We're excited to announce the results of Moz's biannual Search Engine Ranking Correlation Study and Expert Survey, a.k.a. Ranking Factors.
Moz's Ranking Factors study helps identify which attributes of pages and sites have the strongest association with ranking highly in Google. The study consists of two parts: a survey of professional SEOs and a large correlation study.
This year, with the help of Moz's data scientist Dr. Matt Peters, new data partners, and over 150 search marketing professionals, we were able to study more data points than in any year past. All together, we measured over 170 correlations and collected over 15,000 data points from our panel of SEO experts.
Ready to dig in?
We want to especially thank our data partners. SimilarWeb, Ahrefs, and DomainTools each gave us unparallelled access and their data was essential to helping make this study a success. It's amazing and wonderful when different companies—even competitors—can come together for the advancement of knowledge.
You can see all of our findings within the study now. In the coming days and weeks we'll dive into deeper analysis as to what we can learn from these correlations.
Search Engine Ranking Correlation Study
Moz's Ranking Correlation Study measures which attributes of pages and websites are associated with higher rankings in Google's search results. This means we look at characteristics such as:
- Keyword usage
- Page load speed
- Anchor text
- ...and over 170 other attributes
To be clear, the study doesn't tell us if Google actually uses these attributes in its core ranking algorithm. Instead, it shows which features of pages and sites are most associated with higher rankings. It's a fine, but important, distinction.
While correlation studies can't prove or disprove which attributes Google considers in its algorithm, it does provide valuable hints. In fact, many would argue that correlation studies are even more important than causation when working with today's increasingly complex algorithms.
For the study, Dr. Peters examined the top 50 Google results of 16,521 search queries, resulting in over 700,000 unique URLs. You can read about the full methodology here.
Here's a sample of our findings:
Example: Page-Level Link-Based Features
The features in the chart below describe link metrics to the individual ranking page (such as number of links, PageRank, etc.) and their correlation to higher rankings in Google.
Despite rumors to the contrary, links continue to show one of the strongest associations with higher rankings out of all the features we studied. While this doesn't prove how Google uses links in its algorithm, this information combined with statements from Google and the observations of many professional marketers leads us to strongly believe that links remain hugely important for SEO.
Link-based features were only one of the features categories we examined. The complete correlation study includes 12 different categories of data.
10 Ranking Factors summary findings
- We continue to see lower correlations between on-page keyword use and rankings. This could likely be because Google is smarter about what pages mean (through related keywords, synonyms, close variants, and entities) without relying on exact keyword phrases. We believe matching user intent is of the utmost importance.
- While page length, hreflang use, and total number of links all show moderate association with Google rankings, we found that using HTTPS has a very low positive correlation. This could indicate it's the "tie-breaker" Google claims. Negatively associated factors include server response time and the total length of the URL.
- Despite rumors to the contrary, the data continues to show some of the highest correlations between Google rankings and the number of links to a given page.
- While there exists a decent correlation between exact-match domains (domains where the keyword matches the domain exactly, e.g. redwidgets.com) and rankings, this is likely due to the prominence of anchor text, keyword usage, and other signals, instead of an algorithmic bias in favor of these domains.
- Our study showed little relationship with the type of top-level domain (.com, .org, etc.) and rankings in Google.
- While not quite as strong as page-level link metrics, the overall links to a site's root and subdomain showed a reasonably strong correlation to rankings. We believe links continue to play a prominent role in Google's algorithm.
- Use of anchor text was another prominent feature of high-ranking results, with the number of unique domains linking with partial-match anchor text leading the way.
- Always controversial, the number of social shares a page accumulates tends to show a positive correlation with rankings. Although there is strong reason to believe Google doesn't use social share counts directly in its algorithm, there are many secondary SEO benefits to be gained through successful social sharing.
- Time until domain registration expiration was moderately correlated with higher rankings, while private registration showed a small negative correlation.
- Engagement metrics from SimilarWeb showed that pages with lower bounce rates, higher pageviews, and better time-on-site metrics were associated with higher rankings.
Ranking Factors Expert Survey
While correlation data can provide valuable insight into the workings of Google's algorithm, we often learn much more by gathering the collective wisdom of search marketing experts working at the top of their game.
