As a publisher, slow loading ads negatively impact the user experience for our visitors. And unfortunately, the type of ads that partners are most eager to run, like video and rich media, are often the slowest and bulkiest ads. How do we know ad speed matters to users? On the PubNation platform, which now manages ad quality for billions of ad impressions per month, ad speed has historically been the top concern reported by users. To quantify the value of the decreased user experience, we gathered data and analyzed the speed of ads versus the ad CPM to figure out if slow ads pay more. The results are clear: our slowest ads often are our lowest paying ads.
As demonstrated by our latest study on site speed, sites without ads load more than 2x faster than sites with ads. As publishers, we believe that if an ad is paying a good CPM to run a large number of network requests and files, it may be worth the reduction in user experience, sometimes. In our minds, the reason we tolerate slow ads is because we think they contribute significantly to revenue. To test this assumption, we conducted this test using PubNation, and collected CPM, network requests, and file size downloaded related to our ads.
When we analyzed the data, we saw virtually zero correlation between metrics related to ad speed and CPM of ads.
Looking deeper into the data, we realized the ads that are loading the largest number of files and resources are often some of the lowest paying ads! In fact, about three quarters of the worst performing ads pay less than a $1.00 CPM. For us, it was a no brainer: we needed to remove the slow ads that paid very little. We started by removing campaigns with ads that pay less than $0.30 CPMs but download more than 200 resources, as illustrated in the red box below:
For each publisher, the specific guidelines will require careful consideration after looking at all relevant data. However, removing the slowest ads that pay little should become an element in managing ad quality and user experience. For these ads, user experience is being damaged significantly for very little revenue lift.
By separating out the data into CPM intervals, the picture becomes clearer - higher paying ads on average load fewer resources and have smaller total file sizes. In fact, by comparing ads that pay more than $4 CPMs vs. ads that pay up to $0.50 CPMs, the higher paying ads load ⅓ fewer network requests while paying at least 8x in revenue. Ouch. Not only are we being paid very little on the bottom of the CPM range, we are also paying higher penalties in user experiences.
Not all partners are created equal in ad quality or ad speed. Certain partners perform better than other partners - they drop fewer pixels, load fewer resources, and download smaller files. By benchmarking partners, campaigns, and creatives, we can isolate and take action on the ads that are negatively impacting user experience without contributing to our bottom line.
Just looking at our data by partner, we noticed that 2 partners perform significantly worse than the others. Partner F downloads 2x as many resources as partner B for less than ¼ of the CPM; partner C downloaded 60% more resources for ⅔ of the price of partner A. Partner C and F potentially are disrupting the user experience on our site with minimal contribution to our bottom line. Going forward, we will be reaching out to these partners to ask them to improve their ad speed performance. We will also consider handicapping these partners’ bids in our header bidding setup to help reduce low-paying ads from slow loading partners. For our site, at a sub $1.00 CPM level, it makes sense to take a slightly lower CPM for a much better performing ad from another partner.
In conclusion, while ads are a necessary part of our business model, we should not have to worry about cutting down on the low-paying slow loading ads for fear of losing revenue. By comparing the quality of ads across partners and platforms, our ad ops team can maximize revenue and minimize the impact of the speed of ads. By simply removing the worst performing ads, we can increase our user experience without paying significant penalties in revenue.
Want to conduct a similar analysis on your website? Join PubNation’s Ad Speed Beta today by signing up at PubNation.com.
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...