MayDay - What Are the Quality Signals?

[hoot33]

I think google is gearing there broad match search more toward an exact match search whether we like it or not.

But on second thought, I just did a search for the first to random words that came to my head which for some reason were "halo lights", the results are interesting.

For a broad match search google returns listings with the following words highlighted, "headlights" "halo" "halo lights" and "lighting"

Have a look

http://img143.imageshack.us/img143/5695/testeg.jpg

now what this seems to suggest is they have geared there search more towards exact/related keywords.

Eg if they dont think any of the exact serps are what your looking for they hit a database of related keys, yank them, rank them and serve results.

 

[MisterGemini]

I think google is gearing there broad match search more toward an exact match search whether we like it or not.

But on second thought, I just did a search for the first to random words that came to my head which for some reason were "halo lights", the results are interesting.

For a broad match search google returns listings with the following words highlighted, "headlights" "halo" "halo lights" and "lighting"

Have a look

http://img143.imageshack.us/img143/5695/testeg.jpg

now what this seems to suggest is they have geared there search more towards exact/related keywords.

Eg if they dont think any of the exact serps are what your looking for they hit a database of related keys, yank them, rank them and serve results.

That is interesting. Where do you suppose that data is coming from?

They translated your lights into headlights, which was correct. This result doesn't look like its about what they think you are looking for, but a match to what they have available that they think you are looking for. Lets see:

#1. Lighting (extension of root word) Halo exact match
#2. Halo exact match Headlights (related to root word)
#3. Same
#4. Same but without Title match
#5. Same but without Title match or related phrase match in content

So if it doesn't get a match, it is using synonyms and/or extensions of the root terms first, then doing the standard relevance matching.

Does this mean that in order to recapture the long tails, we need to start generating every variation of words/terms used in order to gain exact match?

Seems that way.

Let me try... ummm.. how about black leather sofas

So what I am seeing here is sort interesting without going into backlinks etc.

#1. Exact match with title at front, in content, and category top level to domain
#2. Partial match in title and broad in content and partial in page top level to domain
#3. Same as #2, only the page name is in a sub directory
#4. Partial match in title with partial content match and again in document top level to url
#5 Typical MNS type setup, only with a critical error, The phrase match is not at the beginning of the Title. Everything else is a related match (sofa instead of sofas) except for the domain, which is an exact match.

I didn't bother to go into the rest, it is more of the same after that except for #7, which looks like it should be #1, but when you look at the site... awww guess what, it's a crappy MNS site with the ugly green template that is a puny 1 pager site with adsense thrown on it. Appears to have over 2700 links too. Just can't save it though. Even has two other categories that just go to a dummy 404 that advertises hostgator, but attempts to cookie stuff me (I think).

Now to go into the backlinks here is the kicker. Almost all of the top 5 results had next to no backlinks to those particular pages. A handful at best for each of them, and sometimes none at all. The #7 guy has the most to his page/domain, and it's been up since November last year.

So it appears from these two random samples, that google is getting much more specific in it's matching, and the weight from links doesn't matter as much as the weight in the content matching.

Before I share any more thoughts on this, I'd love to hear what you all think.

 

[ipopbb]

What if a google engineer starts working at yahoo and tells all there secrets.

The engineer goes to jail for stealing trade secrets and yahoo gets sued for damages and is put under court order to not act on or share the trade secrets.

Stealing trade secrets is covered by existing federal laws. Patent law doesn't protect those.

 

[cody41]

Gemini,
I think it's a combination of factors as you've discussed thus far. Content matching is, imo, huge. If you're nesting exact keyword terms in the title/url/header/body content and infusing natural language inclusion into that same content set, you're ahead of the game. I can build a site tomorrow, rougly 10 pages with 1 or 2 backlinks that will get me into a top 10 position for a relatively short period of time. That is a first indicator to me of whether that site has any long term value or not. If i get a positive feeling that it does, i'll build the hell out of it. Like you, I write most of my own pages. It's easy to do once you get a system in place..

Time and time again, the mantra FROM google has been about providing the end user with quality content that goes towards a positive user experience.

Now, if that positive user experience is merely to get that person on page and then quickly off page to allow that person to make a purchase - that's a win in Google's eyes. The kind of landing page(s) that presents the end user with a choice however will make or break the success of the webmaster.

To be honest, i think you're on to something regarding the multi-content inclusion/ranking as a high level factor. Just like every other algo roll out, more and more testing is needed to gain any kind of understanding.

