Welcome to the Q&A Forum

You could use the 3rd domain as the "face" of the brand, temporarily split traffic to the 2 other domains and then, if that gives you trouble, integrate those 2 domains as sub-folders of the 3rd domain (if that makes sense). Then, the user experience would at least be unified over time.

Unfortunately, I'm not sure what the payment/technical issues are - there may be ways around it (there usually are), but they can get tricky fast. I've coded a few payment gateways in the past, so I know what a pain it can be.

I'd agree but probably feel a bit more strongly about it - post-Panda, these kinds of pages can definitely cause you harm (depending on the scope). If you're not linking to them and they're off on a sub-domain, they may not be creating any practical issues, but it's still better to prevent problems. META NOINDEX probably is a good bet here for prevention.

Are there specific keywords you were ranking on that drove traffic to this page. I don't see that this page has many direct links, and it's not optimized for any popular terms, so I'm surprised that login form was getting much direct traffic.

Is it possible your internal link architecture changed? Maybe you're pulling a ton of search traffic to the home-page, but it's not getting pulled through to the deeper page now?

I'm also a bit confused, because Rediff.com is an outside site service. If you can explain what you were ranking for and dig into the specifics, that would certainly help.

It also appears that you had some pretty strong social signals to that page. It's possible that those gave you a strong short-term rankings boost, but that it faded over time. The social signals aren't as long-lasting as permanent links.

The new Difficulty score gets pretty aggressively scaled, because we found that the distribution of PA/DA was bunching up a bit, and fell almost entirely in the 25-85 range (looking across an entire SERP). So, a 25 gets scaled down to nearly zero, to give the new metric more granularity.

It looks like the CTR-adjusted PA for "Naturmode" on Google.de (for example) is right around 26, so it may be that the PAs have gone down a bit over time as well. Even by the non-weighted metric, I'd expect that to be a solid 10 points lower than the old difficulty score.

We're going to dig in and try to find out if the rescaling is too aggressive, once we have more data from real-world KWE usage. In general, though, you're going to see a bigger range of difficulty scores with the new metric. It's pretty tough to meaningfully compare the old and new scores.

My short-term goal is to provide some at-a-glance drill-down metrics (like domain diversity or EMD influence) on the Mozcast site. That way, when something happens, people can immediately  look at half a dozen metrics and see if there's a pattern across the data as a whole. No ETA on that yet, but hopefully within a few weeks.

It sounds like you've got 2 different questions here:

(1) In terms of the new/3rd domain, what's the goal? If it's just to get an exact-match domain and then send traffic to one of the 2 main domains, that could look like a doorway page. At best, it's a much weaker tactic than it was a few years ago. Keep in mind that you have to support these domains, build links to them, etc. In most cases, you're much better off focusing on your core domains.

(2) @Highland is right - your current domain may not pass all link-juice to the new domain, even with 301-redirects. You also may lose/gain depending on the domain name itself. That's not to say that you should never change domains, just that it carries risk and should be done carefully.

I'm wondering if a sub-directory structure would make more sense for your regions. Are these within one country (ccTLD) or international. You might be better off merging everything into one domain, honestly. Of course, this is both an SEO and a business decision, so it's hard to give responsible advice in just a few minutes.

If it were a brand new site and the target markets were pretty equal (or the Brazilian market was relatively small), I think the sub-folder approach is a good bet. You could use <rel="alternate" hreflang...> for language-equivalent pages, and the combination of a .com plus language/country targeting should work fairly well.

In this case, though, since the Brazilian market is so strong, and you're already dependent on that .com.br domain, I think making the switch completely would be very high-risk. From an organic standpoint, I would advise against redirecting the .com.br domain to the Brazilian sub-domainfolder

I'd suggest either launching the new domain with  <rel="alternate" hreflang...>and no Brazilian sub-folder, or, as you said, setting up <rel="canonical"...>from the sub-folder to the .com.br domain (and keeping that the main site for Brazilian visitors). As the new site gains ground and you see how it performs, you can always switch them out later.</rel="canonical"...></rel="alternate" hreflang...>

As Michael said, make sure you're not double-dipping on the Adwords side with two domains in the same region. That could land your account in hot water. For now, if the .com.br stayed canonical for Brazil, AdWords should continue to use the .com.br site. You can set up regional/country campaigns for the .com site and target them specifically. That shouldn't be a problem and will give you time to build up the quality score and history on the new domain.

