Some people are wildly speculating that Google and other engines may create historical databases of SEOs and site relationships to identify spam. I have no doubt that some sites that go way too far stay penalized for a long time, and that some penalties may flag related sites for review, but I think search engines have enough data and most people leave enough footprints that search engines do not have to dig too deep into history to connect the dots. And there is little upside in them connecting the dots.

If they did connect the dots manually that would take a long time to do it broadly, and if they did it automatically they would run into problems with false relationships. Some sites I once owned were sold to people who do not use them to spam. If ownership relationships took sites out by proxy I could just create spam sites using a competitors details in the Whois data , or heavily link to their sites from the spam sites.

Where people run into problems with spamming is scalability. If you scale out owning many similar domains you are probably going to leave some sort of footprint: cross linking, affiliate ID codes, AdSense account numbers, analytics tracking scripts, a weird page code, similar site size, similar inlink or outlink ratios, similar page size, or maybe some other footprint that you forgot to think of.

Many of those things can be spoofed too, (what is to prevent me from using your AdSense ID on spam?), so in many cases there has to be a hybrid of automated filtering and flagging and manual review.

And even if you are pretty good at keeping your sites unique on your end, if you outsource anything they are going to have a limited network size, likely a routine procedure with footprints, and if their prices are low they are probably going to be forced to create many obvious footprints to stay profitable. And if you use reciprocal or triangular links associated with those large distributed link farms that puts you in those communities far more than some potential historical relationship of some sort. By linking to it you confirm the relationship.

Search engines do not want to ban false positive, so many spammy link related penalties just suppress rankings until the signs of spam go away. Remove the outbound reciprocal link page that associates you with a bad community, get a few quality links, and watch the rankings shoot up. The thing is, once a site gets to be fairly aggressively spammy it rarely becomes less spammy. If it was created without passion it likely dies then turns into a PPC domainer page with footprints. Hiding low value pages deep in the index until the problem goes away is a fairly safe idea for search engineers, because after a domain has been burned it rarely shifts toward quality unless someone else buys it.

One hates to give Jason Calacanis any additional exposure, but how can a person be so against SEO while selling text links for scuba blackjack online? Is he ahead of the market on global warming?

Grow up.. the only thing you're ever going to prove by trying to game my SeRP is that you're low-class idiots.

True, or maybe we are looking for scuba blackjack customers, and knew that you publish high quality original content and ads for that market.

Webmasterworld recently had a good thread about signs of low quality websites. The less a person knows about your topic the more likely they are to rely on general signs of quality (or lack of) when consider if they should link at your site or not.

Common Quality Questions:

Is the design clean? Is the content well organized? Do they have major misspellings on their homepage? Who is behind the site? Is it easy to contact them? Are they referenced by any other credible sources? How unique and useful is the content? How aggressively are ads blended into the content? etc. etc. etc.

Why Proxies for Quality Are Important:

Recently someone spread a God hates fags song website. Friends were instant messaging me about whether it was real or not. Some journalists guessed it wrong. People are getting better at creating fakes. The easier we make it for people to trust us in a snap judgement the more people will trust us (and link to our sites).

These proxies for trust are important, especially when you are new to an established industry, are in a new industry with a small community of support, are in a rapidly growing industry that the media is having a feeding frenzy over, or are the seedy arm of a larger industry.

Example of the Importance of Outside Perception:

If an industry is new, the early leaders of that industry might be determined by mainstream media perception (or other perception outside of that industry). Using blogs as an example, if the media did not constantly pump up the Weblogs Inc. story that company still might be unprofitable today. That media exposure lead to more media exposure, gave the sites the link juice to help them rank, and gave them brand exposure that brought in advertisements.

Relating This to the SEO Industry:

With SEO it is easier to be seen as a SEO expert if you are first seen as an expert on search. It is easier to be trusted as an expert on any topic if your site does not flag common signals of crap.

I just got a link from the WSJ to my keyword research tool, but if I would have scored lower on the proxies for value maybe they never would have linked. And when you get that type of link you can leverage it as an additional signal of trust that makes it easier for others to link at you.

With BlackHatSEO.com, I mentioned as seen in Clickz and Search Engine Watch, but what I didn't mention was that both mentions were brief and in the same syndicated article. When the London Times interviewed me about that site I quickly put up another as seen in at the top of the home page, which will make it easier to get more exposure. You want your press coverage to lead to more press coverage, because those are some of the most trusted links and links that money alone usually can't buy.

