Google Changes the Web Graph

Google has made some SEO business models less profitable and their search results less useful by placing too much trust on old domains, with the goal of forcing new publishers to be interesting or profitable without love from Google. Of course there are consequences to Google's moves.

Now those older sites have a certain value, and many are drastically underpriced. If you are a creative searcher you may be able to find a few sites worth getting links from or buying outright. Based on Matt Cutts recent FUD post it is pretty clear that short of manual review Google is straight up lacking in the ability to prevent older trusted sites from dominating their search index right now. Worse yet, automated content generation is getting so good that most people will not even be able to detect much of it.

If you have old domains, ranking in Google is like stealing candy from a baby, especially for newly established markets, especially because most competing sites will be too new to be able to compete in the SERPs. As long as the new market is not tech heavy and it does not have huge legal and social implications there probably won't be too many powerful sites. The old domain barrier to entry also lowers the number of people competing for the money to further consolidate the wealth (now you need to either be sitting on an old domain or you need to have BOTH money and knowledge). It almost reminds me of the Affinity Index Rank stuff at Become.com (where you trust authority at the exchange for lower relevancy), but Google has a slightly greater weighting on relevancy, IMHO.

Many old sites hold up fairly well with age. Content based articles about slow changing topics can remain relevant for many years. Until the usage data goes down, people stop citing it, and some citations die away the documents continue to rank strongly. And, of course, there is a rich get richer effect to ranking at the top of the SERPs.

As the web continues to grow, with more bloggers and other news sites providing temporal linkage data and sending usage streams I think Google is going to have to put more weight on those rather than raw domain age and how long it has had links for.

As far as aging goes I think about the worst type of site to buy in terms of monetization is a large directory in a general non business category. The big upsides to buying a non business site when compared with a business site would typically be:

• lower competition in the SERPs

• cheaper price
• better link citation data - on average when compare

But if you cant turn that site into something with a fairly commercial twist to it general non business categories will mean little value from cheap (un)targeted contextual ads. For a general site the ads are hard to target to high value relevant niches, and the search results are obviously more competitive for general terms. In affiliate marketing targeting is even more important than with contextual earnings because you need people to click AND buy.

As far as why old directories are bad, in a couple years of aging what was once a well kept directory could end up looking like a war zone. Sure the link titles and descriptions look good at first glance, but then you click through to find things like:

• broken links

• domainer pay per click ad pages
• totally unrelated sites like porn

Plus if you sell ads next to a directory those will get less clicks if the content area offers users many ways out of your site.

Creating useful directories takes a bunch of upkeep. In spite of paying $20 million for Zeal in 2000 Looksmart is now killing it off in favor of a social bookmarking site. As companies that own major directories shift in favor of bookmarking sites you have to think that search engines are going to start moving in the same way - looking more for temporal data (Yahoo! obviously already is with MyWeb and Del.icio.us).

Think of older link lists as being similar to a directory category page. If you were a search engine would you want to trust these older than dirt links as votes next to their votes for porn sites, domaining sites, and broken links? Google does a ton right now, but eventually that will change.

If sites get too junky people use them less and they stop acquiring as many links, and thus would be hurt by temporal effects, but it takes a while for that to filter through the web.

As the web grows what would stop Google from ranking pages with broken links (or other aged and not well kept indication) much lower (they may already do this in some cases) and passing less outbound link popularity on pages chuck full of broken links.

They could stop their over reliance on links from crusty old domains a few ways:

• looking at temporal data more

• if a page has above a certain percent (or number) of broken links do not allow it to pass PageRank (you would most likely have to do that on a page by page level since so many sites eventually decay away and many published sites offer dated articles that do not change)
• if a page has above a certain percent of broken links allow it to cast its voting power proportional to what percent of links work...ie: if only 10% of the links work only allow it to pass out roughly 10% of that page's PageRank (or 8.5% if you wanted to assume the original .85 dampening factor)
• if a site has above a certain percent of broken links or bad links flag it for review

Many spammers are well tooled up and cash flush. As the ad systems grow in advertiser depth and better target ads spamming gets more profitable. Invariably I think Google is going to have to start doing a few things to fend off from what is going to be a very spam filled year.

