it is very frustrating.”
“The only problem I have is with the actual website itself. It constantly glitches or freezes, which gets aggravating. Outside of that it is perfect.”
“Love the website but it is so slow at the moment. Very frustrating.”
“Possibly beyond your control but some of the advertisements really seem to slow performance down, especially on a mobile device (sometimes to the point at which a paper dictionary is quicker).”
“I haven't seen it nearly as much lately, but you guys used to have video advertisements on the site that would automatically play and they couldn't be paused or stopped. It's slowing down the site itself and it is an unholy nightmare. Please, for the love of sweet baby Jesus, stop allowing video advertisements on your site. It has driven me to using other online dictionaries before, even though this one is the best.”
Ouch. It pains us to read these comments. We pride ourselves on delivering a good experience, but with the slowdown caused by the ads, we’ve been coming up short.
The site is fully loaded with no ads in 1.8 seconds:
With ads, it’s fully loaded in 8.8 seconds:
The ads don’t block the critical rendering path of the page, so the content is viewable even as the ads load, but while the ads load, the site feels slow.
Digging deeper, the number of requests made to load the ads can be staggering.
When the site loads with ads, we average 24 requests to load the page:
When the site loads with no ads, we average nearly 200 requests to load the page:
I wish that were the worst of it. Take a look at what happens when in-banner video is on the page:
The videos are insatiable. At 96.4 seconds, the page essentially never fully loads. And the number of content requests, at more than 1250, is shocking. The evidence was piling up about the damage to the user experience caused by in-banner video.
Often unspoken, but essential to the calculus of every publisher, is the visceral gut reaction to seeing an ad. And in this case, it was annoying. Too annoying. The moving image was distracting, sure. That’s true of many animated ads. In the case of these video ads, though, they slowed down the core of the site. Every interaction on the site was penalized. With the video ads on the page, we were just not happy with the product.
It can be hard to make the tradeoff between revenue and user experience. But if the product experience reaches a point where it’s not something you can be proud of, the ad has gone too far. As evidenced by the comments above, our users agreed--and if we didn’t respond, we’d lose them.
We had essentially decided to remove video ads at this point, but we wanted to understand the revenue loss that we’d suffer. While a standard programmatic ad unit may generate a $2.00 CPM, the video ads can easily generate a $4-$6 CPM. That’s 2-3X the revenue potential! Or is it? It’s not realistic to load 3-5 videos on the page. The site would become unusable. And while the demand for video is hot, it’s not so robust that it can fill ads deep in the user session with consistently high paying units. The result is that video ads may only show a high paying ad in 3-5% of the standard inventory. Yet, even at this rate, a 5-10% lift in revenue would be substantial.
But were these videos actually paying a $5+ CPM? When ads are bought and sold through an ad exchange, many are not explicitly tagged as video, and it can be difficult, if not impossible, to find the average CPM for video ads. In this case, we decided to investigate one of the most common video ads that was appearing on our site--the one that the most users had reported through PubNation. We identified the campaign and traced it back to the ad partner that was running it. We found the buyer and were floored by the CPM. $0.97. Take a look:
In this case, the potential for high-paying video CPMs wasn’t the reality. The video units were actually paying far less than the price most of our inventory commands, backfilling after we exhausted the high-paying demand.
We are not opposed to running video ads on our site. In fact, video ads are likely a driver of future CPM increases. They are already central to the user experience on Facebook, Twitter, and Instagram. With the proper user experience, they can produce respectable ads.
But we need standards. We need ads that are lightweight and fast, ads that don’t ruin the web by turning otherwise speedy websites into polluted zones of jank.
We’ve decided to abandon in-banner video before our users abandon us. When the user experience is fixed, we will give them another shot.
As a final note, given our experience with slow ads on SpanishDict--which is now our #1 user experience challenge--we are exploring the development of a product to monitor and improve the speed of ads on publisher websites. If you are interested in learning more about how it works and getting access to an early beta, let us know by filling out this Google form.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form
Six months ago we launched an A/B testing framework for our site, SpanishDict.com, that has allowed us to make ad ops decisions smarter, faster, and with measurable improvements to revenue. Our fir...
In the past two months, we've radically overhauled our programmatic advertising stack on SpanishDict.com, shifting our focus to header bidding advertisers--advertisers that submit the price they wi...