But I am Already Doing Well:

Many people who buy consultations are already doing far better than I would expect them to do giving some of the obvious flaws I see with their site structure and marketing methods. Some state that they are already doing well. The point of these sorts of signs of crap is not that you need to fill all the holes to do well, or that you can't do well if you do not fill them, but imagine how much better a site can do after it fixes obvious errors if it was already doing well when it had many errors that undermined its credibility and linkability.

Ever since Google has got more selective with what they will index, the model for profitable SEO changed from chucking up pages and hoping some of them are profitable, to where it makes sense to put more strategy into what you are willing to publish.

The Supplemental Index Hates Parasitic SEO:

Each site will only get so many pages indexed given a certain link authority. And each of those pages will rank based on the domain's authority score, and the authority of the individual page, but each page needs a minimum authority score to get indexed and stay out of the supplemental results - this is how Google is trying to fight off parasitic SEO.

Given that many people are leveraging trusted domains, it makes sense that if you have one that you leverage it in a way that makes sense. CNN will rank for a lot of queries, but it does not make sense for Google to return nothing but CNN. It is good for the health of Google to have some variety in their search results. This is why smaller sites can still compete with the bigger ones, Google needs to use the smaller sites to have variety and to have leverage over the larger sites...to keep the larger sites honest if they are too aggressive in leveraging their authority, or have holes that others are exploiting.

Extending a Profitable Website:

If you have a 100 page niche website you may be able to expand it out to 500 pages without seeing too much of a drop in revenue on those first 100 pages, but eventually you will see some drop off where the cost of additional content (via link authority that it pulls from other pages on your site) nearly matches the revenue potential of the new pages. And then at some point, especially if you are not doing good keyword research, have bad information architecture, create pages that compete with other pages on your site, are not actively participating in your market (gaining links and mindshare), or if you are expanding from a higher margin keyword set to a lower margin one, you may see revenues drop as you add more pages.

The solution to fix this problem is build editorial linkage data and stop adding pages unless they have a net positive profit potential.

What are the costs of content?

• the time and money that went into creating it

• link equity (and the potential to be indexed) that the page takes from other pages
• the mindshare and effort that could have been used doing something potentially more productive
• the time it takes to maintain the content
• if it is bad or off topic content, anything that causes people to unsubscribe, hurts conversion rates, or lowers your perceived value is a cost

How can a Page Create Profit?

• anything that leads people toward telling others about you (links or other word of mouth marketing) is a form of profit

• anything that makes more people pay attention to you or boosts the credibility of your site is a form of profit
• anything that thickens your margins, increases conversion rates, or increases lifetime value of a customer creates profit
• anything that reduces the amount of bad customers you have to deal with is a form of profit

Mixing Up Quality for Profit Potential:

I am still a firm believer in creating content of various quality levels and cost levels, using the authoritative content to get the lower quality content indexed, and using the lower quality content earnings to finance the higher quality ideas, but rather than thinking of each page as another chance to profit it helps to weigh the risks and rewards when mapping out a site and site structure.

Increasing Profit:

Rather than covering many fields broadly consider going deeper into the most profitable areas by

• creating more pages in the expensive niches

• making articles about the most profitable topics semantically correct with lots of variation and rich unique content
• highly representing the most valuable content in your navigational scheme and internal link structure
• creating self reinforcing authority pages in the most profitable verticals
• requesting visitors add content to the most valuable sections or give you feedback on what content ideas they would like to see covered in your most valuable sections
• If your site has more authority than you know what to do with consider adding a user generated content area to your site

Take Out the Trash:

If Google is only indexing a portion of your site make sure you make it easy for them to index your most important content. If you have an under-performing section on your site consider:

• deweighting it's integration in the site's navigational scheme and link structure

• placing more internal and external link weight on the higher performing sections
• if it does not have much profit potential and nobody is linking at it you may want to temporarily block Googlebot from indexing that section using robots.txt, or remove the weak content until you have more link authority and/or a better way to monetize it

I have seen many ad studies that empirically proved that the person doing the study did not collect enough data to publish their findings as irrefutable facts. While recently split testing my new and old landing pages I came across an example that collected far more data than many of these studies do:

Same ad title. Same ad copy. Same display URL. Same keyword. And yet one has 3x the CTR as the other.

If you collect enough small data samples you can prove whatever point you want to. And you may even believe you are being honest when you do so. Numbers don't lie, but they don't always tell the truth, either.

Want to know when your data is accurate? Create another ad in the split group that is an exact copy of one of the ads being tested. Once its variances go away with the other clone then the other split test data might be a bit more believable.