Searchmetrics seo-ranking-factors-2013-uk

White paper: “SEO Ranking Factors – Rank Correlation 2013” © Searchmetrics – Page 1
WHITE PAPER
SEO Ranking Factors –
Rank Correlation 2013
- Google UK -
Executive summary:
This white paper deals with the definition and evaluation of factors that have
high rank correlation coefficients with top organic search results and aims to
provide a more in-depth analysis of search engine algorithms. The data collection
and its evaluation refer to Ranking Factors and Rank Correlation for Google.co.uk
in 2013.
Authors: Marcus Tober, Dr. Leonhard Hennig, Daniel Furch

About Searchmetrics and the Searchmetrics
SuiteTM
Searchmetrics is a pioneer and the world's leading provider of search and social
analytics software for companies and online agencies.
Our software, Searchmetrics SuiteTM
, empowers users to increase visibility,
market share, revenue and profits by applying optimal strategies in the digital
marketing world.
The main advantages at a glance:
 Identification of competitors and their strategies
 Optimization of the online marketing mix
 Increase in the company's ROI
 Largest international and local coverage in the world, with 124
countries and 155 local rankings, allowing for the optimization of global or
region-specific search campaigns
 Monitoring of brand names in 268 search engine/country
combinations worldwide
 ROI analyses: Discovering untapped potential by displaying keywords
with the corresponding search volumes and CPCs
Would you like to find out more about Searchmetrics SuiteTM
or request a quote?
Contact us now!

Contents
FINDINGS OVERVIEW...................................................................................7
Structure of the white paper ........................................................................ 7
DEFINITIONS, INFLUENCING FACTORS, AND THE DATA POOL.................................8
What is a Ranking Factor? ........................................................................... 8
Ranking Factors: Correlation ≠ causation ...................................................... 9
Influencing factors.................................................................................... 10
Data freshness......................................................................................... 10
Data pool ................................................................................................ 10
Interesting background facts ..................................................................... 11
The correlation coefficients at a glance........................................................ 12
1. ON-PAGE CODING..............................................................................14
1.1 On-page coding: Developments since last year....................................... 15
1.2 Keyword domains are no longer a Ranking Factor ................................... 16
1.3 Presence of keywords in the description, title and H1 remains important ... 20
1.4 WDF*P*IDF ....................................................................................... 21
1.5 Keywords in the title should be placed as close to the front as possible ..... 23
1.6 Page load times as an important performance factor ............................... 23
1.7 DIGRESSION: Authorship integration does not (yet) have an effect .......... 24
1.8 Summary........................................................................................... 26
2. ON-PAGE CONTENT............................................................................28
2.1 On-page content: Developments since last year ..................................... 29
2.2 Higher number of words in the text is ‒ in contrast to the previous year ‒
clearly important...................................................................................... 31
2.3 HTML length is important but only to a certain extent.............................. 32
2.4 Text length is relevant only up to a certain point .................................... 33
2.5 Using media to enrich content pays off .................................................. 34
2.6 Internal links: Number, structure and the text of links optimize link juice .. 35
2.7 Moderate integration of AdSense and other Adlinks is less negative than
before..................................................................................................... 38
2.8 Summary........................................................................................... 41

3. BACKLINKS........................................................................................42
3.1 Backlinks: Developments since last year................................................ 43
3.2 The number of backlinks remains immensely important........................... 44
3.3 Link diversity is always relevant ........................................................... 44
3.4 "Hard" keyword anchor text has lost its relevance................................... 46
3.5 Backlinks from URLs with high SEO Visibility are significantly more valuable
.............................................................................................................. 48
3.6 Summary........................................................................................... 50
4. SOCIAL SIGNALS ...............................................................................51
4.1 Social Signals: Developments since last year.......................................... 52
4.2 Social signals continue to show strong correlation with good rankings ....... 52
4.3 Google+ new in the analysis ‒ with very high correlation......................... 55
4.4 Digression: Google+ will overtake Facebook at the beginning of 2016 ....... 56
4.5 Summary........................................................................................... 58
5. THE BRAND FACTOR...........................................................................60
5.1 High impact of the brand factor on SERPs .............................................. 60
5.2 The top ranks are reserved for brands................................................... 61
6. THE SPECIAL ROLE OF WIKIPEDIA.....................................................62
6.1 Wikipedia results affect correlations ...................................................... 62
6.2 Rank 17 and 27 "reserved" for Wikipedia results..................................... 64
SUMMARY OF SEARCHMETRICS SEO RANKING FACTORS – RANK CORRELATION 2013
..............................................................................................................65
Summary of most important developments in Ranking Factors – Rank Correlation
2013: ..................................................................................................... 66
ABOUT SEARCHMETRICS .............................................................................67

