To "Rel canon" or not to "Rel canon" that is the question

[ReneReinholdt]

Looking for some input on a SEO situation that I'm struggling with. I guess you could say it's a usability vs Google situation. The situation is as follows:

On a specific shop (lets say it's selling t-shirts). The products are sorted as follows each t-shit have a master and x number of variants (a color).

we have a product listing in this listing all the different colors (variants) are shown. When you click one of the t-shirts (eg: blue) you get redirected to the product master, where some code on the page tells the master that it should change the color selectors to the blue color. This information the page gets from a query string in the URL.

Now I could let Google index each URL for each color, and sort it out that way. except for the fact that the text doesn't change at all. Only thing that changes is the product image and that is changed with ajax in such a way that Google, most likely, won't notice that fact. ergo producing "duplicate content" problems.

Ok! So I could sort this problem with a "rel canon" but then we are in a situation where the only thing that tells Google that we are talking about a blue t-shirt is the link to the master from the product listing.

We end up in a situation where the master is the only one getting indexed, not a problem except for when people come from google directly to the product, I have no way of telling what color the costumer is looking for and hence won't know what image to serve her.

Now I could tell my client that they have to write a unique text for each varient but with 100 of thousands of variant combinations this is not realistic ir a real good solution.

I kinda need a new idea, any input idea or brain wave would be very welcome.

[Keszi]

Hi Rene,

The first impression after reading your question is that I meat the same situation as on faceted navigation.

Still this is something different. My advice would be to put the rel=canonical on and get rid of duplicate content. This way you will have one default image, then the visitor can choose what they need (and you just reload the image).

Writing all the hundreds of thousands of unique texts wouldn't be the solution I believe. Still you can use some parameters in the "facets" such as a #color so if people would like to share this content with their friends they can distribute a visitor friendly URL. That would be my choice.

I hope that helped,

Istvan

[ReneReinholdt]

Hi there and thanks for your input. But what you mention is exactly what I already have (maybe I just explain it badly), I was kinda looking for a different amazingly brilliant solutions that I hadn't thought of myself

But your thoughts and time is very much appreciated. If you have any other ideas do let me know

[Keszi]

Ah, sorry. Miss-understood the question then.

[ReneReinholdt]

or maybe my explanation is just crappy

[AnthonyMangia]

This is exactly the kind of situation that rel="canonical" exists for.  Product color is one of those classic examples SEOs bring up when explaining canonicalization.  Don't trust Google to figure things out on their own - make it clear to them that these pages are related and should be treated as such.

[ReneReinholdt]

Hi and thx for your reply.

I agree with you, as I tried to explain in my post. But this doesn't really help me with the users from Google not getting served with the correct picture. Possible leading to a high bounce rate. Plus I have the added problem that Google will see the master as less relevant for the colors as keywords. Since the keyword won't be in the page title, h1,h2,ex.. so all in all the page will have a very low relevance for the key-phrase  "blue t-shirt".

Hence I'm looking for a different solution

[Dr-Pete]

I'm afraid there's no perfect solution. The canonical tag probably is the best bet here - the risk of letting thousands of near-duplicates into the index is much greater than the cost of not landing people on specific colors.

Keep in mind that, once Google removes the color variants, only the "master" product page will appear in search. So, users won't really come into the site with a color intent (except in their heads). Whether that's good or bad for usability isn't clear. On the one hand, it would be nice to rank for every color and have users with a color in mind land on that specific product. On the other hand, some users don't have a color in mind (they know what they like when they see it), and landing on the main product pages shows them all available options. It really depends on your customers, but there are pros and cons, in terms of usability and conversion.

There's no magic Option #3, though - I'm 99% confident saying that. The risks of indexing all color variants post-Panda are relatively high, and I think you'll gain more from consolidating than you'll lose by leaving them all.

[ReneReinholdt]

I completely agree on every point (as I tried to explain above) and I could not myself come up with a better solution, but thought I might give you guys a chance before jumping the rel-canon band wagon

To be honest I didn't expect any amazing ideas but one could hope that I hadn't thought about everything, unfortunately it seems I had.

thx for your time everyone

[Dr-Pete]

Unfortunately, there are still a lot of gaps in how Google handles even the typical e-commerce site. Even issues like search pagination are incredibly complicated on large sites, and Google's answers are inconsistent at best. The only thing I'd say for sure is that I no longer believe the "let us handle it" advice. I've seen it go wrong too many times. I've become a big believer in controlling your own indexation.