Recently, the search community appears to have lost their collective mind over the question: Did Google changing the color of the tiny label denoting ads as paid links from yellow to green dramatically increase their ad click-through rate beginning in April?
Logically, the answer to that question is no, given that Google didn’t even roll green ad labels out to all users until mid-June and were only testing the feature for a limited number of searches until that point.
Even ignoring that fact, our data just doesn’t back up any clear increase in CTR as a result of green ad labels, and there are several reasons why analysis to the contrary might be flawed.
No Change in Text Ad CTR on Google.com vs Search Partners
One analysis on the subject that’s been widely shared is a Wordstream post featuring the following graph, which shows Google.com CTR relative to search partner CTR rising beginning in early April.
The argument goes that this relative rise is the result of Google rolling out green ad markers to google.com ads beginning in early April while search partners continued to use yellow markers.
Looking into Merkle’s own data, split out by device (since this is a crucial bit of segmentation for any trend analysis these days), we see no steady movement in how much greater Google.com CTR is relative to search partners since the beginning of March.
While there are obviously some pretty significant daily swings in this metric, neither desktop nor phones saw even a 3% difference in relative CTR comparing the last day featured in the graph to the first.
I don’t want to say that anyone’s analysis that goes against my own is outright wrong, but the fact that Wordstream’s data appears to aggregate data across devices makes me think they might be getting deceived by the numbers they’re using.
CTR Didn’t Change, but Desktop Partner Share Did
Taking a look at the share of text ad traffic coming from search partners, we did begin seeing a rise for desktop computers beginning in early April – conspicuously around the time Wordstream reports an increase in google.com CTR relative to search partners.
Could this rise in share be to blame for the increase in overall google.com CTR relative to search partners observed by Wordstream, as partner traffic from a new source manipulated the relative CTR of google.com?
Perhaps, but even looking at overall relative CTR across all devices we see only a brief rise and then return to past relative CTR for Merkle advertisers.
It’s possible Wordstream’s traffic mix by device is different than that of Merkle, which could impact this overall CTR look.
However, we feel pretty confident in saying any changes in relative CTR were not driven by a different colored ad button, and that different data sets are not to blame in our reaching a different conclusion than Wordstream.
Either way, this debate brings up another point that’s important to remember in analyzing performance.
The Search Partner Network is Volatile
Comparing google.com performance to search partner performance can be a worthy exercise in determining if you want to target your ads to partner sites. However, using the search partner network as a ‘control’ group is not a good idea.
That’s because specific sites frequently fall in and out of Google’s partner network. For example, eBay started using Bing Ads for mobile searches in 2014, causing the share of Bing Ads traffic coming from mobile devices to increase from 25% in Q1 of that year to 36% in Q3. eBay’s mobile ads are now right back with Google, along with a ton of impressions, but fairly low click volume for Google overall.
Other recent changes to the search partner network include Google and Yahoo rekindling their ad serving relationship in late 2015, AOL moving to Bing Ads as a partner, and Google beginning to show PLAs in image search, with those searches deemed part of the partner network.
Long story short, search partner performance, and particularly CTR, can wax and wane with the simple addition or subtraction of a single partner, and is for that reason not a very good bellwether.
Conclusion
Our data does not point to a rise in Google.com CTR relative to search partners since early April, around the time Google began testing green ad labels.
Even if it did, it’s pretty unlikely that such a rise would have had anything to do with the change in label color, as that change wasn’t rolled out to all users until mid-June -- a time when Wordstream's analysis actually shows google.com CTR decreasing slightly compared to search partners.
Are searchers 15% more likely to click an ad that has a few pixels that are green as opposed to a few pixels that are yellow, per Wordstream’s analysis? No way.