For this reason, every two years we conduct the Ranking Factors Expert Survey.
The survey itself is famously grueling–over 100 questions covering every aspect of Google's ranking algorithm. This year, we sent the invitation-only survey to 150 industry professionals.
Stay tuned for a deeper dive into the Expert Survey later this week. We're honored to have the participation of so many knowledgeable professionals.
In the meantime, you can freely view all the findings and results right now:
Ranking Factors wouldn't be possible without the contribution of dozens of very talented people, but we'd especially like to thank Dr. Matt Peters, Kevin Engle, Rand Fishkin, Casey Coates, Trevor Klein, and Kelly Cooper for their efforts, along with our data partners and all the survey participants.
What ranking factors or correlations stand out to you? Leave your thoughts in the comments below.
Great to see that it's here Cyrus!
First, for someone who is still gaining a deeper knowledge of SEO, this is the best thing that could have happened in 2015 so far.
Second, my question is: how exactly is there a correlation between exact match domains and higher rankings that still exists? Wasn't one of the biggest factors behind Google creating the EMD Update in 2012 to try and weed out websites that were using this as a tactic? While there may be a handful of websites out there that had an EMD long before the algorithm update/change who very well may have been creating that domain in good faith, aren't there too many websites out there who are doing the exact opposite?
I'm just trying to figure out for my own understanding as I was under the impression that an exact match domain would work against you, more often than not.
Thanks for all the information, though. Really helpful!
Great question. This is where we pull out the "Correlation is not Causation" card. While Google very well may not algorithmically favor exact-match domains, but there are other forces at work which may often help the exact-match domain.
(For the initiated, an exact-match domain is where the domain exactly matches the keyword query, i.e. the domain extrawidebrownshoes.com would be an exact-match domain for the keyword search "extra wide brown shoes")
In the past, many believed that Google gave preference to exact-match domains, or that simply using the keywords in the domain was enough to gain a boost. The practice was severely abused, and as you mention, Google stated a few years back they were moving to make exact-match domains play on a more level playing field. That said, exact-match domains often possess the following features:
So although Google likely removed any stand-alone advantages to having an exact-match domain, there are many secondary reasons why we see such a strong relationship between exact-match domains and higher rankings in the real world.
The "Correlation is not Causation" card is not one I enjoy ha. But I get why Google allows for exact-match domains now. It's too easy to point to one exact-match domain website and say, "well this makes no sense why they rank so high because their content is awful." (I've come across this too often with small SEO sites) Before when the bad outweighed the good, as I've been told, purchasing an exact-match domain was the best way to try and improve ranking, especially when trying to gain a higher place in SERPs "legally". It seems that I've been led to believe that these domains should leave a bad taste in my mouth. Now I know to check, and then recheck, to see just how good an exact-match domain website may be.
Thanks for your insights!
This is an awesome study, thanks for doing all the hard work to make it possible!
It's great to see so many partners stepping up to help out, really gives the survey an extra layer of credibility. I know I'll be spending a lot of time looking at these numbers.
Thanks for including us in this amazing research! it's great honor
Thanks Roy for you own and SimilarWeb's contirbutions. It was an honor working together and we couldn't have done it without you.
Already bookmarked this post. Although there's nothing much changes, having a latest survey on your proposal is a big plus for data lovers and successful clients. :)
I knew social media has little significance in SEO but it has its own stand in Digital Marketing.
Excited to read these correlations! It's interesting the way that the search engines and SEO have changed so much, yet the ranking factors really do not seem so dissimilar to me.
Cyrus, Good Job here. I am a Big fan for your ranking Factors Posts and waiting it for 2015 a long time. ( 8 Months) :-)
Man this is better than Christmas to geek like me Thanks so much for sharing!
I lit-up one of our local Phoenix experts for saying it was a part of the algorithm. Talk about backstroking I thought he was in an Olympic qualifier ..lol and I agree with your statement about social shares. It might not be a direct benefit, but its is easy to beleive that there are ancillary benefits when it comes down to SEO.
I would like to see a breakdown of page length as far as the number of words and embedded Items like # of pic, videos, pdf's etc.
No big surprises on the list. Work on content and get those backlinks from authority websites!
