I made a few slight changes to my site layout and made a few more changes with my adsense settings to try and clear up some visual issues I was occasionally having. When done, and with A/B testing active, I monitored the site for performance issues and was stunned by one ad network causing a full 4 second increase on page load times. Without ads the page loads in less than 0.5 seconds, with ads it usually hovers around 1.5 to 2.5 seconds... but this particular network ad pushes it up to 6+ seconds.
I ran speed tests on a handful of pages twice hourly over the span of 3 hours to get a sampling. The ad only appeared twice but the pagespeed and yslow scores on those two pageloads is horrendous, below 70% vs 95 on average.
1 link unit, 1 image/text unit and 1 text only unit per page
- 61 additional javascript calls for this network ad vs adwords ads
- 5 redirect chains of 3 or more steps
- URIs with more characters than a data packet can hold at one time, seriously, 1 ad with thousands of characters and approaching 100 parameters in a uri is a bit much.
- additional requests for logo images, branding
- not using a CDN warnings x 5 per ad
- various css, expires, reduce DNS lookups and make fewer request warnings as well
Compare that to an adwords ad and the difference is night and day. I know it's a network ad because of the domain names in the URI requests. I don't know which ad network because the domain and company name aren't in the ad network list for their name or domain name, and I don't know their name.
Anyway, I did additional A/B testing with different settings, including having ad networks blocked 100% and with all enhanced features off, and that's as good as it gets. I have a choice, increase earnings a little or speed up load times a lot. I chose the later.
Is my account at greater risk of being banned if I'm essentially opting out of 90% of the options available in the dashboard? It's not that I don't want to take Google's advice or participate in all their efforts and features, it's that performance isn't optional for me. I've read of others being banned after major settings changes due to invalid traffic. I also don't want to perform further A/B testing because the testing itself causes pageloads that are technically invalid traffic too.