Welcome to the Q&A Forum

I can't debate one thing - we certainly don't have all the information, and that can lead to bad advice at times.

I disagree on a couple of points:

(1) User-friendly URLs can have both usability and SEO advantages, whether or not they're meant to be typed in directly. Typically, those advantages are minor, but descriptive URLs can certainly boost SEO a small degree.

(2) If your URLs have spaces in them, they are probably being converted in some cases to "%20" (that's the URL-encoded equivalent of a space). It's generally a bad idea to have internal URLs with spaces, and this can lead to minor problems. This explanation sounds a little dubious to me. I'd highly recommend you run an internal crawl with a tool like Xenu or Screaming Frog - you might turn up badly formed internal URLs. I can't prove that, but I'd check if it were me. Hyphens don't "turn into" spaces.

Overall, this reads to me like a list of excuses, not solutions.

Agreed - this is generally an issue with relative paths, and job one is to fix it. In most cases, you really don't want these crawled at all. I do think rel=canonical is a good bet here - 301 redirects can get really tricky with http/https, and you can end up creating loops. It can be done right, but it's also easy to screw up, in my experience.

I tend to agree - these pages are often very low-value for Google and can spin out of control. The canonical tag is a great way to conslidate unavoidable duplicates, but in many cases it's better not to create them at all. Of course, these situations can be very complex, and it's tough to speak in generalities.

If nothing unusual is going on at the sub-domain level, it's probably one of two things:

(1) Either the site is new enough or has a small enough link profile that our stats are just imprecise.

(2) There's a strong bias toward home-page links, either due to the link-building strategy, the type of the content, or the size of the site (if it's small, for example).

It's not necessarily something to worry about. The only red flag we look for is a low MozTrust to MozRank ratio (could mean low-quality links). It might indicate that some link-building to deep pages could help you, but that really depends on the site, content, etc.

Is the home-page on an unusual sub-domain, or something other than the root?

It's a bit hard to speak in generalities, without seeing the page/site, but is this your home-page or a deeper page? In some cases, it is possible for one page on a site to get a lot of authority, but for that authority to be spread thin due to weaknesses in the rest of the site.

What do you Page Authority and Domain Authority look like?

Looking at some Zemanta links in the wild, it doesn't seem too aggressive. Take for example, the bottom of this blog:

http://kevinwmccarthy.com

It depends on your goals. From a traffic perspective, as long as the links are relevant, it's fine. I can't tell you whether Zemanta will drive traffic to your particular site (I haven't used it), but it's certainly plausible. It probably depends a lot on your niche and the population of sites using the system.

From an SEO standpoint, my gut reaction is that it's fairly low-risk but may also be low-gain. Google could easily recognize these plug-in links and simply devalue them. I doubt a block of four relevant links is going to get outright penalized - it's more likely Google will just see this as a link-exchange of sorts and simply dial down the volume on those links.

That's not to say it's risk-free (we've certainly seen many link networks get hit in the past year), but I suspect that John was speaking in generalities and not about Zemanta in particular. On the other hand, no tool like this should be seen as some kind of panacea for link-building. It's just one piece of the puzzle for getting exposure and traffic.

Unfortunately, it can vary wildly - in some cases, those variants aren't really crawlable and/or Goolge will ignore them. In other cases, they can spin out a lot of duplicates. It's hard to speak to the crawlability without digging into the site, but you can test for indexed copies with a Google query like:

site:example.com inurl:scid

If you're seeing versions of these URLs indexed, then you may have a problem. GWT parameter handling is one option, but you could also use rel=canonical to point these variants back to the original page. It depends a bit on the situation and the scope.

In some cases, we return a warning if the canonical doesn't match the display URL. I realize this can be confusing, because often canonicals don't match the page, by necessity. It's essentially just a heads up, in that case, to make sure no one does anything dangerous. There are two canonical messages, though - one is an error or warning, and one is just a notice. I'm not sure which one you're seeing.

As Sean said, though, I'm not seeing any obvious issues with the canonical tag on your blog. This may just be a hyperactive warning on our part.

Haven't heard of that with IE9, but from an SEO standpoint, 302s everywhere is much more risky than a few 301s mis-firing as 404s. I get why they're concerned, but this is the wrong solution. Is there a way to set up the redirects within the page headers and only returns 302s for IE9, for the short-term. That's not ideal, but it's at least a stop-gap solution. I'm sincerely afraid their short-term "fix" could cause you long-term problems.

I tend to agree with Moosa that cross-domain canonical can be a bit unpredictable. In theory, it can work well for syndicated (or anything like syndicated) content, and I've heard of examples where it worked very well. Some big newspapers use it, for example. There are also cases where Google can choose to ignore cross-domain canonical, so sometimes you have to try it and see. The good news is that the risk is low - worst case, it just won't do much.

Of course, it can keep your other sites from ranking for that same content, but it sounds like that's what you intend anyway. My gut reaction is that it should boost the main site somewhat - I just don't know how much.

