Affiliate Link is Trumping Homepage - URL parameter handling?

[LawrenceNeal]

An odd and slightly scary thing happened today: we saw an affiliate string version of our homepage ranking number one for our brand, along with the normal full set of site-links.

We have done the following:

1. Added this to our robots.txt :

User-agent: *
Disallow: /*?

2. Reinserted a canonical on the homepage (we had removed this when we implemented hreflang as had read the two interfered with each other. We haven't had canonical for a  long time now without issue. Is this anything to do with the algo update perhaps?!

The third thing we're reviewing I'm slightly confused about: URL Parameter Handling in GWT. As advised - with regard to affiliate strings - to the question: "Does this parameter change page content seen by the user?" We have NO selected, which means they should be crawling one representative URL. But isn't it the case that we don't want them crawling or indexing ANY affiliate URLs? You can specify Googlebot to not crawl any of particular string, but only if you select: "Yes. The parameter changes the page content." Should they know an affiliate URL from the original and not index them? I read a quote from Matt Cutts which suggested this (along with putting a "nofollow" tag in affiliate links just in case)

Any advice in this area would be appreciated.  Thanks.

[gazzerman1]

hmmm.. seems like Google is getting some strong linking signals that this is the popular page to arrive at.

The canonical tag on the homepage is the right way to go.

You could 301 redirect any customer that lands on you with an affiliate code in the url? This would be a very simple bit of code you could even put it in an an include at the top of each page. This way those pages never even exist and you get all the link juice.

One other thing might be to put a noindex on any page that has an affiliate parameter. But you would lose the link juice.

[LawrenceNeal]

Thanks for your reply, Gary.  I'm not entirely sure how our (far reaching and lucrative) affiliate tracking/logging works, but I would have thought 301ing all the links to the original page would sabotage it, no?!

The canonical will certainly work but we've only reinstated it on the homepage as we have 6 other sites that have hreflang alternates in place and the canonical seems to interfere with their function.

[Laurensvda]

I don't know if this has anything to do with the algo update, but at least your not the only one. I saw a competitor ranking with a second version of their homepage. The second version had utm parameters behind them.

Luckily the page with the utm parameters disappeared from the serps this morning. He was actually ranking first with the normal version and second with the version with the url parameters. This was on some pretty competitive keywords and lasted almost three days.

[gazzerman1]

Lets say your site is using php?

Your system no doubt picks up the parameter with a php get and stores it as a session variable.

That is likely all that would need to be done before the page is 301 redirected.

Best thing to do is create a test page with the cod mentioned above on your site and try it

xyz.com/test.php?affif=12345

have the page redirect to the homepage and see if that affiliate code is stored.

[LawrenceNeal]

Luckily it's disappeared today, which leads me to believe it was a Google-side algo error that was swiftly corrected (nothing we have done will have reflected in the serp so quickly, I doubt)

[Everett]

I'm glad to hear you've been sorted out Lawrence Neal. I find it interesting the the other Lawrence saw something similar, and I'll ask around to see if it was a glitch that other people have noticed too.

For anyone reading this wondering what Mr. Neal was referring to in regard to rel canonical / href lang conflict, there's a good writeup of it over at Dejanseo.com and Gianluca Fiorelli mentions it in his comment on Dr. Pete's Rel Canonical uber post here on Moz.