 

[maildigger]

Same here ... I first thought it was because of a suddenly hot summer day

Well this explains why my traffic dropped in half starting today (Friday, June 4th). I've got about 50 niche sites that are keyword specific with exact domains. Most were on page 1. ALL text was unique and had about 600 words per page. Plenty of Amazon products, photos, links, videos, etc. (nothing rehashed from Amazon). Most of the sites have dropped about 10-15 spots. I thought it was maybe the recent very short Scrapebox runs I've done. Guess not.

I have one authority site that I've built over the years that was sitting at #3. It's still sitting at #3. Tons of content on there. Gee...do you think Google wants to see content heavy sites?

 

[MisterGemini]

Gemini,
I think it's a combination of factors as you've discussed thus far. Content matching is, imo, huge. If you're nesting exact keyword terms in the title/url/header/body content and infusing natural language inclusion into that same content set, you're ahead of the game. I can build a site tomorrow, rougly 10 pages with 1 or 2 backlinks that will get me into a top 10 position for a relatively short period of time. That is a first indicator to me of whether that site has any long term value or not. If i get a positive feeling that it does, i'll build the hell out of it. Like you, I write most of my own pages. It's easy to do once you get a system in place..

Time and time again, the mantra FROM google has been about providing the end user with quality content that goes towards a positive user experience.

Now, if that positive user experience is merely to get that person on page and then quickly off page to allow that person to make a purchase - that's a win in Google's eyes. The kind of landing page(s) that presents the end user with a choice however will make or break the success of the webmaster.

To be honest, i think you're on to something regarding the multi-content inclusion/ranking as a high level factor. Just like every other algo roll out, more and more testing is needed to gain any kind of understanding.

Something else I noticed about this sampling is that google clearly identified this search query as a product. What I didn't include in my capture was the top bar that was all shopping results.

I think that they have gotten smarter in terms of what kind of content matches the query as well for a better match.

In this example, each site that ranked highest were the sites that had lists of leather sofas on them. Not the site that had a ton of writing about it.

So back to what you mentioned about quality user experience. I think before you start on a term, you need to see how google views what type of content the user is seeking. black leather sofa obviously is seen as a buy word to google, and thus the user experience is 'shopping'. You don't see discussion, blogs, images or anything else like that showing first, its shopping. Which is correct.

I am going to go out on a limb thinking about my example, and say that when it comes to shopping content, the one with the biggest e-dick and correct title/content match wins. How else is an algo going to determine the quality for this type fo content? Perhaps user review content is a signal also.

I think we need to look at some other types of content and their qualty to see if this is part of what is going on. Bottom line, this comes back to google has categorized the web, and all that traffic is being diverted to a more specific type of content relating to the terms category.

This could spell trouble for some traditional promotional methods of getting links.

 

[Frankie4Fingers]

Well this explains why my traffic dropped in half starting today (Friday, June 4th). I've got about 50 niche sites that are keyword specific with exact domains. Most were on page 1. ALL text was unique and had about 600 words per page. Plenty of Amazon products, photos, links, videos, etc. (nothing rehashed from Amazon). Most of the sites have dropped about 10-15 spots. I thought it was maybe the recent very short Scrapebox runs I've done. Guess not.

Similar experience here. My niche exact-match domains got penalized the same day (interesting to note that I work in the Italian online market, and this means that the algo changes are happening globally simoultaneously).

I'm trying to implement on the sites with the same templates and similar rank droppings different changes to see if I can make a come back to page 1.

The aspect I am working on right now is the load speed of these sites (which I am a bit surprised the OP didn't mentioned as a hypothetical quality signal ). They all were WP powered site on shared hosting, which means above average load times at best.

So on one of this site I implemented WP Cache, on another one WP Super Cache, and on a third one I dumped WP and hard-coded a much lighter template (mix of PHP and HTML) to make its loading times uber-faster.

Let's see what happens in the next days, I'll post the results in this thread. I just hope we'll find soon some solid methodology to work on with this new algo to get decent rankings, because aside the rankings we just lost, I don't know how the hell to procede with some other projects I was going to put online

 

[polocolto]

I stoped all new projects and i'm working to improve the sites i have that haven't been penalized. Untill we have a better understanding of this new algo.

 

[MisterGemini]

Similar experience here. My niche exact-match domains got penalized the same day (interesting to note that I work in the Italian online market, and this means that the algo changes are happening globally simoultaneously).

I'm trying to implement on the sites with the same templates and similar rank droppings different changes to see if I can make a come back to page 1.

The aspect I am working on right now is the load speed of these sites (which I am a bit surprised the OP didn't mentioned as a hypothetical quality signal ). They all were WP powered site on shared hosting, which means above average load times at best.