Just to add to Ryan's comments - if you had massive 500 issues, then you might theoretically argue that 301'ing would keep Google from crawling so many errors. At best, though, it's a band-aid, and maybe even a poor-fitting one. The better question is - why are those 500s occurring. Ultimately, they should be fixed, not patched.

Usually, Google isn't going to penalize a one-time 500 error or a short-term server problem. The only time I could see 301'ing is if you knew you had a major problem and couldn't fix it for a few days. The 301 (or possibly 302, in this case) could buffer you from crawl problems while you made the fixes. Obviously, that wouldn't be an ideal situation.

Unfortunately, since MozCast is a side project, we haven't set up an SSL certificate yet. So, you're likely going to have problems using the widge on secure/https sites. How much of a problem it causes (warning vs. error) may depend on your site setup.

I don't have an ETA on fixing that, but we'll add it to the queue, and I apologize for the inconvenience.

I suspect we count each character as just one character, so we'd see your title as six characters long. Since most "words" in Chinese are 1-2 characters long (I speak a tiny bit of Mandarin, poorly), it's fair to say that this guideline isn't going to be very useful. Generally, a title being too short is just a heads up - it's not the kiss of death, but something you might want to consider.

Do you have any notion of whether their posts are getting indexed faster than yours? Even if they re-post your content, they could get indexed first if they seem to have more authority. You can ping certain services before you roll out to your RSS feed, for example. This post at BlueGlass talks about it:

http://www.blueglass.com/blog/content-scraping-prevention-benefits/

There is always the legal route, include DMCA takedowns. At some point, unfortunately, it may come down to that. If it's just one site (and not 100s), that's more feasible.

I've been through something like this once, and, unfortunately, even with substantive ad spend, it took a lot of time and effort to get an AdWords rep involved. You have a couple of choices:

(1) Complain very loudly, including Twitter and Google Webmaster Forums. Yes, it's not the ideal solution, but it works more often than it should.

(2) Contact anyone you can at AdWords and try to get rep time. I'll tell you what we did - I'm not saying you should do it, but desperate times, etc. We went to a Google Analytics meet-up and got a few cards from Googlers. Then, we emailed all of them. Since they were out of town (for the meet-up), we got their vacation auto-responders, so we emailed all of their bosses. This continued until some VP got so annoyed he demanded someone assign us a rep to shut us up.

Look, I'm not proud of it, but that client got their ban lifted.

If there are links (from external sites) to these images/pages, I'd probably use the canonical tag (to the library, as you said). It's not technically a true duplicate, but that would preserve the link-juice. If there are no links, you could just META NOINDEX them.

I'll add this, though, in your case. Are these the only images (I count 21 of them)? Technically, they're near duplicates, in the sense that they have the same TITLE tags, etc., but if you're talking 2-dozen pages,  I wouldn't lose sleep over it. That scale of duplication isn't likely to cause you much trouble.

I'd tend to agree with Miriam. While I understand your concerns, we're still not really clear on how Google uses user metrics, and their impact on ranking is limited at best. New developments like RankBrain are more likely to factor them in, but I don't think high bounce rates from poorly localized visitors is a huge concern.

Now, weight that against the risks of just cutting these pages out of the index or ham-stringing them somehow, and I think the risk of damaging your site is much more than the risk of some mediocre user metrics for some parts of the country. Your ability to rank nationwide may not convert well, but it could have general organic benefits, be attracting links, etc.

If you were asking me if you should pursue a nationwide strategy when you're a local business, then I'd say "no" - that would be money and time poorly spent. In your case, though, you already have that traffic - I fear that trying to remove it out of uncertainty over quality is only going to do you more harm than good.

It's tough to say without seeing the pages and searches in question, but if you feel these pages are generally of decent quality to organic search visitors, I'd leave them alone.

