Aaron Wall of SEOBook recently predicted that, in 2013, SEOs who “remain overly-public will continue to invent language to serve their own commercial purposes while chastising those who do not fall in line.” I appear to be living up to (the first part) of that promise because I’m calling it: the breakthrough ranking factor of 2013 will be “waves,” a term I just made up.
This will be a somewhat speculative post, so I feel compelled to say that these opinions are my own, and don’t necessarily reflect the opinions of Northcutt as a whole.
Where did this crazy idea come from? It started with the realization that, back in 2009, Google’s Chief Economist told McKinsey Quarterly “I keep saying the sexy job in the next ten years will be statisticians. People think I’m joking, but who would’ve guessed that computer engineers would’ve been the sexy job of the 1990s?”
Now, besides the encouragement that I picked the right major, I also immediately realized that this would have implications for the future of search. There is certainly no doubt that Google is hiring statisticians. Furthermore, the Panda update was almost certainly the result of algorithms that leveraged correlations between various factors and what surveys deemed “bad content.”
After giving the matter some thought, I realized that Google’s massive dataset (the entire internet) offers plenty of opportunities to detect authority and relevance on the web, and links wouldn’t necessarily be the best way to do it in the years to come.
In theory, the best way to identify an online influencer isn’t with links. It’s by measuring the waves they make online.
Okay, So What Are “Waves?”
Let’s start with a real world example. As a science blogger, I know one of the go to places for science news is EurekAlert. Whenever an organization publishes new scientific research, they go to EurekAlert to publish their press release. It then gets syndicated and spread throughout the web. From there the story gets picked up by The New York Times and an onslaught of bloggers (like me) who comment on the news.
The site recently published a press release entitled “Climate model is first to study climate effects of Arctic hurricanes.” A search for this phrase in Google reveals EurekAlert’s version of the article first, followed by several copies of the article and similar articles throughout the web.
This is what I’m calling a “wave.”
In other words, if a site produces content that gets emulated by others and spawns related stories throughout the web, it’s an influential site. That is, in fact, pretty much the definition of influence.
Is it Really Possible for Google to Measure “Waves?”
As somebody with a decent amount of experience in online journalism and copywriting, I can tell you that websites copy each other all the time. They just don’t, for the most part, plagiarize each other. I am also convinced that the technology already exists to make a reasonable guess about whether one article was based on another one.
So I feel fairly confident in saying yes, it is possible for Google to measure such waves. I don’t believe they are doing so currently, but I believe the technology and the statistical techniques are already there to make it happen. Such an algorithm would work something like this:
- Look for statistically significant fluctuations in the terminology, phrases, and subjects discussed on the web, and searched for through the search engine. Look for pages that have similar subject matter and article structure.
- Identify the turning point when the subject increased in popularity, and identify the sites that covered the subject first based on index date.
- Look for correlations between mention of the subject and mentions of specific brand names (co-citation).
- Supplement the process with link data.
If you think this is speculative, you’re right. But if you think it’s far fetched, consider how the Panda update most likely works:
- Use human quality raters to categorize content as either low or high quality.
- Use a machine learning algorithm to find correlations between various factors and which pages were considered low or high quality.
- Use those correlated factors as a proxy for quality, and adjust rankings accordingly.
Panda may also already be using pattern recognition to identify redundant articles: articles that don’t use the exact text as another article, but have a similar topic and structure. In that case, identifying at least some waves would be as easy as tracing these redundant articles to their source.
Furthermore, the search engine has already become very good at interpreting query meaning. Anybody involved in keyword research should already have noticed that minor changes in keyword hardly change the search results, especially in the top three. Combine this with Google Insight and Google Correlate and I’m convinced Google could identify shifts in the chatter across the web with relative ease.
If you still think this is unlikely, consider this:
- Suppose that whenever the algorithm claimed to detect a page that was based on another one, there was only a twenty percent chance it was correct.
- In that case, it would still only need to detect 21 “positives” in order to be 99 percent sure that at least one of those pages really was based on another one. (Use this binomial calculator if you don’t believe me.)
And, if you didn’t know this, Google has already built an algorithm capable of training itself to recognize pictures of humans faces, body parts, and cats with 82, 77, and 75 percent accuracy respectively.
I repeat, those are pictures, not text.
Why Would Google Do This?
- Hard to fake. Chatter spreads to far more corners of the internet than links do (or social does, for that matter). This would give the search engine a lot more data to sort through for unnatural signals than links can alone. Measuring “waves” would be tantamount to measuring influence directly, rather than using links as a proxy.
- Noncommercial. For the most part, only noncommercial subjects propagate through chatter. People don’t stand around chatting about commercial products in forums, social networks, and blogs. This is good news for Google, because it wants most of the commercial content on its own properties, not the organic search results.
- Brands. The one exception here, of course, is big name brands, which people do mention frequently. Since Google would look stupid if big brands didn’t show up in the search results, this also suits the search engine’s motives.
- Intimidation. Such an update would likely cause many SEOs to call it quits and move over to AdWords, a good move for Google.
There’s more than one way Google might use such technology if they end up implementing it.
- As a supplement to links – It may turn out that waves aren’t quite as easy to detect as I’m theorizing. Under these circumstances, they may be used as a method of detecting unnatural links. For example, if a site has a much higher link to wave ratio than average, the link profile is probably artificial.
- As a domain level signal – There is a good chance that waves would be too hard to measure for individual pages. It’s more likely that the search engines would use it to measure the influence of a domain, except in the case of very influential pages.
- As a “replacement” for links – While I don’t think this is likely, it’s possible that waves and other statistical signals may eventually become better signals than links, and end up replacing them as the dominant ranking factor. This very well may be coming, but I doubt it’s coming in 2013.
Why it May Not Happen
It’s possible we won’t see this happen in 2013, but if so, I don’t believe it would be because the technology isn’t there to do it. Instead, it would be because:
- Google’s search results are “good enough” and they would prefer to focus on expanding Google Shopping and their other properties
- Google is too busy building tablets, Project Glass, and cars that drive themselves to invest in search the way they used to
- Google will be too invested in the push toward mobile and social to advance their core search technology
Implications for SEO
Whether or not “waves” become the new ranking factor for 2013, I have no doubt that Google is looking for new and innovative ways to measure influence. If I were in Google’s shoes, “waves” would certainly be my next step forward. Either way, SEOs should be considering expanding their influence by:
- Becoming known as a source of content that can’t be easily found anywhere else
- Building relationships with other influential people online
- Creating products and services that naturally lend themselves to online discussion
- Engaging in community building, crowdsourcing, and gamification
- Focusing on getting indexed rapidly so they are attributed as one of the first sites to cover a subject
So, what do you think? I’d love to hear how people weigh in on this. Is this the direction the search engines are headed, or am I crazy?
Carter Bowles is a freelance writer, science blogger, and SEO enthusiast. He lives in Idaho with his wife and daughter, where he is pursuing degrees in physics and statistics. Carter writes for Northcutt, a Chicago based SEO and Inbound Marketing company. Like Northcutt on Facebook, follow them on Twitter, and check out all they have to offer at www.northcutt.com.