So on one of this site I implemented WP Cache, on another one WP Super Cache, and on a third one I dumped WP and hard-coded a much lighter template (mix of PHP and HTML) to make its loading times uber-faster.

Let's see what happens in the next days, I'll post the results in this thread. I just hope we'll find soon some solid methodology to work on with this new algo to get decent rankings, because aside the rankings we just lost, I don't know how the hell to procede with some other projects I was going to put online

I'm surprised I didn't mention that either. Another good cache plugin to look at is the W3 Total Cache Plugin. Check it out.

Load time could definitely be a factor. It doesn't explain how the crappy MNS fall out though, as they are about as basic as you get for load time. Though it might have a factor in the ones that do vie for the top.

Generally it's always good to optimize your load time. I hit some pages that take 10 seconds to load and in that 10 seconds I am thinking 'what kind of site is this that is taking this long? Am I going to get this huge page of junk?' That's just me. However load time has been raised as a factor recently this is true.

I would add this to the mix, but I don't know how much weight to put on it. Has anybody run a test of the top sites to see if the load times on them correlated with rankings?

 

[bowcot84]

A couple of thoughts. All my hit sites have been Scrapeboxed. Could over reliance on blog comments hit the quality score?
And what about "over-optimised" sites i.e. H1=title, stuffing in alt tags etc.

Not sure I buy either of these, but thought I'd open it to the floor!!!

 

[GreyWolf]

Although #2 i gotta disagree with... No facts, just logic: how could google NOT know your bounce rate? It only makes sense that everything you keep track of in Ganalytics, they are already keeping track of (probably even before they released Ganalytics)... they're just sharing now.

Well a lot of it they can't keep track of without Ganalytics. Analytics requires you add some code to your pages. Without the code a lot of the information can't be collected.

I'm sure that if you have analytics those factors are being considered. It would be ridiculous if they aren't. So if you're site has an extremely high bounce rate it's possible that analytics could actually be hurting your seo efforts. If you're using the data you get from analytics then you should be working on those factors as part of your seo. If you aren't paying attention to the analytics data, then you very well may be better off not to include it for your pages.

 

[ipopbb]

A couple of thoughts. All my hit sites have been Scrapeboxed. Could over reliance on blog comments hit the quality score?
And what about "over-optimised" sites i.e. H1=title, stuffing in alt tags etc.

Not sure I buy either of these, but thought I'd open it to the floor!!!

Stuffed sites appeared to perform better now in my opinion. In most retail scenarios you 50-100 matches in source to even be competitive.

 

[MisterGemini]

A couple of thoughts. All my hit sites have been Scrapeboxed. Could over reliance on blog comments hit the quality score?
And what about "over-optimised" sites i.e. H1=title, stuffing in alt tags etc.

Not sure I buy either of these, but thought I'd open it to the floor!!!

I would say yes. I have always believed in link diversity, or at least ratios that reflect a more natural progression of links. We already seen that google can tell if a link is sourced out of commentary, and what type of sites are out there based on their site structure and content format.

A diversity of links that starts from a mix of commentary followed by articles, and/or forum posts/site posts can give a good mix of sources.

As we have seen though, the amount of links doesn't appear to have as much impact as the quality of the content matching the genre of the query type. I plan to do a few more tests on this tonight just to see if it's consistent in other categories.

As for the over optimization, I think ipobpp would be the first to tell you stuff like crazy. Basically signals from code tags that match the keyword targets appear to have a dramatic impact on results. These being part of the markup language doesn't come across as content keyword stuffing, but the tag coding certainly contributes to the relevancy of the keyword terms. ipopbb would also lead you to look at additional tags you probably have never used or heard of to use in your pages to help with this.

Personally I don't think there is such a thing as over optimization. You should follow proper sematic stucture in your pages. Though you should not keyword stuff the content. Keep that in the tags and you should be alright.

 

[MisterGemini]

Well a lot of it they can't keep track of without Ganalytics. Analytics requires you add some code to your pages. Without the code a lot of the information can't be collected.

I'm sure that if you have analytics those factors are being considered. It would be ridiculous if they aren't. So if you're site has an extremely high bounce rate it's possible that analytics could actually be hurting your seo efforts. If you're using the data you get from analytics then you should be working on those factors as part of your seo. If you aren't paying attention to the analytics data, then you very well may be better off not to include it for your pages.

Check out their privacy policy for analytics. They straight up say they will use the data internally. They say that they won't pass it on to a 3rd party, that's about it.

