We recently discovered how to radically increase our income from our mobile ads that were served by Google AdSense. As a result, we not only multiplied our mobile income, but a single mobile ad on our web app now also outranks the performance of all of the other banners like Skyscrapers or Superbanners on the desktop version of the same site. In this article I am not only going to reveal the basic trick, but also share the ideas behind it, including the tests we conducted and the steps we took to multiply our income from our mobile ads with simple pay per click optimization.
Update: AdSense and Sticky Ads
As Fedor pointed out in the comments, AdSense is now clearer in its terms of services stating that “fixed” or “sticky” ads are not permitted. I decided to neither remove nor change the content of this post for two reasons.
1. AdSense is still testing their own implementation of sticky ads, called “anchor ads”. Hopefully, every publisher can use this method without coding skills and worrying about the TOS in a few months. Moreover, as a beta tester, I can say: they seem to work great!
2. You might still want to use sticky ads with ads from other ad networks with click based payments. Just ask them if they will allow you to use sticky ads.
Why even bother with mobile ads?
There was a lot of noise about mobile in 2012, but I didn’t have to listen to understand the importance of a mobile web app. I just had to look into the web statistics of our word game site to see the enormous growth in traffic that we achieved from mobile devices. Now, on both our German and English versions, the number of unique visitors from mobile devices is higher than from any other device.
I didn’t really bother with web app advertising until the first significant increase of mobile users before the holiday season in December 2011. I would have thought the increase might have happened AFTER the holidays, but it actually did so BEFORE. The same happened in December 2012, and so there must be a reasonable explanation. I always had Facebook’s more or less failed IPO in mind that analysts also saw as a result of a bad performance of their mobile ads. Anyway, Facebook managed to significantly increase its income from mobile ads, and so I felt that I had to give it a more serious try. It took me until fall 2012 to come up with a solution to include ads depending on the browser width so that I would be technically able to serve mobile ads as well.
Mobile Ads or Mobile Banners are 320 x 50 px and are designed to fit most smartphones. Google AdSense recently included them in their “normal” ad portfolio, and so, technically, there is no reason not to use this kind of ads for your desktop site.
As a website owner, you probably know about responsive design. This means that the size of your site is adapting to the width of each visitor’s browser. When done right, your site just magically fits into every browser and device. Anyway, for a web developer this is quite easy to set up. The hard part is for any publisher to put ads on such a page and optimize these “responsive ads” properly. We use our own sites as sandboxes for testing different optimization methods. Whether you visit the pages with your desktop browser, tablet, or smartphone, you will always see the same content, but with a responsive layout and responsive ads that fit into the available space.
The solution I developed for WordPress is available as an add-on for my Advanced Ads plugin on wpadvancedads.com.
Bad performance of mobile ads
Imagine that you have already placed your mobile ads on your responsive website. There is a high probability that you placed them just where you might have placed your superbanner or another ad – in the header, footer, or directly within your content. This is what I did at first.
I didn’t really create new ad placements for the mobile version. Therefore, in a “normal” page, there were about two or three banners embedded in the content. This was working great from a technical perspective, but it monetized rather poorly. The click-through-rates and costs-per-click were below the numbers from the “desktop” banners. That wasn’t really a motivation to optimize the mobile banners because any slight improvements on the well performing desktop ads seemed to be more important than the mobile ads, until I came up with a great idea in December 2012.
Use sticky ads
If you have a smartphone, you might have already downloaded some free apps. A lot of them are financed by advertisements. I especially know this from a lot of word game apps out there. Where is this ad normally placed? Right, you can find it very often with a fixed placement, like in the portrait mode on the very top or bottom of your smartphone screen. And this is exactly what I did with my mobile ads in our web app. I cleaned up the content from any ads and placed just one ad at a fixed position at the bottom of the screen. I called them sticky ads, but I seem to not be the first one with this idea.
Wait, I already expected this might result in a better performance of my mobile ads, but I didn’t expect what happened next. The sticky ad shot from being the lowest performer to my best ad, outranking all the others in its total income. In the rest of this article, I am going to explain the numbers that I experienced in detail. I can compare them quite well because of a split test that I conducted first.
