Search and Referral verions of site - permanent change - penalty?
Can I use Optimizely to permanent change of my site?
I am wondering if I can set two versions of site. First for search traffic with a large amount of content and second for referral users with little amount of content. Won't it cause a penalty from Google?
More here: https://community.optimizely.com/t5/Using-Optimizely/How-does-Optimizely-affect-my-SEO/td-p/3116
You can send all your traffic to a winning variation / launch a winning variation and can find more about that here: https://help.optimizely.com/hc/en-us/articles/200040115-Traffic-Allocation-Limit-experiment-traffic-...
As for your second question, robertchan is spot on in that you cannot redirect a user to a different page with the intent to display content other than what was made available to the search engine crawler. Here's more information on that, directly from Google : https://support.google.com/webmasters/answer/2721217?hl=en
I hope that helps!
Technical Support Engineer
Not to be "Negative Ned" on this topic, but you need to be careful in how you read the exact phrasings of the statements by Google and pay attention to the details of what to do to protect your SEO energy.
First key point in the linked article:
"Google has stated that performing an A/B or multivariate test poses no inherent risk to your site’s search rank"
What this means is that you are not penalized for running an experiment.
This does not mean that the experiment itself may not negatively impact your rankings.
for example (totally oversimplifying SEO to illustrate a point): if you have a primarily text based landing page (lots of good key words, etc.) and you are testing an image based approach, Google will start to see a drop in the number of times your page appears with all those juicy key words. Depending on how long you run the experiment, this change will evenually eat into your rankings (or, conversely, help your rankings if you are now adding more key words than before!)
Second key point in that article:
"if you're running a redirect test with multiple URLs, place the rel=canonical link attribute on all of your alternate links"
This is easier said than done... you would need to add HTML such as the following for every alternate page you present to the user:
<link rel="canonical" href="http://example.com/original_URL_Here" />
Depending on the complexity of your experiment, this may or may not be a burden to figure out how to accomplish, but is a step that I have overlooked in each and every redirect experiment I have ever run.
The conversation I would like to see (and perhaps this is addressed in one of the Personalization blogs and I missed it) is what effect personalization has on SEO. If a Google-bot is able to somehow get into one of the personalization buckets and they start seeing an alternate version of the homepage, a curated navigation menu, and a different assortment of products, how is this *not* going to affect SEO?
I presume the answer is "ensure google bots are excluded from personalization", but this is easier said than done.
For example, 24 days ago a bot out of Portland, OR, started hitting our site. It visits the same 5 pages every day at 10:00am. There is nothing unique about the User Agent that lets us know programatically that it is a bot (we know this only from the behavior - 120+ hits to the same pages every day at the same time). Based on the pages that it hits, our alogorith presumes that this "user" is shopping for "Mens" products and for every a non-gender specific page this "user" sees, they receive content tailored to "Mens" shoppers.
In reality, our personalization is very very mild and would not affect SEO at all, but that is only because our personalization implementation is very mild. If we went whole-hog and *really* made our homepage gender biased (e.g., removing all of the female CTAs), I see no reason to think that SEO would not be impacted by the personalization.
Perhaps in the big picture, only 1 spider out of 1000 that this company uses to crawl our site is affected by personalization, and as such, this change is statistically insignificant in both the short and the long run. But, as I mentioned, I have not seen this topic addressed yet and I'm not an SEO expert so cannot speak to the ins and outs of exactly what each search engine uses.
Hey guys, I'm the SEO Manager here at Optimizely. To chime in:
I agree with robertchan-- as long as you are showing Googlebot and search visitors the same version of the page, you should not be penalized by Google for cloaking. Google considers it cloaking only when you show Google one version of a page, and search visitors a different version.
To address nap0leon's concern, you should not see any change in rankings for keywords, because Google will be shown the original keyword-rich version of the page, not the image-heavy version. Even if Google were to crawl the variant, you can stick a canonical or noindex tag on the variant to prevent Google from getting confused by the new page.
As far as personalization as is concerned, the key is to make sure Googlebot sees the original version of the page and not the variations and to ensure that personalization isn't being used for cloaking.
Takeshi Young - SEO Manager, Optimizely