We just rolled out a new ad system, and nobody noticed. In this line of work, that means I did my job.
Our platform shift from Open Ad Stream (OAS) to DoubleClick For Publishers (DFP) went smoothly. From the business side of things, this resulted in improved content targeting and increased revenue for the company.
From the technical end, it meant improved performance for our readers, less maintenance , and greater reliability. Here’s a quick comparison of technical features:
|Old Ad System
||New Ad System
Our new DFP-based system is a step up from where we were previously. It’s better organized. It employs meta tags from our CMS for content targeting. The new code is also asynchronous, injected into the page by Google Tag Manager, with fewer HTTP requests to boot.
New Targeting System
Example META Tags:
Why implement another tag standard? What about Opengraph Tags?
It was tempting to use many of our existing vendor-specific tags, such as Opengraph. We probably could have, and it’s possible that we might in the future. However, this is internal standard follows our own set of business rules, not the whim of Facebook and Twitter.
Meta tags are processed by the method
this.getTag("mpr-content-topic"), then normalized and pushed to the ad-loading object.
Targeted requests are formatted by the library in a request like this:
Our library comes with a couple of tools for debugging ad slots and requests. First, the global
gptadslots tells us about our targeted request:
We can also take advantage of the Google Publisher Console to gain an understanding load times and other performance metrics. For example, see the Page Request tab:
Ad Units Console:
Perceived Performance Gains
It’s a challenge to precisely measure performance gains of the new system. I can say for sure that perceived load time has been greatly improved by asynchronous loading of ads. Google Publisher Tags come with the added benefit of single request mode to reduce the number of HTTP requests required to render ads. Things feel faster.
Measurable Performance Gains
I have been able to measure improved DOM rendering time after the new system went live. However, I can’t tell if the change is due to ads loading in iframes yet or not. Time will tell. We will be working to improve page load times later this year, using Pingdom Real User Monitoring to measure our progress. Here’s one report:
Overall, the system launched with an improved developer toolset, more options for revenue growth, and for our readers, improved page load times.
For more on the jargon behind advertising platforms, see this Metafilter article, What is the difference between DART, Open Adstream, and AdSense?.
Hat tip to Will Johnston and Will Lager for all their help on migration day.