Again good job and thanks>
Always nice with a fresh list thank you
Learn something new everyday - I was not aware of: Time until domain registration expiration was moderately correlated with higher rankings, while private registration showed a small negative correlation
Good stuff!
/Chenzo
Right?! I've been personally curious about the relationship between private registration and rankings for years, so I was happy that DomainTools was able to help us out with this data.
Cyrus,
I'm sure the list isn't ordered by importance, but it's great to see user intent mentioned right off the bat. I still see a lot of clients who seem open to the idea of factoring user intent into their content marketing efforts, but they still cling to ideas of keywords being all-important. Ironically, these same folks often ask "How does the competition outrank me for my main keywords, and they aren't even using my exact match keywords?" Hmm...
RS
Thanks! To think I use to work for somebody who is STILL deploying the "flood keywords all over the page strategy"... Just makes me wonder how far some people will go for "quick" results, even in 2015.
Yeas! Link is still main factor of SEO
In fact, it's almost impossible to rank with links.
This is fine ... I think some updation taken maximum time during moz crawl report ,,,,,,,,,,,,,,,,,,,,,,,
Hey Cyrus, perhaps you and the team would be interested in my unscientific take on the matter of ranking factors?
https://crankedseo.com/google-ranking-factors-intro/
Interesting work - it's obvious you've put a lot of thought into these factors and how they apply to actual SEO services. I also appreciate how you flat out state that some of your analysis is based on guesswork, as these things are usually open to debate+discussion. Thanks for sharing.
Thanks Cyrus, lets face it: anyone stating that they know for 100% sure is not worth listening to or hiring, because that person would be closed to the need to adapt and change their plan of attack on the basis of new discoveries.
In my opinion Google does not take social links under consideration but can take social users experience based on their behavior on the website.
In addition, by signals of social media the value of the website increasing, making it more resistant to any filters and punishments.
I greatly appreciate the Page-Level Keyword Usage Features section. It is helpful to remember that there are several ways to grade and analyze a page and chances are that using more than one method will offer different insights which is like gold at reporting time. Thank you Cyrus and team.
Good read.... I would bet that speed of site is going to get more important as well with he growing mobile segment.
Hi Cyrus, You have done nice job. I have one question : I am trying to get Ranking for the keyword 'Perlen' in Germany but i am amazed that after implementing everything we are not in google searches even and our competitors are in top positions. They don't even have sufficient content which is 200-400 words.
My page was cached on 25 June and then on 2nd July. Although we are getting good ranking from other keywords but not from Perlen and my client first ask about that one keyword.
I hope we will soon get somewhere in searches.... Any suggestions appreciated :)..
Thanks for this heavy work!
Regarding h1 characteristics and how it can influence rankings, in "Page-Level Keyword Usage Features" section there is a relatively hight correlation of 0.09 for "LDA similarity between H1 tag and query". In contrast, the "# of keyword matches in h1 tags" correlation to ranking is very low (0.02).
So what exactly does LDA similarity mean? In investigating I found that LDA similarity is sometimes used as method for measuring similarity between questions and answers. I also found that "Latent Dirichlet Allocation is an unsupervised semantics model that takes a corpus of documents--in this case, the blog posts--and estimates what "topics" they are about. A "topic" is a set of word frequencies ..."
So in result can we say that ...
?
Thanks for your help in getting into H1 ranking power ...
Question: Trying to gain traction for Structure Studios, What do you pros think is best way to get links to your website?
Hi there! This would be a great question for our Q&A forum. :) You might also check out our Beginner's Guide to Link Building.
Thanks Matt! will do.
I am relative new to SEO but this is awesome. Thanks Cyrus
Very impressive ranking factors of 2015. Its like search engines have made changes so much when it comes to ranking specially anchor part!
Thank you for the awesome study! I was expecting link features to be at the top of the list, but I'm happy that content-based features & engagement are also extremely influential.
Ahhh... thank you! I've had a bit of trouble finding stats that relate to the relationship with the type of top-level domain (.com, .org, etc.) and rankings in Google. It's interesting (and encouraging!) to know that there really doesn't seem to be a relationship.
This sentence make think about our websites "We continue to see lower correlations between on-page keyword use and rankings". It's a common mistake full up our websites with the keyword that we want to position in serach engines, but in this article our friend Cyrus explain us that it isn't a good methodology.