Piwik can provide you just as much, if not more intel about your site as GAnalytics. Nobody but you sees it.

 

[GreyWolf]

Check out their privacy policy for analytics. They straight up say they will use the data internally. They say that they won't pass it on to a 3rd party, that's about it.

Piwik can provide you just as much, if not more intel about your site as GAnalytics. Nobody but you sees it.

Well it just depends I think on what the data is.

If you're site is getting a lot of visitors, the bounce rate is low or at leat reasonable, there is a lot of interactivity between the visitors and your site, then it's a really good idea to use Ganalytics. It would be a good idea to let google use that information if it wants too.

If on the other hand, your site sucks. lol. It has a high bounce rate, and the ones that don't bounce don't seem to chack anything else out. Then you're probably better off using a different analysis program. I've never checked out Piwik, but if using Ganalytics is going to hurt your site by providing info to google that would be better it not have, then I'd say that would be a better option.

 

[Tim03]

I've noticed that my trademarked domains, the domains with a company name, are ranking significantly higher. Anyone who noticed the same?

 

[bhatsuccess]

Y'know everyone gets tired of all these updates and sudden mega-shifts we're supposed to make in our SEO but to be honest I have only made minor changes in my overall approach to rank.

Content-rich sites that people are likely to link to rank well period.

So if you have to keyword spliff then by all means do it, but we were doing that 3 years ago. There are only so many dang things you can do to a website.

 

[MisterGemini]

I have just gone over a total of 40 search queries and looked at the results I was getting from google and checked out their sites.

It looks like my initial observations were close to being right.

I don't have the time to underline and point out every single query and show it to you here.

Google has indeed categorized their rank results. That side bar as many have already noted from its first introduction, has taken a chunk of search, but the primary listings on first search, have also gotten far more precise in the type of sites it will first display.

I took samples of keywords from google trends, from shopping.com top searches, and then just thought up some random terms as well like 'poll dancing' 'made by hand instruments' 'how to draw' and so on.

Here are some things I noticed:

1. Wikipedia is always coming up #1 because it seems like everybody is linking to each page! I'm not kidding. I can understand though. People want to add a reference to something, they link it to the appropriate wikipedia page. The end result though has been it getting all these rankings. So next time you think it's a good idea to link to wikipedia for something related, think again about the beast you helped create.
2. Using a variation of a long tail search brought up significantly different results. By adding 'where to buy' before the key term, the side bar menu shifted to shopping at the top and video below, and the #1 spot was no longer wikipedia, but an about page going into an extensive 5 part series about the product. Only snagged that spot because there was no other close title match available.
3. Search phrase order brings up significantly different results. eg. 'where to buy graphics cards' vs. 'graphics cards where to buy'
4. URL based keyword matches just do not come up as much. Partial matches are making apperances though. It seems good to have just one keyword out of a phrase in your domain with the full phrase in your document + title.
5. Document most of the time is showing up higher when its a top level page.
6. epinions is doing something right because they are showing up for all kinds of stuff and their pages are nothing but a list of companies and product parameters that then links to all the various products related.
7. PR means nada. I'm even seen a PR N/A beat out at PR4 because it was a far better match.
8. Keyword phrasing order, google gets the terms even when the words are backwards, but the properly ordered one to match the query goes higher.
9. Video is still a great way to get in the SERPs if your site can't
10. Quality of links appeared to make little difference. From all the sites I went over links were coming from all kinds of sources good and sometimes not so good.

Overall, it seems the quality signals appear to be based on the categorical nature of the content you are presenting. The most important factors I observed and that I have reported previously were:

• Title Exact Match High Density
• Partial URL and/or Top Level Document Exact Match
• Content Exact Match Phrasing
• Content Category Match

That's all there is to it really. The bottom line to this seems to be that traffic is more likely to hit your page based on an exact match rather than on a variation of the keyword, and if the content you present is related to the keyword query type. By this I mean, a buy search term 'buy wii games' produces ecommerce results, not 500 word article pages, or blogs for that matter.

How they understand the various forms of content and what constitutes an ecom site vs. an informational site is still not clear. I can only surmise the signal for that is 'add to cart' and/or 'buy now' within the content to determine that it's a selling site.

What was surprising how often exact match results with maybe one hundred backlinks to the page trumped results just below them that had sometimes tens of thousands of links to their page.

The necessity to optimize for every longtail variation is definitely there. How that takes form in your site us up to you.

Building links is still an important part, but not nearly as important as the attention you put into the pages you build. You have to understand your keywords and the mindset of the person who entered them to deliver the appropriate content. This is basically what it all comes down to. Frankly I think it's a great step forward.