My only warning about the 301-redirect is that these approaches are very different in one respect - the visitors. A 301-redirect will take visitors from one site to the other. A cross-domain canonical only impacts Google. The 301-redirect is typically more powerful in this case, but it may not be appropriate for your scenario.

I like that approach a lot. Thanks for the heads up.

I have to disagree about link juice. In many cases, canonical tags will work much like 301-redirects, and do seem to pass link-juice. I've even seen experiments where people used canonical tags to move an entire domain. I wouldn't recommend it (except in rare cases), but it seemed to work.

I am concerned, though, that you're linking internally to one version, but then using canonical to point to another version. I find that's a bad idea - while it sometimes works, you're sending a mixed signal, and it can cause problems for your SEO efforts. I personally think that a truly canonical URL should be used consistently across the site, including in internal links. Internal links are one of your strongest canonicalization cues.

Unfortunately, it's hard to say if link-juice is being passed, given the mixed signal. I'm afraid there's no great way to measure it, at least on the level of the individual link. I should add that Google also isn't a big fan of setting a canonical to page 1 of search results. They'd generally rather you canonical to a "View All" or use rel=prev/next. Canonical isn't always a good bet for pagination these days.

We don't usually take private info in public questions, but if you want to, Private Message me the domain (via my profile). I'm really curious about (1) and I'd love to take a peek.

If you're willing to piece together multiple sources, I can definitely give you some starting points:

(1) First, dropping from 21K pages indexed in Google to 240 definitely seems odd. Are you hitting omitted results? You may have to shut off filtering in the URL (&filter=0).

(2) You can also divide the site up logically and run "site:" on sub-folders, parameters, etc. Say, for example:

site:example.com/blog

site:example.com/shop

site:example.com/uk

As long as there's some logical structure, you can use it to break the index request down into smaller chunks. Don't forget to use inurl: for URL parameters (filters, pagination, etc.).

(3) This takes a while, but split up your XML sitemaps into logical clusters - say, one for major pages, one for top-level topics/categories, one for sub-categories, one for products. That way, you'll get a cleaner could of what kind of pages are indexed, and you'll know where your gaps are.

(4) Run a desktop crawler on the site, like Xenu or Screaming Frog (Xenu is free, but PC only and harder to use. Screaming Frog has a yearly fee, but it's an excellent tool). This won't necessarily tell you what Google has indexed, but it will help you see how your site is being crawled and where problems are occurring.

I wrote a mega-post a while back on all the different kinds of duplicate content. Sometimes, just seeing examples can help you catch a problem you might be having. It's at:

http://www.seomoz.org/blog/duplicate-content-in-a-post-panda-world

How is your transfer page working, technically - it seems to be resolving as a straight 200 for me. Those transfer pages are usually where affiliates start to get creative (they might 302, for example).

Yeah, I think Chad has the bases covered - sharing an account is really only an issue if you don't want Google to know that the two sites are connected. For example, if you're cross-linking a lot of sites, you might want to keep their ownership as separate as possible. Most of those cases are at least gray-hat, though, and often Google has other cues.

In general, sharing an account is no big deal. Sharing the same IP address has some minimal risks (in the past, if a site on that IP was penalized, it could cross over), but even that seems to be going down over time. Now that we're running out of IPv4 addresses, shared hosting is a lot more common. Personally, I still like unique IPs whenever possible, but sharing a host is no big deal.

I guess the only other exception would be from an engineering standpoint. If you were running two sites for reliability/up-time (in case one went down), then sharing a host would defeat that purpose. That doesn't sound like what you have in mind, though.

Just one adjustment to this - although I think David's right that the canonical tag can be a good solution. Although Google can index https: fine, the issue is whether you're creating duplicates. If you have duplicates, then it's possible that the https: version could be the one you want as canonical. In this case, it doesn't sound like it, but I just wanted to point that out.

Of course, long-term, you should sort out why these are being created. A desktop crawler like Xenu or Screaming Frog may be the best bet, but I'd hit the WordPress forums, too. Odds are it's a common issue. Typically, it happens when some deeper page (like a shopping cart) on a site is secure, and then the links are all relative ("/about.php", for example). Then, those links get crawled as both secure and non-secure.

Unfortunately, I'm not a WordPress expert, so I can only speak in generalities.

Since these are really external links, it's a bit different calculus, especially since they're deeper on the site. "Nofollow" won't really help from a link juice perspective (it still gets burnt), although it would disavow those links, in a sense.

How is the target page redirecting to the outside site? I'm having a hard time telling with header checkers. That could impact the SEO implications quite a bit. This almost seems like an affiliate link sort of scenario, so it's more a matter of how Google sees the hundreds of links between these two sites.

They're definitely burning up some internal link equity, but at this level of the site (search results to product pages), that's not as big of an issue as if this were the home-page or a major category page, etc.

Yeah - I'm really afraid that stacking all those sub-domains is going to cause you long-term issues with your link-building, and that some of those sub-domains could fragment. If the country needs to be in a sub-domain, then I think the hybrid approach (with "/shop" as a sub-folder) may cause you less trouble.

