Did Analytics change the way to handle Google images searches on Dec 12?

[AABMarketing_Frank]

Dear all,

One of the sites I'm monitoring receives a lot of traffic from image searches or images that appear in universal search results.

On Dec 12th, 2015, the bounce rate for these sessions went from around 30% the day before to around 87%. See screen shot below.

Did anybody notice similar bounces in the bounce rate? Did Google change something in the way that image search is handled?

Looking forward to your ideas!

large?v=mpbl-1&px=999

[Dr-Pete]

Interesting. At this point, you have a lot more data on the subject than I do. I know the changes to how Google cached/displayed images caused a lot of headaches for enteprise SEOs in the US, but I don't know much about that situation with Google.fr. Your explanation seems plausible, given the data.

If your explanation is true, I'm not sure what you can do about it. These referrals are just inaccurate, and that bounce rate is meaningless. As you said, the design basically makes the bounce inevitable.

I would be reluctant to remove it completely, because you might want to be able to track any changes Google makes to how this is handled, but I would certainly remove it from your overall metrics somehow. You and/or your team shouldn't be judged negatively on this bounce rate.

[AABMarketing_Frank]

I spend some time observing GA's Real Time reports and here is what I found.

I first noticed that also on Dec 12th, the source "images.google / organic" makes its appearance in GA.

If you use google.com (or probably all other versions of Google that use the same interface), GA doesn't log a visit till one clicks on either the enlarged image or the "Visit page" button. (screenshot with French flag attached, does this interface have a specific name?) The visit is logged as "google / organic", not "images.google / organic".

But if you use google.fr (or probably all other versions of Google that use the same interface, I confirmed with google.de), GA logs a visit even if you haven't really left Google, when the image is shown hovering above its host page. The source is "images.google.fr / referral" at this point.

But when you then click on the cross to close the image or on the "Site Web pour cette image" link in the side bar, hence if you really go to the site hosting the image, the source information is replaced by "images.google / organic". (screenshot with map of France attached)

So it seems quite logical that
a) the bounce rate for the source "images.google.fr / referral" is close to 100% and that
b) the source "images.google / organic" appeares at the same time

This raises three questions for me
1. How was the behaviour before Dec 12th?
2. Wouldn't it be appropriate now to exclude entirely trafic provided by the source "images.google.fr / referral" (as well as images.google.de etc.), as this is only an enlargement in Google's search results and not a visit of the site?
3. How is it possible that the bounce rate of "images.google.fr / referral" is not 100%? Why do certain sessions still get multiple page views?

I also looked into what you had suggested, if only certain images or phrases had the high bounce rates.

The answer is that for the traffic logged as referral traffic, there is no keyword data. But if I look at the landing pages, the general rule is a bounce rate between 75 and 95% for "images.google.fr / referral". The landing pages with lower bounce rates, simply have very few sessions, so that's probably just a coincidence. I noticed though that there are quite a few pages that are redirects for images that don't exist anymore that have a low bounce rate or even bounce rate 0%

How do you think, we should deal with these new settings now?

2016-01-25_images%2Bafter%2Bclicke%2Bon%2Bimage.jpg 2016-01-26_google-fr%2Bafter%2Bclick%2Bon%2Bimage%2Bin%2Bwidget%2Bfrance.jpg

[Dr-Pete]

I'm sorry - I misread the bounce rate part. So, image search is definitely driving clicks, but your bounce rates in GA skyrocketed - gotta. Unfortunately, I'm not aware of anything on the Google side that would blow up bounce rates once someone got to your site.

Have you checked out what that actual experience flow looks like right now? First thing I'd do is try a couple of your most popular image searches and make sure nothing obvious is acting up.

Do the bounces seem clustered around any particular images or phrases, or are they across everything?

[AABMarketing_Frank]

Hello Dr. Pete,

Thank you for stepping in!

Did you see similar changes in bounce rates at the time?

Also, I am a bit confused that you speak of "image impressions" while I referred to bounce rates. My understanding was that if Analytics loggs a session with for example referring site / images.google.fr as source, it means that the user is actually visiting the site, in other words that the user has clicked on the "Consulter la page / Go to the page" button. Am I wrong?

Just to avoid misunderstandings: I am not referring to the number of sessions with "referring site / images.google.fr" as source, but to what Analytics shows as bounce rates. The number of sessions even has increased for my particular site.

Thanks for your help!

[Dr-Pete]

Here in the US, Google changed image search a few months back and started caching everything, which killed image impressions overnight. I thought that roll-out was international, but I'm not experienced enough with vertical search to know for sure. Did that potentially just hit France?

[Martijn_Scheijbeler]

No sorry, I don't have a French one to check.

[AABMarketing_Frank]

Thanks Martijn,

Did you check on a single site or on multiple site.

Is this maybe a francophone thing? In a Francophone forum, I found at least one other webmaster who reported the same observation
http://forum.webrankinfo.com/google-images-hausse-soudaine-taux-rebond-t185121.html

Do you have a .fr site that you could check?

Best wishes

Frank

[Martijn_Scheijbeler]

Hi Frank,

No we don't see it in our dataset, I've checked around 30k sessions and the bounce rate definitely isn't seeing changes like what you're seeing.