There's no easy answer, I'm afraid, and if an answer looks too easy, I'd stay away from it. Building low-quality links might help in the short-term, but it's too high-risk in the long-term. Plus, if you're combining it with duplicate content, you've got multiple quality issues in play (at least, in Google's eyes - I'm not making a judgment calling about using product descriptions, which is very common).

You say that unique text is proven to have worked, and yet it isn't an option. Why? If it's a matter of time/cost, I'd strongly consider not only the long-term ROI but the possibility of investing selectively. For example, you don't have to write unique text for every product you sell (or re-sell) - you could pick the top 10% of products (which may account for 90% of sales) and start with those. Even the top 1% would be a start. Small investments in the right places could yield large returns here.

The other option that people don't like to hear but really is powerful is to consider more carefully focusing your link equity on a smaller number of products. The more products you list, the more duplicates you have, and some of those products are probably very poor sellers or have very poor profit margins. What if you focused your site architecture on 25% of the total products? You'd focus your authority more and each page would be stronger, relative to your competitors.

One easy win is to make sure you're not dealing with any internal duplicate content (product options pages, search filters, etc.). If you're compounding external duplication with internal duplication, it's only going to make all of your problems worse. The internal duplication is much easier to solve.

Yeah, it gets tricky, as this data seems to come from multiple sources. Google is probably using local business (G+) data and industry data, but this is also, at least in part, a pilot paid program:

http://adwords.blogspot.com/2015/09/google-hotel-ads-makes-it-easier-for.html

The problem is that this program isn't open to everyone. So, make sure you're listed consistently across data sources, and, if you have an AdWords rep, push on them for your options. Beyond that, it's a waiting game, I'm afraid.

When you're looking at 200+ ranking factor, there's a lot we can't account for even across all of our metrics. A few possibilities:

(1) The #2 site's links are being devalued, for some reason, due to quality issues.

(2) There are geo-targeting signals out of whack or focused outside of the US market. If the site is ranking #1 on Google.co.uk and has a clear UK connection, that could lower it's value on Google.com slightly. It's not the kiss of death, but it can make a difference.

(3) The #1 (currently on Google.com) site has recent activity that we're not aware of yet (link-building, especially).

(4) As Joshua said, #1 could be targeting inbound anchor text better.

(5) As Brandon said, #1 may have an advantage on user signals.

I don't think that Google currently "sandboxes" new sites in they way they may have once. What I think we're seeing is a grace period where new sites get a chance to rank while Google evaluates their link profile. If, after a couple of months, those links look spammy, the site may drop. In most cases I've seen, though, that's not a #1 vs. #2 sort of thing.

Unfortunately, it's a bit of a black box right now. The timing is unclear (it's usually not immediate for the information to go from Wikipedia to Freebase to Knowledge Graph), but a couple of notes:

(1) If you can, make a direct edit to Freebase. I'm not sure how/when the Wikipedia to Freebase transfer happens.

(2) Definitely get the information into another source, like Google+. Google seems to look for multiple sources, whenever possible.

Andrew Isidoro has some great posts on the subject (and he's one of the only people writing about it):

http://moz.com/blog/i-am-an-entity-hacking-the-knowledge-graph

Thanks - happy to pass that along. We're actually in the middle of a long-term spam detection project to help notify people when a site seems to be suspicious or is likely to be penalized by Google. Eventually, this may find its way into DA/PA. We don't want to use ranking and Google's own numbers, as it creates a bit of a problematic data dependency for us (especially long-term).

Our general opinion (obviously, we're biased) is that the SEOmoz stats are better at tracking high-quality links, but the Majestic stats are fresher and track recent links better - if I had to over-simplify it. When you say your "domain value number" is 7, do you mean MozRank? What are your DA/PA stats like?

Keep in mind that, regardless of which numbers you use, these are only two pieces of the ranking puzzle, which has hundreds of factors. Our MozRank and MozTrust are analogous to Google PageRank and TrustRank, but there are many other factors in link quality and there are many ranking factors that have nothing to do with links.

If your site is only a month old, it's even harder to get a clear picture, as Google may still be assessing you. Your ranking could move a lot in the next few weeks.

