Why Google’s Content Experiments API is Part of the Larger Plan

Posted by | Comments

Tags: ,


Many marketers who want to perform split or multivariate tests have little to no budget for it. That’s the nature of the space - companies are always looking to do more with less. So when Google shut down Google Website Optimizer and launched Content Experiments inside Google Analytics, many in the space were rightly concerned. The relatively full-featured tool was going away, replaced by a competent, but lacking, replacement.

The move drew considerable ire, and left many marketers asking the same questions:

Why had Google watered down their full-featured testing app in favor of a relatively lightweight testing tool?
Why did Google allow it to launch without multivariate tests?
Does Google care about this space at all?

Of course, there were a few benefits to having tests live inside GA. First, the scripts for GA and GWO no longer need to run separately, and there’s greater democratization of data within teams. Second, the integration allows more marketers in the space to get their feet wet on conversion rate optimization.

And now, as Google releases the Content Experiments API, the bigger play becomes even clearer - first, Google is taking out as many of the silos as reasonable, and integrating everything under Analytics and AdWords. Next, it’s building deep experiences and heavy integration on top. Developers will be able to start building on top of Content Experiments, and be able serve rich testing experiences:

1. Server side testing. This will allow for things that traditionally run in the backend to be tested, like search queries.
2. Google math. By default, the tool will not split traffic 50/50. It’s going to use the the “one-armed bandit” problem in balancing between exploiting a version that’s performing well, with exploring lesser-performing options to see whether the performance is just a statistical anomaly.
3. Custom math. In case companies are running tests with very specific reasons for the sampling distribution, the tool allows for customizations.
4. Zero usability impact. Page-level testing is handled by redirects. That’s great for simplicity, but not so great for UX - usability takes a slight hit. With the API, that’s no longer something companies need to balance - the easy-to-set-up tests are available without redirects.

Google already tied Webmaster Tools to GA, and dissolved Website Optimizer in favor of GA. The tool is competing with heatmap tools like Crazy Egg with In-Page Analytics, with dedicated testing tools like Optimizely with Content Experiments, all the while working in the background to release APIs like this.

Make no mistake - Google is looking at the marketer toolset, and changing the game. This is just the next step in the giant’s general direction.