I believe now I have a firm understanding of the quality signals that were referred to in this latest update with google. I am going forward with my projects now with all this, and see what happens.

I hope this makes sense to those who read it, because I'm half asleep finishing this off.

Mister Gemini

 

[Frankie4Fingers]

Hey Mr. Gemini, I gave you rep for your latest post since I'm pretty sure it was a tedious process and I appreciated you shared in detail the results.

Still, I have to disagree on the importance of the Title tag: I'm pretty sure they diminuished the weight of this factor in ranking positions, since different of my sites that lose first page rankings had exact kw title (besides exact kw domain and good on page SEO - and zero or so incoming links) and now in the first page I see sites that barely have the main query kw in the title.

 

[thestig]

I have just gone over a total of 40 search queries and looked at the results I was getting from google and checked out their sites.

It looks like my initial observations were close to being right.

I don't have the time to underline and point out every single query and show it to you here.

Google has indeed categorized their rank results. That side bar as many have already noted from its first introduction, has taken a chunk of search, but the primary listings on first search, have also gotten far more precise in the type of sites it will first display.

I took samples of keywords from google trends, from shopping.com top searches, and then just thought up some random terms as well like 'poll dancing' 'made by hand instruments' 'how to draw' and so on.

Here are some things I noticed:

1. Wikipedia is always coming up #1 because it seems like everybody is linking to each page! I'm not kidding. I can understand though. People want to add a reference to something, they link it to the appropriate wikipedia page. The end result though has been it getting all these rankings. So next time you think it's a good idea to link to wikipedia for something related, think again about the beast you helped create.
2. Using a variation of a long tail search brought up significantly different results. By adding 'where to buy' before the key term, the side bar menu shifted to shopping at the top and video below, and the #1 spot was no longer wikipedia, but an about page going into an extensive 5 part series about the product. Only snagged that spot because there was no other close title match available.
3. Search phrase order brings up significantly different results. eg. 'where to buy graphics cards' vs. 'graphics cards where to buy'
4. URL based keyword matches just do not come up as much. Partial matches are making apperances though. It seems good to have just one keyword out of a phrase in your domain with the full phrase in your document + title.
5. Document most of the time is showing up higher when its a top level page.
6. epinions is doing something right because they are showing up for all kinds of stuff and their pages are nothing but a list of companies and product parameters that then links to all the various products related.
7. PR means nada. I'm even seen a PR N/A beat out at PR4 because it was a far better match.
8. Keyword phrasing order, google gets the terms even when the words are backwards, but the properly ordered one to match the query goes higher.
9. Video is still a great way to get in the SERPs if your site can't
10. Quality of links appeared to make little difference. From all the sites I went over links were coming from all kinds of sources good and sometimes not so good.

Overall, it seems the quality signals appear to be based on the categorical nature of the content you are presenting. The most important factors I observed and that I have reported previously were:

• Title Exact Match High Density
• Partial URL and/or Top Level Document Exact Match
• Content Exact Match Phrasing
• Content Category Match

That's all there is to it really. The bottom line to this seems to be that traffic is more likely to hit your page based on an exact match rather than on a variation of the keyword, and if the content you present is not related to the query. By this I mean, a buy search term 'buy wii games' produces ecommerce results, not 500 word article pages, or blogs for that matter.

How they understand the various forms of content and what constitutes an ecom site vs. an informational site is still not clear. I can only surmise the signal for that is 'add to cart' and/or 'buy now' within the content to determine that it's a selling site.

What was surprising how often exact match results with maybe one hundred backlinks to the page trumped results just below them that had sometimes tens of thousands of links to their page.

The necessity to optimize for every longtail variation is definitely there. How that takes form in your site us up to you.

Building links is still an important part, but not nearly as important as the attention you put into the pages you build. You have to understand your keywords and the mindset of the person who entered them to deliver the appropriate content. This is basically what it all comes down to. Frankly I think it's a great step forward.

I believe now I have a firm understanding of the quality signals that were referred to in this latest update with google. I am going forward with my projects now with all this, and see what happens.

I hope this makes sense to those who read it, because I'm half asleep finishing this off.

Mister Gemini

I got to agree with Mister Gemini, it seems like there looking to give the end user what they really need, which is to match physical products with the exact keyword match, this really can work in our favour in regards to content issues being unique which i think is just google trying to trip us up, as all i'm seeing is alot of sites with more images than content, got an domain that is and exact match which i have, going to build a site simlar to say amazon and see how that works, i really think this is the end of micro niche sites such as made for adsense type.