|Maile Ohye - what exactly is she saying about the Long Tail|
Back in the Over Optimization June 2 and ongoing June 2010 Updates threads there was a link to a report on Maile Ohye's Q&A Keynote [toprankblog.com] at SES Toronto. One of the things she touched on was the Mayday update.
|We tweak little things in our algorithm all the time. Mayday was a significant update that really impacted long tail terms. A lot of people were leveraging long tail phrases for lots of traffic but it was frequently done via automated methods. We’ve looked to eliminate spam, and that’s been a big priority for us. At the same time, there were people developing not quality content (not a violation of guidelines, but also not providing value). What it does is for long tail queries, is we now just consider them queries like anything else. We are going to put as much value in those search results as all search results. |
Besides my inference that Mayday only affected long-term queries, one other thing stood out to me:
|What it does is for long tail queries, is we now just consider them queries like anything else. |
That raises the question, "How was Google treating long terms before Mayday?"
Maybe if we can answer this we might be able to suss out how long tails are being treated post-Mayday.
I don't have any real answers here; I'd like to get a discussion going so I'm just going to throw something out for starters.
How about this: The supplemental results never went away -- we just forgot about them?
We know that when supps were labeled that a heck of a lot long-term query SERPs were pulled from them. Then, because of a lot of confusion, Google decided to no longer label supplemental results. After just a short bit of time there was very, very little continuing discussion about them.
But that doesn't mean that Google wasn't still pulling long tails from the supplemental index.
So coming back to "we now just consider them queries like anything else," might mean that *all* queries are now answered only by pages in the main index.
It almost makes sense. But that's only one possibility. What's yours?
I think previously, due to processing power and speed issues they had to treat the query as an 'extension of the head' meaning they would look for the 'two word phrase' to narrow the results, then look for the rest, but now they are able to 'individualize' the entire phrase.
So before 'green widget information and resources' would go to the top 'green widget' page containing 'information and resources', and now they search for the entire phrase rather than 'head' results pre-ordering the 'sub-query' to show on the results page.
The baffling things is that while many of our pages might be "optimized" for the long tail keyword (meaning that the longtail appears in the URL and in the H1 tags), it also appears in the body of the text, and the content is unique, and is primarily focused around that particular longtail phrase.
For our longtail keywords like "porcelain handmade widget" the whole page is quite literally about porcelain handmade widgets. While the majority of the site (over 200 eccommerce pages) is about widgets in general.
the ironic thing is that we expected our rankings and traffic to IMPROVE as google got better with indexing long tail keywords, not take a nose dive.
In the quoted paragraph Maile Ohye mentions two types of websites that the update was aimed at:
1. Sites with content created by automated methods.
2. Sites with low-quality content that doesn't provide value.
She indicates that excluding pages from these two types of sites improves the search results.
you mean the exact word must be mentioned like "Cheap Red widgets in New York City" ?
Google toned down the importance of exact match titles, perhaps all the way to zero. They were being gamed.
Searching for the example above "Cheap Red widgets in New York City" used to return a lot of exact match pages but now you may see none of them in the top 10 (20-30?). It just means you can't go out of your way to target just longtail (at least not with titles). When every longtail query is met with 1000's of pages with those exact titles the titles became useless anyway.
Auto-generated content and scraper sites simply created a page targeted at every long tail query they knew was getting searched for (thanks to things like Google's keyword tool etc)
I think this is a sledgehammer to crack a nut, baby out with the bathwater and other popular rant openers.
What's the effin' point of having a title tag if Google is going to ignore it. So people were spamming the longtail. So what? Google's quality filters should still have been able to reward relevant and reputable content. Relevant and non-reputable would come lower down.
But the whole point is that it *was* relevant. For a few years if I'm doing a really specific search I've not used Google because they're trying to be too clever. Nowadays, if I didn't need to stay in touch with Google for work, I wouldn't use it.
Stop second-guessing me! I want what I searched for, not something like it, not a great site that mentions what I searched for in passing somewhere on the page. I want a page about WHAT I SEARCHED FOR!
<end of rant>
No, I think 'it's all relative' any more, so attorney = lawyer and there are semantic relationships, etc. but what I think changed is the way they determine which page to show... (It might not be 'all the way there' yet, but it's definitely where things seem to be going IMO.)
Previously, I think (semantically of course) "Cheap Red widgets in New York City" would yield results from Red Widgets (main phrase) with 'Cheap, Discounted, Low Priced' + 'NY, New York' and now they're looking for the entire 'semantically related' phrase.
It could possibly have been 'Widgets' -> Main Phrase; 'Red', 'New' -> Dial Down Phrases; 'Cheap', 'York', 'City' -> Sub-Sub-Phrases. IOW: I think the searches were 'broken up' by some type of 'phrase prominence' and ranked from there, so the 'best Widget page' containing the sub-phrases would be first, but they are now selected for the entire phrase, with phrases determined to be semantically (essentially) the same factored in of course.
So, previously, I think there was a 'dial down' from the main phrase to the sub phrases in the results returned, and now they're looking for the phrase in it's entirety, which would IMO have significant impact on the results for many searches, hence what we are seeing with the 'MayDay' update.
Remember this is only an example, so it could have (likely) been:
'New York City' -> Main Phrase; 'Widgets' -> Dial Down Phrase; 'Red', 'Cheap' -> Sub-Sub Phrases
this is very interesting, we've seen recently that a newly launched bricks and mortar site is acting strangely compared to how they used to on longtail terms.
ie it was built and optimized for
then plus the geo modifier
"widgetorial services any town"
now it's currently knocking around at #15 to 20 for "widgetorial services any town" the same area for "widget services any town" , it's even appeared into the top 100 for "widget services" but is as far back as page 16 to 20 for "widgetorial services IN any town"
this has never been the case before, ordinarily I would have always expected it to be at least as high as without the "in" if not higher.
there definitely seems to have been a change with they way G handles "in" as a part of this.