I will warn, though, that any change like this carries some risk. You'll have to put proper 301-redirects in place.

I might try the href lang tags first, though, and see if it helps the current problem (it may take a few weeks). Changing too many aspects of the on-page SEO at once could cause you a lot of grief.

I'm happy to hear you've found it useful. We're definitely working on next steps to make it more powerful - should be some cool data coming out on the site and blog over the next few months.

Yeah, the diplomacy is the hard part. I had a client who I literally spent about 4 hours of CRO work on (as a test project) with surprisingly good results, and they didn't care. They then went out and spent $10K on a new logo (I'll save my opinions about that project).

Middle ground - could you create 1-2 vanity metrics of your own. Take all the keywords tracked, show how many your beating the competition in right now, and try to get that number, up, for example. Or create some derivative of organic traffic (put it on a 100 point scale, with 100 being the best-of-all-possible worlds goal). Obviously, these paths can be dangerous, too, and any one metric can always be misleading. I'm just thinking that, if you have to pick one, better for you to pick it than him.

If you have separate GWT profiles, then I think the XML sitemap may have to be under the sub-domain - Google has to be able to access it from a sub-domain URL. It doesn't have to be in the root of the sub-domain.

I'm not clear on what the "shop." pages are, but stacking sub-domains like that sounds like it's getting pretty messy. Why the separation?

Have you seen a corresponding drop-off in the ListFinder pages over that time. If the canonical is kicking in, you should see some of those pages fall out as more ConsumerBase pages kick in.

Is there a reason your canonical'ing from the more indexed site to the less indexed one. It could be a mixed signal if Google things that ListFinder is a more powerful or authoritative site. Cross-domain can get tricky fast.

Unfortunately, beyond NOINDEX'ing, it's about your best option, and certainly one of your safest. It's really hard to predict what the combo of cross-domain canonical plus link would do. From a dupe content standpoint, it's risk free. From the standpoint of creating 80K links from one of your sites to another of your sites, it's a little risky (don't want to look like a link network). Since you're only talking two sites, though, it's probably not a huge issue, especially with the canonical already in place.

Google interprets cross-domain canonical heavily, so it can be a little hard to predict and control. Interestingly, the ConsumerBase site has higher Domain Authority, but the page you provided has lower Page Authority than its "sister" page. Might be a result of your internal linking structure giving more power to the ListFinder pages.

All else being equal, I think it's very similar - you're just diluting your unique keywords. The only exception would be if the phrase "cheap 2012 honda accord" is very high volume. Then, you might want to target that on the page. I'd be selective about that, though - don't just add it to every URL.

Unfortunately, "cheap" is just one of those words that looks low quality to Google, because everyone is chasing. So, I think you have to be all that much more careful. If you sold only used cars and you wanted "used-" ("used-2012-honda-accord", for example) in your URLs, that might be different. I'd still proceed with caution, but at least "used" would be a meaningful aspect of your business and the user experience. "Cheap" is a value judgment that everyone wants to claim.

Tend to agree with EGOL, but I'll add some specific thoughts of my own:

(1) Try to come up with a tangible metric, like average ranking across a set of keywords, or total organic search traffic, or total queries, etc. that you can show the boss is more connected to actual traffic and sales. If you can get him to accept a metric that's more on your terms, it'll be better for both of you. I completely agree with EGOL that chasing PR or "authority" is a bit of an affectation.

(2) If he's stuck on something like authority, make sure to define it, and stack the deck. Our metrics - DA and PA, are tricky, and are based on machine learning. In reality, we're trying to predict how well you'll rank. So, we can't really say that authority comes from any one source. Traditionally, authority is things like high-trust, high-PR links, social mentions, and other "brand" signals, but that's very difficult to measure.

(3) I'd also say that whether on-page or links are more valuable really depends on your situation. I've seen a client's organic traffic triple in a few months because we fixed some on-page messes. Likewise, if you've got a Panda penalty due to thin content, then increasing your "authority" 50% could do nothing until you fix the on-page mess. If you've got a beautiful site, architecturally, but have no inbound links, building authority could be like magic. It's a mistake to focus on only one side of the equation.

I think (1) and (3) are intertwined. You really need to make the case: "Do you want a vanity metric, or do you want results?"

I'm not seeing strong evidence of an algorithm update, but it's far from a perfect science. There's flux every day (a lot of it). We're testing an expanded data set that will let us drill down by 20 industry categories and 10+ data centers, but we have to collect enough data to tell what's normal. It could be that 65 degrees is a "hot" day in consumer electronics but a cool day in real estate. It can get subjective fast.

Definitely working on tools to answer those questions, but I don't have much solid information for you about yesterday. Best we know, the moderately high temps we original observed (and they weren't even all that high) were a fluke of changing our data collection method.

It might be worth exploring NOINDEX'ing the useful pages and 410'ing the non-useful ones, if only because sometimes a mix of signals is more palatable to Google. Any time you remove a swatch of content with one method, it can trigger alarm bells. I'll be honest, though - these situations are almost always tricky and you almost always have to measure and adjust. I've never found a method that's right for all situations.