One of our community members recently wrote a post that's a pretty good introduction to other ranking factors:

http://www.seomoz.org/blog/going-beyond-moz-metrics-to-answer-why-is-this-site-outranking-me

Huh, that's weird. Haven't seen that, no.

If they're really just low quality (spun articles, blog comments, etc.), then I agree with Casey. You can cause more trouble taking a machete to your links - move forward with stronger, better links, and the past will probably be forgiven. My only caveat is that if these links appear to be paid or farmed, I might be a bit more aggressive with removing some of them.

I'm honestly not sure if there's a great way to use that process to drill back into the links to see which ones are low quality - it was intended more as a big picture look at your entire profile. We're actively working on more spam detection metrics, but it's a tricky business and we're really concerned about potentially giving people false alarms, especially with so many people hacking at their links.

I know some folks use OSE to export links and pull out specific bits of data, like very low PA pages, pages with exact-match anchor text on keyword-heavy phrases (non-brand), etc., but even then you've got to be a bit careful. A low PA page isn't necessarily "bad", just relatively weak, from a link perspective.

One spam marker we've found pretty relevant is sites with a disproportionately low MozTrust to their MozRank (or to their DA). Sites with high authority but low trust may have spammy link profiles in general. These are the kinds of factors that 3rd-party link evaluation tools look at. Some of them are a good starting point, but you still need to fact-check.

I'd definitely start with some kind of domain-based analysis. Your 40K links might be coming from 4K domains, or even 400. Getting a broad sense of linking-domain quality will be an easier starting point.

It depends a bit on the scale - if you have a massive amount of affiliate links, they could get devalued, so some people have masked them in the past. The 302 does that, in a sense, but it also blocks the link-juice (or most of it). The 301 redirect would pass the link-juice, but there's a little more risk. If you're talking about 1000s of affiliate links and that's 90% of your link profile, I'd be careful. If it's just one part of your link-building strategy, it's probably not a big risk.

As Istvan mentioned, the canonical tag is another option. Visitors would still see the affiliate URL (which may or may not be ideal, depending on your own business model), but search engines would pass link-juice to the target page. In that case, I'd make the canonical page match the landing page (the home-page might not be a good bet) - it's a bit situational.

This post is a couple of years old, but it covers the basic options:

http://www.seroundtable.com/archives/022446.html

If you had a very large number of internal links, that could dilute your authority in some cases, but some of that should be factored into MozRank. Having too few internal links shouldn't be an issue, unless you were artificially restricting your internal PageRank flow somehow.

This is almost impossible to diagnose without seeing the specific case, but keep in mind that none of our metrics really account for any filtering or devaluation Google might do. So, if they felt some of your links were low quality, or there was a bad pattern to them (too much reliance on one type of link, such as footer links from clients), or you were hit for excessive exact-match anchor text, etc., that wouldn't necessarily show up in our metrics.

The other issue would be any sort of technical issue with the page or even an on-page problem (a low ratio of crawlable text, too many ads, bad targeting, etc.).

Checked my data, and I get a handful of queries with Google Maps URLs, but they're all very specific, branded things, like "google driving directions" or "google earth street view".

I think John is right on all counts. I'll just add that, while the canonical tag should work fine, you may want to consider whether it makes sense to index the premium content at all. It really boils down to whether you think people will link to the premium versions, in which case the canonical tag would help consolidate that link-juice.

The other option, of course, would be to roll-out the wrappers/upgrades dynamically and not even have two copies - everyone would see the same URL, but premium members would see it differently. In most cases, I think that would be the best bet for SEO. It also means you're not maintaining two copies (although I'm guess your current set-up is at least partially dynamic).

Our new Site Crawl can provide a full audit of your site (whether or not pages have issues) with information like crawl depth and Page Authority, but both tools have pros and cons. We use a cloud-based crawler (pardon the term) and Screaming Frog runs on your own machine -- even this has pros and cons, depending on your use-case. We now allow on-demand recrawls for Medium Tier on up. Screaming Frog has more customization, if you're looking for specific patterns and generally offers more raw data, I'd say. Our new Site Crawl is a bit more accessible to most users and helps provide insights, but doesn't have a lot of customization options right now.