Split testing our mobile ads
There is a lot you can do for pay per click optimization. I personally like to compare an old and a new version and totally understand it. For this reason, I am running split tests. In case of the mobile ads optimization, I was afraid that people might be irritated by the sticky mobile ad and leave the web app. Therefore, I didn’t just want to switch from ads places in the content to a single sticky ad at the bottom of the mobile screen. I also only implemented sticky mobile ads on our word search pages and not on pages with articles or other static content. This was based on my assumption that the tools are more like an app as well as on useful and free apps it is acceptable for users to have a sticky mobile ad.
When it comes to pay per click optimization, I use Google AdSense for tests, because it is very easy to handle. I already had different ad codes for the two existing banners – one above and one below the search results like from our Scrabble Helper. To be able to track the performance of sticky mobile ads, I created a completely new banner code.
Setting up a real split test
With a self-developed split test class for WordPress, I marked 30% of mobile visitors to see the new sticky banner and 70% still saw the old version. I performed a test on a visitor and not a page view basis, and so the surprise of switching banner placements would not affect the result. I also wanted to see how the test group reacted over time and if the click through rate would decrease. At first, I wanted to run the test for 3 weeks, but I already finished after 2 weeks because the results were really great and I didn’t want to “lose” more money by not switching completely to the new sticky mobile ads.
The result of our split test
The results of the split test were amazing. Due to the AdSense policy I am not allowed to share specific numbers, but the click through rate of the sticky version was eight times higher than the ad in the content. As the click prices were almost the same, the overall RPM raised by 800%.
Here, you can see the ad requests compared to each other. The difference in ad requests is explained by the new sticky mobile ad to only show for 30% of visitors. The previous banners in the content were displayed on the same page view, and so they were almost identical. The difference is explained by the banner above the results that was only being shown when the search was successful. Remember from this pie chart that the new sticky mobile ad is displayed a lot less than the others due to the test settings.
This pie chart shows the distribution of the income from mobile banners on word search pages. Despite the fact that the sticky mobile ad was displayed a lot less, it generates 2/3 of the ad income from mobile devices. Compared to the other ad blocks on the site, the mobile ad outranked a lot of well performing desktop ads as well.
Best performing ad in the long run
You probably understand now, as to why I didn’t wait the whole test period to change all the mobile ads on word search pages to the sticky version. The numbers were so incredibly good that even a medium decrease would have changed the fact that they were better than the previous version. Still, a month after I completed the test and switched banners, the sticky banner is the banner with the highest income on the whole site.
In the second week after completely switching to the sticky mobile ad, the income increased by 445% compared to the week before the first test with this new ad placement.
Even though the click through rate decreased by almost half, the sticky mobile banner has still one of the best click through rates. Visitors lost interest in the new banner position, of course, but they didn’t lose interest in our site. The bounce rate increased and visit duration decreased by just 1%. This might as well be unrelated to our changes with ads and related to the bigger increase of the rate of new visitors.
As another result, we started the test on one of our other sites. Why test again? We tested again because visitors are different and from our experiments we expected the other sites visitors to be more critical about such a change. In the meantime, the results were also positive and we switched to the sticky mobile ads. Still, the numbers did vary in a lot of ways.
Understanding sticky mobile ads
Even without thinking twice about changing my mobile ads to a sticky version, I am still curious about what the reasons for this are and how to improve the results. It seems like a no-brainer that people are interested in something new. Due to less space on mobile screens, people really notice the new ad and since it doesn’t “flash” them away and is quite short, they might even read it because they are really interested in the content.
At first, I also was afraid of a lot of people clicking by accident. Since Google AdSense reduces your ad revenue at the end of a month also for invalid clicks, I expected it to be a lot higher than it actually was. Perhaps well integrated and, laid out mobile ads really make sense to the visitor. I am definitely going to explore this in more detail.
Use sticky banners with caution
For all of those who read this article and who want to get started right away please be warned. I am still cautious about annoying visitors with sticky mobile ads. We only show them on pages that a visitor really wants to see. Therefore, there is no sticky mobile ad on any landing page. They just appear on search result pages after a visitor has actively used one of our tools. Here, we know for sure that he/she is interested in the results and is not being driven away by an ad that is not a popup or overlay.