The Self-Deflating Link Strategy

There has been a great deal of buzz lately about the so-called new Google “brand algorithm”. I’ve been asked to form an opinion but the only opinion I can form so far is, “Where’s the proof?” The idea seems intuitively acceptable for several paranoid reasons (don’t take that the wrong way — I’m actually paid to be more paranoid than most people).

Nonetheless, you cannot develop strong, efficient search engine optimization methodologies on the basis of intuition. It just doesn’t work that way.

I have looked at several of the publicly discussed queries where the so-called Brand Invasion has been noticed. Yes, I see the brand name sites appearing in search results. Do I have an explanation for what happened? No — but then, neither does anyone else. People are just shooting in the dark at this point.

However, it’s easy enough to shed a little light on the situation. Of course, what follows assumes that Google is reporting accurate, reliable information. Given all the complaints people have made about the quality of Google search results over the past couple of months, it’s reasonable to ask if the data Google reports for special queries isn’t reliable.

I don’t know the answer to that question.

I did, however, look at some inanchor variations on the so-called Brand Invasion queries. I found that the results were very similar to the natural queries. Usually when inanchor closely matches a natural query that is a sign that hyperoptimization is at work. However, it seemed odd to me that hyperoptimization might be the villain in a brand query where government Web sites and university Web sites were appearing on the first page of Google’s search results.

What’s up with that? We don’t normally associate government and institutional pages with brand value. Clearly, people are not looking very closely at the content in the search results.

Having found some rather odd brand-value listings, I ran a link check for the oddball sites on Google. Naturally, Google reported over 100 links for each site — a clear sign that a large number of links have been pointed at those destinations. I checked a random sampling of the (random selection of) backlinks to see who was linking to what.

Interestingly, I found a number of articles linking to government Web sites and institutional Web sites as a smokescreen for building links. Smokesreen linking is nothing new. The idea is pretty simple. You want to establish credibility for a link to destination A, so you include links to credible destinations B, C, and D on the same page. Any search engine looking at the links sees you’re linking to good neighborhoods and therefore (hopefully) grants at least a probational credibility value to your preferred destination.

People began building smokescreen links in volume when the whole “bad neighborhood” thing became a regular topic on the SEO circuit. That means a lot of gratuitous links have been built up over the past few years — links to highly credible, authoritative Web sites.

It shouldn’t require the genius mentality of a rocket scientist to understand the inevitable consequences of targeted link building. That is, after all, the preferred method of “optimization” for most SEOs. If you point enough smokescreen links at an otherwise obscure Web site, sooner or later your hyperoptimization will overwhelm the natural search ranking factors and push your obscure content to the top of the search results.

It’s not that links matter most or outweigh all other factors to the search engines — that’s completely unnecessary. Rather, it’s that links matter most and outweigh all other factors to the SEOs. Think of it this way: you have 100 muscles and you only choose to exercise 1 of them. Which muscle are you going to mistakenly believe is Nature’s Gift To You, because you can do so much with it?

The hyperoptimization of smokescreen destinations for brand-value queries is only one piece of data in an emerging puzzle. I can certainly speculate about possible algorithmic factors but such speculation is pointless. Neither you nor I can show that an algorithmic change focusing on brands has occurred (no one has yet produced any strong or conclusive evidence of this).

The intuitive evidence is pretty strong. Several people have published some interesting footprint profiles but their data is incomplete. I’m not sure I would want to wade through enough data to make a pattern-analysis case for anything specific. I already have to look at lots of data every week.

Nonetheless, the fact there is a lot of anecdotal analysis going on only means that something is developing. The early analyses all seem to be wrong if only because their arguments do not hold up under close scrutiny. In fact, all the analyses start out with the presumption that Google must have invoked a brand-friendly algorithm and then set forth to prove that presumption correct.

There are probably several things happening at once. After all, Google changes lots of things. If I had to guess I would guess they are engaging some of that new semantic technology they promised to roll out this quarter. Semantic search can still work with links — I have no doubt that Google wants to continue working with links (as few links as possible, but that’s a different story). But the promise of semantic technology is that it enables a search engine to look beyond the links.

Let’s suppose, however, that the inanchor query no longer works as it should. That almost invalidates the coincidental hyperoptimization hypothesis. “Almost”, I say, because I can still find relevant smokescreen links pointing to highranking destinations, even in some cases where the query terms only incur in anchor text for the real destinations (not in the smokescreen links).

It’s possible that Google has implemented a TopicRank semantic algorithm in its Web search. I speculated a couple of weeks ago they might have introduced this ranking methodology to their Blogsearch (I also suggested the names StoryRank and ConceptRank for the hypothetical algorithm).

There is, in fact, a TopicRank paper that was presented at an Information Retrieval conference in Singapore last year. I have only read the Abstract but that TopicRank’s high-level description closely matches what I have observed in the Brand Invasion queries.

That is not to suggest that Google need have implemented that specific algorithm. Rather, I just think there is considerable evidence (including Google’s own admission in their 1st Quarter 2009 earnings call) to show that Google is going semantic.

They seem to be taking baby steps and people who are emotionally tied to the Brand Invasion hypotheses are free to continue hanging their egos on it by suggesting that an artificially manipulated brand query space would be one of many small steps Google could be taking to get to full semantic search. I don’t buy that but we all know we’re going to be hearing endless drivel from the SEO community about the Brand Invasion for up to two years, so we might as well start bracing for more of the same.

Who knows? There may be a common ground somewhere in between all the hypotheses. Google finally conceded not long ago that it had indeed been closely monitoring “certain verticals” after years of speculation and accusation by a small corner of the SEO community. The old adage, “Fool me once, shame on you; fool me twice, shame on me” comes to mind. I won’t be so quick to say Google wouldn’t do that again.

Nonetheless, I think the Smokescreen Link Piper came calling in early 2009 and in some cases it appears that SEOs may have found themselves hoist on their own petards (or maybe one segment of SEOs used brand-value queries to create smokescreen links that ended up hurting another segment of SEOs).

Let’s assume for a moment my smokescreen link hypothesis is right. That means you can rev up your link building engines and charge ahead full speed — you may be able to recapture lost territory.

Now let’s assume my smokescreen link hypothesis is wrong. That means we don’t know what happened, so maybe the old link building trick will work once again.

But let’s make one last assumption: that Google has gone more semantic than we expected. That means you need to start building a lot of relevant copy around your not-so-powerful links. And it could mean a whole lot of other things, too.

These are by no means the only possible explantaions. After all, Google could merely have launched another wave of link executions in its ongoing war against paid links. Matt Cutts’ recent comments on that front are similar to arguments he has made in the past, which coincided with new link pogroms. It could be that Google targeted certain verticals for heavy-duty cleanup operations.

There is certainly a financial incentive for Google to improve its search results, which have been declining in quality (in my opinion) for over a year. The better the quality of the sites shown to users, the more likely users are to click on the ads. How does that work? Well, actually, I don’t know. But I’m sure someone will come up with an explanation — they always do.

In the meantime, if you’re one of those SEOs who likes to build smokescreen links, ask yourself: whose cause are you really helping? Are you SURE you want those sites to rank for that anchor text? It doesn’t have to be the anchor text you point to them. It might be the anchor text you hope to pass to yourself.

Think about it.