If you've got a Moz plan that covers the size of your full site, then either Moz Site Crawl or Screaming Frog should be able to help you audit your full list of pages prior to an HTTP-->HTTPS migration.

Is that the source or target of the redirect? When I check the header on that page, I'm seeing a regular 200 (OK).

Since you use the non-www version of the URLs, I tried them with "www.", but those seem to be 301-redirecting properly. I'm not seeing any 302s offhand.

I'm working on a couple of cases now, and it is extremely tricky. Google often doesn't re-crawl/re-cache deeper pages for weeks or months, so getting the canonical to work can be a long process. Still, it is generally a very effective tag and can happen quickly.

I agree with others that Robots.txt isn't a good bet. It also tends to work badly with pages that are already indexed. It's good for keeping things out of the index (especially whole folders, for example), but once 1000s of pages are indexed, Robots.txt often won't clean them up.

Another option is META NOINDEX, but it depends on the nature of the facets.

A couple of things to check:

(1) Using site: with inurl:, monitor the faceted navigation pages in the Google index. Are the numbers gradually dropping? That's what you want to see - the GWT error may not update very often. Keep in mind that these numbers can be unreliable, so monitor them daily over a few weeks.

(2) Are there are other URLs you're missing? On a large, e-commerce site, it's entirely possibly this wasn't the only problem.

(3) Did you cut the crawl paths? A common problem is that people canonical, 301-redirect, or NOINDEX, but then nofollow or otherwise cut links to those duplicates. Sounds like a good idea, except that the canonical tag has to be crawled to work. I see this a lot, actually.

Honestly no idea at all. We're starting to process some true CTR (clickstream) data at Moz, but it's tough, and nearly impossible for something that's in testing.

Sorry, the thread was getting a bit nested, so I'm breaking out this response, even though it relates to a couple of replies. Although it is tough to tell without specifics, I suspect that you do need to canonicalize these somehow. Geo-pages that only differ by a keyword or two didn't fare well in the Panda updates, and Google is starting to see them as "thin" content. I do actually think the canonical tag is preferable to a 301 here (at least from my understanding of the problem). As Aran said, the 301 will remove the pages for visitors as well, and they may have some value.

While they're technically not 100% duplicates, Google is pretty liberal with their interpretation of the canonical tag in practice (in fact, I'd say they're actual interpretation is much more liberal than their stated interpretation). If the pages really have no search value, you could META NOINDEX them - it's unlikely that they're being linked to.

First, let me say something that may sound strange coming from a Moz employee. Don't try to raise DA. DA is a proxy for your link authority, and if you get obsessed with trying to raise that score alone, you may find yourself link-building in artificial ways. This isn't unique to Moz -- I'd say the same thing about raising toolbar PR in the old days.

In a broader, conceptual sense, DA attempts to measure your overall authority. Ultimately, this is a product of PA and means getting high-quality links to a variety of pages (ideally). From a pure DA perspective, that means links from high-DA domains and a diversity of domains (not thousands of links from a handful of sites). From a broader perspective -- which DA doesn't measure -- pursuerelevant links that will drive traffic and produce value.

So "How do you raise domain authority?" really becomes "How do you build high-quality links?" and that's a tremendously difficult question to answer. Or, maybe better to say that the answer isn't one most people want to hear, because it's going to require a lot of time, energy, and probably money.

I will add this -- don't get tunnel-vision on links from super-high-DA sites. Not every link has to be from the New York Times. Links from low DA sites that are relevant and legitimate are perfectly fine. Plus, it's going to be much easier to approach these sites and build relationships with them, especially in your own industry. Find the sites that would naturally want to link to you, because you offer something of value to them. They don't have to be mega-sites that dominate the web.

I think that Daniel and Gianluca are on track here, but I just wanted to try to structure it a bit. Google has over 200 ranking factors now, and we use "natural" link profile to cover a pretty big chunk of them. Generally, I think natural means:

(1) From a diversity of sources (blogs, forums, PR, etc.)

