Will your website/blog make more money, if you use Google’s Auto Ads?
We conducted a 40 day split testing experiment, comparing Auto Ads to Fixed Ads displayed by our AdSense plugin for WordPress!
The results – Google Auto Ads made 38% LESS revenue!
Here is a video review of this experiment:
What is the hype with Auto Ads?
Google states: “AdSense Auto ads is a powerful new way to place ads on your site. Set it and let Google do the rest”.
If you think about it, its a pretty incredible innovation. You don’t have to spend time and energy to figure out how to monetize your blog or website. Using artificial intelligence Google scans a particular page on your site and figures out which type of ad, what size, as well as where it should be placed to maximize revenue.Google Auto Ads make 38% Less than Fixed Ads served by Adsense Ninja plugin Click To Tweet
In fact, ad placement is the newly added bonus, since Google Adsense was largely automated anyways. Many web masters, and especially ones who are just starting out, are confused about where to place the ads on their pages, in order to make more money. So auto ads essentially takes care of this challenge and allows many more bloggers to sign up with Adsense.
Moreover, the system is supposed to adjust and improve ad placement as it gathers performance data from your pages.
Ultimately, this gives website owners the opportunity to focus a lot more on creating quality content and improving their online services, without having to worry so much about monetization.
Since Google Auto Ads has such incredible potential, we were excited to see whether it can actually earn more money, compared to going through the trouble of inserting and optimizing ads on your own.
After 40 days of testing, we can definitely tell you that if you properly optimize ad placement and ad sizes yourself, you can (and most likely will) make more money, than if you rely on Google’s machine learning AI.
Testing methodology: 50/50 page views split between Auto and Manual ads in WordPress
Before we jump to analyzing the results of our A/B test, it’s important to discuss methodology – how the test was performed.
Also, note that AdSense does not have proper Auto Ads performance reporting. This makes it very difficult to get accurate results. Moreover, WordPress is not an ideal platform for running these types of experiments.
However, with a lot of manual configuration, we managed to have 2 identical WordPress Themes setup to show on 50% of the page views. One with Auto Ads (Variation A), and another with Manual Ad placements (Variation B), set up in the same way as it was, before the test began.
WordPress Theme AB Testing:
To conduct this experiment, we had to use several specialized tools:
Each identical WordPress theme had about 50% of page impressions (views), served by AB Theme Testing at random. This means that each site visitor saw only one ad variation!
We modified the plugin to have 600 seconds “cookie expiration” time, so that each visitor would see only Auto Ads or Manual Ads.
We used a modified Point theme (free version) with most options disabled, and some CSS/HTML modifications to remove unnecessary page elements.
We ran the test for 40 days, and the entire website had about 41,577 pageviews (reported by AdSense).
Ad Placement – Manual vs Auto Ads
AUTO ADS: We placed Auto Ads snippet in the header.php file of Variant A, between
MANUAL ADS: we used AdSense Ninja WordPress plugin to place ads above and below content, as well as in comments – after every 10th comment on each page (note that Auto Ads also places ads in comments, which we took screenshots of).
Additionally, we had ads inside content, about 4-5 paragraphs down, manually placed by
Measuring the results – Auto VS Manual Ads earnings
As I mentioned earlier, it’s fairly easy to measure income from Auto Ads, because of the way Google implements them.
There is (currently) only one way to see the revenue from Auto Ads. Select the Ad Formats report, and then add the Site, which you want to track.
Unfortunately, there is no Custom Tracking Channel available for Auto Ads.
To see results, you need to:
- Got to Performance Reports inside AdSense dashboard.
- At the top, where it says Report Type > Days, click down arrow, and select Ad Formats.
- Where it says Filter, select your Site, where you ran Auto Ads.
Here are the “formats” to identify which ones are attributed to auto ads income:
- In-page (Manual)
- In-article (Auto)
- Anchor/overlay ads (Auto)
- List view Matched content (Auto)
- Auto text & display ads (Auto)
- Vignette ads (Auto)
- Large Expendables (Auto)
Basically, every ad format that is not In-Page is an Auto Ad type. You select which ones you want to run when you initially setup up auto ads on your site.
Results of the experiment
Here is a screenshot that shows the total combined ad revenue for both Auto Ads and Manual Ads. As you can see from the graph, “In-page” format (manually placed ads) earned $260.32, while Google’s Auto Ads made $161.36 during the same period, with equal number of page views.
If we take the total site revenue of $421.68, then we can see that Auto Ads earned only 38.3% of total. Manually placed ads made 61.7% of all revenue for the duration of the experiment.AdSense Ninja plugin outperformed Google Auto Ads by 61.33% Click To Tweet
In short, AdSense Ninja plugin outperformed Auto Ads by 61.33%
(260.32/161.36 = 1.613287)
Here is another screenshot that shows daily breakdown of each ad format earnings:
From this graph you can see that Manual Ads earned more money nearly every day of the experiment, with only 3 exceptions (3 days), when Auto Ads made more.
Now let’s look at performance of Manual ads on this site, to confirm that “in-page” ads are in fact manually inserted ads, and are not somehow mixed into the Auto Ads.
Here is a line graph report for “all inclusive” custom channel that tracked all manual ads during this experiment.
Note that I made a mistake and did not include “Desktop Comment Ads” into “SE – MANUAL” tracking channel until late into the experiment. That’s why custom channel shows $242.55, whereas “in-page” ads shows $260.32 … We are missing about $18 here.
Finally, here are Google Analytics for both Manual and Auto ads theme. Note – I did not setup Analytics until April 2nd. That’s why you see zero traffic on March 30, and April 1. However, the page views for each variation are within 2%
(14,858 for Auto Ads vs 15,178 for Manual Ads) which is negligible, when we consider 61% difference in earnings!
Here is Analytics screenshot for Auto Ads:
And here is Analytics Screenshot for Manual ads:
Once again, page views for both versions were nearly identical in April. Unfortunately, I do not have data for the end of March. However, we assume that since AB Theme Testing plugin did its job just fine for 29 days, March stats are the same.
Analyzing potential losses
I like to play the “what if” game. What if I used Auto Ads (continued to run my experiment) for 12 months straight. How much would I lose? I did some calculations.
If I ran the 50/50 split for 12 months, and assuming traffic would stay the same, I would lose around $903! If I ran ONLY Auto Ads for 12 months, I would lose $1782. These numbers are based on a little spreadsheet that I made.
Conclusion – with Auto Ads you are losing money!
I don’t want to beat the dead horse anymore, but from 40 days of testing, we can clearly see that I lost 38% of revenue ($98.96) during this experiment, and my manual ads earned 61% more.
While I take into consideration that my ad optimization is somewhat better than most people’s who will read this, most web masters / bloggers should be able to get at least 25-30% better results by using fixed size ad units.
Future prospects of Adsense Auto Ads
While so far the success of auto ads has been rather mixed, it is important to remember that this is a rather new Adsense feature. As more bloggers sign up for the service, Google will have more information to further improve the performance of its auto ads.
Its highly possible that in a few years, auto ads will outperform manual ad placement efforts, as well as existing WordPress Plugins for Adsense.
For now, I recommend running tests, similar to the one I described above, to see what would work best for your website.