I worked this out in email this afternoon. It’s stone obvious, once you think about it, so I think it might be on the horizon now — if it’s not already happening.
What the hell are you talking about, Greg?
I think the next level of search engine algorithms is going to be based in an heuristic observation of end-user behavior to correlate keyword relevance to actual relevance.
Do you see? Google and other search engines identify patterns of keywords in static HTML documents to try to identify keyword-relevant content. They do this because it’s cheap. Before that, they used gunk like meta tags because that was even cheaper — because that had too little hardware and software and too many documents to index.
The hardware and software problems are gone, plus Google has a huge and growing database of user behavior that is has harvested from the many bits of Google software people load on their systems. Moreover, Google has learned to draw sophisticated inferences from user behavior.
So consider two web pages. One is very strong on relevant keywords, but weak on useful content. The other is not as strong on keywords, but it delivers an ocean of very useful data. When users click into the first page, they tend to click out right away — high bounce rate, short time on site, few pageviews per visit. Users of the other site stay for hours and read everything twice — low bounce rate, long time on site, many pageviews per visit.
Assuming Google or another search engine can measure all of this end-user behavior, which site is actually more relevant to real people?
This is so obvious that it has to happen. If Google doesn’t do it, its successor as the number one search engine will.
What does it portend for you? For one thing, dumbstunt SEO plays like Localism are doomed. But more importantly: Now and forever, content is king. A highly-passionate, well-written, deeply-informative weblog is going to kick the ass of any site trying to get by on money and high-gloss lipstick.
If you deliver the goods, the search engines are going to find a way to deliver the users.
Technorati Tags: blogging, disintermediation, real estate, real estate marketing, technology
Kay says:
YES!!!
July 9, 2008 — 4:29 pm
Todd Carpenter says:
I totally agree. Google took off because Yahoo was getting to easy to manipulate. Now Google is in the same boat. They will change or someone will replace them. Just think how Powerset Labs, Mahalo, or a del.icio.us influenced Yahoo might change the game.
July 9, 2008 — 4:45 pm
Tom Vanderwell says:
I believe it was Jeff who said….
“Do the right thing no matter what.”
The right thing is providing a “highly-passionate, well-written, deeply-informative weblog” and if we do it right, it will work.
Tom
July 9, 2008 — 5:31 pm
Susan Zanzonico says:
Greg, I think you’re right. It makes perfect sense and seems so obvious and logical once you say it.
July 9, 2008 — 6:54 pm
Anonymous says:
“Now and forever, content is king”…kind of says it all…and that’s why I spend a couple of minutes each day right here.
Thanks
July 9, 2008 — 7:11 pm
Tony Sena says:
I would have to agree with the assessment above. Google is all about content and if they can ranks sites by useful relevance as opposed to key word relevance, I’m all for that!
July 9, 2008 — 7:26 pm
John Sabia says:
just one more reason to “write for people not search engines” people like reading information not spammy words.
July 9, 2008 — 7:40 pm
Hunter Jackson says:
Why not provide both? Highly passionate, useful, important information that is seo’d as well?
July 9, 2008 — 7:42 pm
Barry Cunningham says:
Hunter you got it! Greg makes an excellent point.
I was wondering about Localism myself today. If they imported 200,000 posts as they say..many from years ago…why were’nt those posts already a problem to overcome in the serps? Additionally, with the content most of the Ar members put out, are they expected to now write 2 posts..one for localism and one for the watercooler?
July 9, 2008 — 7:56 pm
Michael Rahmn says:
Thus, free Analytics.
And they don’t really need that data set to deliver the goods. Given long-lived cookies and the fact the back button from an irrelevant (keyword-rich, content-poor) site often leads right back to Google, it’s fairly simple math to extrapolate time on site(s).
July 9, 2008 — 8:58 pm
Brad - Dakno Web Design says:
I think it’s already happening on two accounts.
1) Google’s Search Results do not link to the website but rather to a Google tracking page (right click and “copy link location” in Firefox to see what I am talking about) This gives Google the ability to track and time bounce rates.
2) Google doesn’t just look at how many times a keyword is repeated but instead uses “Latent Semantic Analysis” to determine relevancy. Basically it means that a website has to do more than just repeat a keyword over and over to get ranked. It has to also use keywords that are related to that main keyword. Example: Chicago Bears – keywords such as NFL, Football, Tickets, Team, etc. The thought process is that “useful content” will contain these ancillary keywords in addition to the main keyword.