Even though it's 4 months old, this should still be top of everybody's Christmas reading list, (that's if you've not read it and digested it 10 times over already).
Very useful compilation of the SEO Ranking factors! I had missed this post earlier and just saw the link from the Moz Blog highlting the top posts of 2015.
Awesome study! I think that DA and PA still remains a relevant factor
Thanks You guys for this
I am so thankful for the point #1 in your summary. We constantly get clients wanting to create 'Franken-pages' to rank for so many different (but essentially the same) keywords. Now we can just focus on the core purpose of the page and let Google figure out the small keywords differences, because the user is essentially searching the same thing.
Cyrus, in the correlation study moz.com/search-ranking-factors/correlations there are several double entries showing different correlation values for same metrics ("Query-Document relevance score using ..."). Maybe you want to correct this.
Stop scaring us Cyrus with this bunch of data :)
On a serious note there are few points where i am confused?
1> if you mean "total number of links" then is it only total number of links or just "dofollow" links.
2> Why Page has link to twitter, Facebook, twitter card, Search Action Schema, URL has a hyphen. is all in negative? Is there no use of having them in the website?
Definitely Backlinks and Social Media keep working well with high rankings.
Thanks for sharing this awesome work and keep us update about which factors should be consider to improve our Website.
We will share this information to our audience.
Have a nice day!
[URL removed by editor.]
Excellent! So as expected Page Authority continues to be the king. Moving forward, mobile-friendliness, UX score, user engagement etc are going to the driving factors of page authority. At the end of the day, it's the users that are going to win.
Thanks guys for this. Just wanted to clarify the Keywords in domain name thing, is that related to "Exact match domain"? Are we seeing any positive impact of this ?
By the way, now we have enough material to revise SEO strategies for our clients :)
Spamming is another factor of Search Engine Ranking.. Do you find what is correlation with other matrix and Spamming?
Thankyou @Cyrus for sharing guide of 2015 SEO factors. Great question "Correlation is not Causation" card. I really liked the topic.
EMD is still a ranking parameter and can make a difference to your SERP. As per our experience, the real issue with exact match domains lies in Google’s problem with over-optimization and excessive on-page branding, which you have also used in article, which is appreciable.
But, If we create high quality or good content website, which have a good linking structure with quality backlinks and if we use intelligent SEO tactics to our site, then we can get higher ranking if we also have a exact match domain name for the keywords used to find it. With this, yes, Link factor is important aspect in SEO.
But @Cyrus Does social media have any bad effect on exact match domains? Does it effect the website ranking?
Thanks for it. big surprises on the list. Work on content and get those backlinks from authority websites!
I am very excited after reading the title "Ranking factor". I am waiting for your next post "Expert Survey" and I am sure we will get the chance to learn something new and advance.
Thanks to share.
Great to see it is Cyrus and love the analytic elements behind it! I <3 data!
The ratings factor now is wonderful . Thank you for sharing
I am surprised that the Social and Brand factors section of the study and survey did not attempt to measure the impact of press or news mentions.
Great to read and to keep in mind this study. Always good to know the biannual time of Ranking Factors is here again. Don't think this should be a strategy changer. But it is always good to have a source where we can trust on because the sample with respected professionals in the field.
I would love to see the opinion of the Google people about this one. Maybe they are reading this and looking at each other that this study is really close to the actual algorithm. That would be really cool.
This is a worthy gift for all SEO Professionals. As being in SEO Industry, we wonder every time to get the proper information, this article is just beyond the expectations. A very very helpful and effectively done research..
its really a comprehensive study. but i would have to loved to see one more metric up there, i.e. the relevancy of a particular page with a search query
i did a few test of my own regarding this and found that relevancy is a very potent ranking factor
thoughts?
Thank You Cyrus Shepard for Search Engine Ranking Correlation Study:)
Good Job Cyrus. Good link building is still the order of the day
A special Gift for all the guys girls working in SEO field. Thumps Up for this.
Already I still see a lot of owners Aspam buy backlinks forefront of Google results
Great stuff and LOTS of data to review!
Thanks Cyrus! Great post!
A few surprises and a few reminders in this data. Thank you.
Nice sharing!!!
up