(2) Using a diversity of tactics (comments, contextual links, etc.)

(3) Having a diversity of anchor text, including keyword-targeted and brand-targeted. This is probably one of the easier cues for "unnatural" links. If you're SEO'ing for "medium blue widgets" and 99% of your anchor text is "medium blue widgets", it's not natural.

(4) From topically relevant sites. This is a nice-to-have, but it's not clear how well Google can measure relevance. General authority and "trust" may be more important, although I think relevance will matter more over time.

(5) Not from paid sources. Technically, ads should be nofollow'ed, and some paid placements are a lot easier to spot. If you're right under a "Sponsors" banner, in a box with 10 links from wildly irrelevant and disconnected sites, or in a sitewide footer, your link is going to look unnatural).

I'll give you that - it's true that the result highlighting alone doesn't prove Google is viewing "shop ipad" as 2 words for ranking purposes. The tough part is separating out why these pages rank - I've seen plenty of URLs with underscores rank perfectly well, but there's no way to prove it isn't on other factors.

I think we're generally agreeing, though, that the potential risks probably outweigh what would probably be a very small boost at best. I'm also a little hesitant on having some pages with dashes and some with underscores, but that's probably just the UX side of me - it feels messy.

I should add that there are sites that hide links without trouble, but it's usually in very standardized places (dropdown menus, for example).

The ticker probably is a bit safer. I wouldn't load it up with links, but at least the behavior is more like tabbed content - you're showing chunks of content at a time. I'm worried that the combo of hidden text and color change on the other approach is just going to set off alarms.

Odds are that this isn't causing you major problems yet, but if the variants are indexed by Google (you can check with the "site:" operator), I would use the META robots tag. Robots.txt is good for prevention, but it's not great at clearing pages out of the index, in my experience.

I hate to say it, but I'm with Marty. There's just no way to tell, and it's completely dependent on the editor(s) for your niche. Recently, I'm seeing many cases where people get a DMOZ listing and it's so deep in the site that the page isn't even indexed - so they wait months for this "great" link that then passes no link-juice at all. File a good submission and get on with life - that's all you can do, unless you know an editor personally.

I should actually add that Google doesn't condone using rel-canonical back to the main search page or page 1. They allow canonical to a "View All" or a complex mix of rel-canonical and rel=prev/next. If you use rel-canonical on too many non-identical pages, they could ignore it (although I don't often find that to be true).

Vanessa Fox just did a write-up on Google's approach:

http://searchengineland.com/implementing-pagination-attributes-correctly-for-google-114970

I have to be honest, though - I'm not a fan of Google's approach. It's incredibly complicated, easy to screw up, doesn't seem to work in all cases, and doesn't work on Bing. This is a very complex issue and really depends on the site in question. Adam Audette did a good write-up:

http://searchengineland.com/five-step-strategy-for-solving-seo-pagination-problems-95494

You mention adding and rewriting content, but are you adding it to existing pages or to new pages. Adding new pages may actually make Panda problems worse, even if they have original content. At some point, you're going to need to reduce your indexed pages (maybe dramatically) and canonicalize the new content to the old, thin content. Just adding content is probably a bad idea.

If the crawlers aren't getting into your deeper content, you may have to focus on some deep link-building, either external/inbound or internally by featuring certain pages (and possibly rotating until they get re-cached). You've got to give Google a push to re-crawl and simply adding content to a page they think is low-value probably won't be enough. Injecting solid links at various levels of the site architecture could help.

The tough part of this is that the update cycle for Panda-related problems is out of sync with the main index. What might traditionally take days or a couple of weeks to fix could take many weeks with Panda. They seem to be apply a separate data adjustment or possibly a machine-learning algorithm, and it's not part of the main algorithm. So, it's possibly you're on the right track, but simply getting pages re-cached isn't going to automatically reverse a Panda problem.

To be fair to Highland, I do think canonical is a good bet here, but I just have to comment that I don't think Google handles these kinds of URLs very well. They should, in theory, but in my experience they rarely do. The problem with order variants is that you can easily spin 100s or 1000s of them and create serious indexation and ranking problems.