July 9, 2008 — 9:05 pm
jaybird says:
For many of us RE bloggers this is rad–assuming you know how to write deep, sticky content about RE of course versus what you ate for breakfast or the new store opening up on the corner….
July 10, 2008 — 4:16 am
Greg Cremia says:
I believe they have been doing this for a while now. They patented the methodology a couple of years ago and the google tool bar/spy ware tracks everything and everywhere a user goes on the web.
Watch and you will see sites from the back of the serps get a couple of days on the front page to test their stickiness. If they outperform the site they replaced then they get more chances to stay on the front page. The stickiest sites get to stay. The least sticky sites fall to the back where they belong.
If you are on page 2-3-4 you had better make sure your site is sticky because you might not get 2 chances when your turn at the top comes along.
July 10, 2008 — 5:57 am
Eric Bramlett says:
I agree with you that this would improve search. Couple of points:
1) On-page content analysis is only a small part of the Google algo. In years past, you didn’t even need to have a keyword on your page in order to rank for it. Google changed this to curb Google bombing. It’s why George Bush is no longer #1 for “miserable failure.”
2) If Google is using this, they’re doing so gingerly. Why? It’s VERY easy to manipulate. You can write a script that hops IP’s, or go old school and hire some cheap cross-continent labor to find your site a specific way, and behave a specific way.
3) They might or might not use analytics data. Until Google publicly recommends everyone to install analytics in order to rank (which will probably be never) then it won’t be a requirement to run the script in order to rank. There are plenty of killer sites that don’t use analytics, and belong in the serps.
Great post! It really is the next logical step in search.
July 10, 2008 — 6:22 am
Greg Swann says:
> Watch and you will see sites from the back of the serps get a couple of days on the front page to test their stickiness. If they outperform the site they replaced then they get more chances to stay on the front page. The stickiest sites get to stay. The least sticky sites fall to the back where they belong.
This is what I would expect to see, although there is no reason not to do this continuously, with each data center working independently to improve the data set. Likewise, there is no need to monitor only searches. Experienced end-users are going to have a long list of site they go to regularly without going through Google. Those can be correlated with other people’s search results as well. And Google knows more about feeds than anyone…
The days of SEO gimmicks could be nearing an end.
July 10, 2008 — 6:27 am
Carolyn Gjerde-Tu says:
This does seem like a a logical step for the future. The benchmarking tool on google analytics also seems to confirm stickiness of a site and is a good check for what needs improvement.
July 10, 2008 — 7:18 am
Sam Chapman says:
G has been shuffling results except for the top 2 on page one for a good while now. The thought is that they are checking user behavior for stickiness in order to wrewrite the algo based more on human behavior than on keywords and inbound links. Only time will tell if that is right.
July 10, 2008 — 7:45 am
Greg Cremia says:
>Likewise, there is no need to monitor only searches. Experienced end-users are going to have a long list of site they go to regularly without going through Google. Those can be correlated with other people’s search results as well.>
The patent I mentioned also went into some detail about monitoring peoples favorites/book marked pages. The google tool bar is very intrusive and is providing immense amounts of info from millions of computers.
Google would be foolish to not include this info in their algo.
July 10, 2008 — 7:59 am
Mark McGlothlin says:
Amen. Strong, unique, well crafted content is the cream that will rise to the top.
July 10, 2008 — 9:16 am
Eric Bramlett says:
Like the algo, they’ll only evolve.
July 10, 2008 — 9:40 am
Malok says:
There will always be a cat/mouse game going on between the persons working on the search engine algorithm, and those looking to beat the system.
Hopefully, the algorithm will eventually become so smart/good, that it will make it very difficult for the spammers.
Its best to stick to the old cliche of: act like there aren’t search engines, and create a useful experience for your prospects with good content.
July 11, 2008 — 9:34 am
Joel Ives says:
For some reason I already thought Google did this already. Tracked the bounce rates and etc. to see if the content was good. Big Brother Corporation indeed.
July 11, 2008 — 10:22 am
Sue says:
>>For some reason I already thought Google did this already. Tracked the bounce rates and etc. to see if the content was good.
Me too, but I’m not sure it is currently used for site rankings.
July 11, 2008 — 2:52 pm