Just about everyone with a year or two in the SEO game knows going old is going gold, but outside of the link graph so many other factors are easy to manipulate, and relying too heavily on traffic streams or linkage data could result in them creating an index where every site had some annoying viral marketing aspect to it. Where do they go from here?

Surely posting about hosted content pages over and again isn't going to get people to stop doing it. Also note that Google still serves AdSense ads on the sites mentioned for "still hosting doorway pages," which means Google is still paying them to play.

Google is going back in time with their relevancy and trust to a time when the web was less commercial...largely because their current business model does not foster a functional web. Google placing so much weight on the past is, in my humble opinion, a fundamental admission of that they need to change their business model to inspire higher quality content.

Which of the following two are more profitable (especially when you consider upkeep costs):

• a site with editors that creates original useful compelling content and frequently cites external resources

• an automated algorithmic site that typically limits choices to that which is recommended by third parties, or more often to automated ad optimization services

Google is making more of the web like itself. In a sense, Google's current business model limits the quality of information that you can editorially produce and maintain if the content production needs to be a self sustaining project. They are squeezing the margins on all media by making the ad market so efficient and easy to track.

Having recently bought an old directory I cringe at the idea of sprucing up the directory by cleaning out dead links and replacing them with other similar resources costing nearly as much as the site itself did.

People doing things with passion may not be so easily deterred at creating content at a loss, but for those looking to go independent it probably is nowhere near maximally profitable to have high quality content unless you can create industry standard resources, make others want to create content for you, or use your content to leverage subscriptions, high priced consulting fees or other business relationships.

Here is a review of a 6 page JupiterResearch report that sells for $750. With that price point they probably are not selling a bunch, but price points are a signal of quality. Paying a lot for information means you are more likely to act on it, but paying more for ads just means you are wasting money.

"I know I waste half the money I spend on advertising," department store pioneer John Wanamaker said. "The problem is, I don't know which half."

Some people are touting the cost per influence concept to set up ad networks, but you really don't buy influence. If you do then the people selling it LOSE their influence. As more pages become ad cluttered people will become even more skeptical about what they click.

If you want a parallel about how well the cost per influence strategy works take a peak at how well advertisers and publishers enjoyed Google's print ads. Influential magazines...horrific return on investment.

I think while Google's competitors are scrambling to catch up on the search ad front Google realizes that they have squeezed out much of the efficiency out of that market that they can, and that they are going to have to create a framework that helps:

• lower the cost of content production

• lower the cost of content distribution
• increases demand for media consumption and creation
• helps content publishers create and distribute premium content and / or subscription based content products.
• gives consumers adequate samples while allowing publishers to protect their rights
• organize vertical data streams for data consumption
• takes a cut on the value they create

Peter D (who seems to be on a back and forth link thingie with me right now) recently wrote:

We started with the site being the destination (Yahoo! directory listed sites), then the page (AltaVista serp, or Adwords landing page), and now - the data unit.

In practical terms, it might work like this: if you've got something to say, or to buy, or to sell, make sure that chunk of information is in GoogleBase. The publishers who pull the data from GoogleBase will do the rest, potentially giving you much wider distribution, in the blink of an eye, and with little effort on your part.

Compare that to building a website, marketing it, and managing it. There may, in future, already be an existing, third-party expressway between you and your audience.

I think Google (and others) want to keep making it easier and more efficient to access quick data streams, but I think Google also wants the other end of the market...Google wants a cut on the high end content that isn't profitable enough being monetized by contextual ads.

Their attempts at print ads are probably not only so they can become a market leader in selling them, but also so they can learn how efficient the different media models are, how much value they have, and try to approach publishers with different models that puts Google at the center of the information world...for both free and paid information.

Yahoo! would like some people to believe that they are ahead of Google on the content syndication business, but I don't think they are. The thing that I think will kill Yahoo! in that fight is that they will go after partnerships with bigger companies, whilst Google will get some of those, but also attack that from another angle. Using low cost structure (partly from low overhead from independent publishers, partly from their cheap clock cycles, partly from the value of all the media consumption data they store) and quirky nature and expertise of amateurs to build so much demand that it forces the larger high quality official publishers to need to be part of GoogleBase, either by monetizing the pageviews they get from it or selling their content in it.