For this particular example, the canonical tag is probably best, but there may be cases where certain parameters have no particular value (like a "sort by" parameter). Those are sometimes better off blocked.

I cover a bunch of examples in my mega-post on duplicate content:

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

I'm not seeing any evidence that Google is indexing multiple versions, and your canonical tags appear to be correct. Practically speaking, I think you're ok here.

I'd make sure that you're at least linking consistently (internally) to the "/" version and not both the "/" and no-slash version. If you're using both links, it could confuse crawlers (and our tools). Your search parameters should be ok, though. I don't see any evidence that those have been indexed. See this query:

site:lamangaclubresort.co.uk inurl:datefrom

Sorry, I think my statement may have been unclear. I was throwing out some numbers to try to illustrate what $1,000 could buy someone in low-quality link-building (number of hours, etc.). My broader point, though, was actually AGAINST that kind of approach. I'd rather see people invest in content than cheap links.

What's a "reasonable" number really depends a lot on quality vs. quantity. Someone could get you 100 links/day, but they'd be mostly junk. Some links could take months to build (if they involve building relationships with people and organization), but they might be well worth the investment. It's a balancing act. I think it's fine to get some blog comment links, guest-blog, do article marketing, etc., as long as the quality is reasonable AND you diversify that with hard-to-get links with higher authority.

I'll also make another pitch for content marketing, because it's the only way to put link-building on autopilot, so to speak. Once you invest in content and start to get the word out, that content can naturally attract links. You could get 10/day some days without lifting a finger, once the content is in motion. I think some manual link-building is fine, and probably necessary, but if you can put both in motion, you'll be in much better long-term shape.

I agree with Syed - it really depends on what you're doing with them. If they're just placeholders that have never had a site on them, Google probably won't notice at all. If these were previously interconnected micro-sites and then you turned around and funneled them into one site to capture all the link-juice, then it could certainly look a little manipulative. What's the purpose of the 20+ domains?

Yeah, I'm 100% with Keri - that seems like a really dangerous strategy - 1000s of sub-domains are likely to fragment your site and this is going to look like a low-value tactic to Google. The ranking ability of that extra keyword will be very small, and you could just as easily put it in a directory.

You're also at some Panda risk generating 1000s pages (sub-domains, etc.) just to rank for keywords. In most cases, this is thin content. Granted, I don't know the details of what you're doing, but I'd strongly advise against this.

Google can be very stubborn about wanting to see content at the root level. The other common problem, though, is that the crawler just may not see the other language variants. Are you auto-redirecting by IP (i.e. geo-location)? One major issue is that Google crawls from the US, so you'll need crawl paths to all of the variants.

The first thing to check is that Google is actually reading/honoring the 301. You'll need to see what the crawler is seeing.

Google has added some new tags for sites in multiple languages (especially if the same language is shared across regions/countries). I don't have a lot of good data on them, but Google reps are encouraging their use:

http://googlewebmastercentral.blogspot.com/2011/12/new-markup-for-multilingual-content.html

You may find life is easier, long-term, if you put the dominant language at the root level, along with navigation options for other languages, and then use geo-location to send visitors to other pages. It also saves you a 301-hop for the primary market and will strengthen your SEO a bit for that language.

As much as we love OSE here, it is not a good bet for doing a complete inventory of your entire current site. I'd probably recommend a desktop crawler, like Xenu or Screaming Frog. See my comparison post here:

http://www.seomoz.org/blog/crawler-faceoff-xenu-vs-screaming-frog

I've got to believe that there's some logic that could connect the old and new URLs, though, even if it has to go through a database look-up. Creating 15,000 manual redirect rules isn't only going to be a huge pain, it may cause a huge processing bottleneck. I'd really push back on what "can't" means and why it can't be done. Too often, "can't" is code for "don't want to", in my experience.

Good discussion here, but I endorsed Keith's comment because I do think you have to trust your gut. Directory links are fine, to a point (especially higher-quality ones), but they should be part of a broader, more diverse strategy. If this firm seems like a one-trick pony and you just get a bad vibe, that's important. I can't count how many times I've ignored that gut feeling with a new vendor (or client) and later regretted it.