List of Figures
Figure 1: Searchmetrics ‒ SEO Ranking Factors – Rank Correlation 2013 ........ 12
Figure 2: On-page coding ‒ Rank Correlation 2013....................................... 14
Figure 3: On-page coding 2013 ‒ Changes since 2012 .................................. 15
Figure 4: Correlation ‒ Keyword in Domain / URL......................................... 16
Figure 5: Screenshot ‒ Losers at keyword level for an example domain after
Penguin update........................................................................................ 17
Figure 6: Correlations ‒ Existence of on-page keywords / description, H1 and H2
.............................................................................................................. 20
Figure 7: Average by ranking ‒ Similarity.................................................... 22
Figure 8: Correlations ‒ Position of the keyword in the title ........................... 23
Figure 9: Average by ranking ‒ Page load time ............................................ 24
Figure 10: Average by ranking ‒ Presence of authorship integration ............... 26
Figure 11: On-page content ‒ Rank Correlation 2013.................................... 29
Figure 12: On-page content 2013 ‒ Changes since 2012 ............................... 30
Figure 13: Correlations ‒ Content quality .................................................... 31
Figure 14: Average by ranking ‒ Word count in text ..................................... 32
Figure 15: Average by ranking ‒ HTML code length in characters ................... 33
Figure 16: Average by ranking ‒ Text length in characters ............................ 34
Figure 17: Average by ranking ‒ Image count.............................................. 35
Figure 18: Average by ranking ‒ Number of internal links ............................. 36
Figure 19: Average by ranking ‒ Number of internal links pointing to the same
website ................................................................................................... 37
Figure 20: Correlations ‒ Ad integration...................................................... 38
Figure 21: Average by ranking ‒ AdSense ................................................... 39
Figure 22: Average by ranking ‒ Adlinks total (including AdSense)................. 39
Figure 23: Average by ranking ‒ Number of AdSense ads.............................. 40
Figure 24: Backlinks ‒ Rank Correlation 2013 .............................................. 42
Figure 25: Backlinks 2013 ‒ Changes since 2012 ......................................... 43
Figure 26: Average by ranking ‒ Number of backlinks................................... 44
Figure 27: Average by ranking ‒ number of words in anchor text ................... 45
Figure 28: Average by ranking ‒ Percentage of nofollow backlinks.................. 46
Figure 29: Average by ranking ‒ percentage of backlinks with keyword .......... 47
Figure 30: Average by ranking ‒ Percentage of backlinks with stop word......... 48
Figure 31: Average by ranking ‒ Average SEO visibility of the link source ....... 49
Figure 32: Social signals ‒ Rank Correlation 2013 ........................................ 51
Figure 33: Social signals 2013 ‒ Changes since 2012.................................... 52
Figure 34: Average by ranking ‒ Facebook signals in total............................. 53
Figure 35: Average by ranking ‒ Twitter signals........................................... 54
Figure 36: Average by ranking ‒ Pinterest signals ........................................ 55
Figure 37: Average by ranking ‒ Google+ signals......................................... 56
Figure 38: Percentage growth and absolute number of Facebook shares vs plus
ones ....................................................................................................... 57

Figure 39: Facebook shares vs. plus ones ‒ growth forecasts in % ................. 58
Figure 40: Average by ranking ‒ H1 title tag................................................ 60
Figure 41: Wikipedia correlation................................................................. 62
Figure 42: Average ranking ‒ Wikipedia URL................................................ 63
Figure 43: SEO Ranking Factors – Rank Correlation 2013 – Changes since 2012
.............................................................................................................. 65

Findings Overview
On-page factors such as the technical side structure and good content are basic
requirements for good rankings. The keyword on-page is important, but seems
to have been devalued by Google in its relevance as a domain name, as a part of
the URL and as the link text.
In contrast, the number and quality of backlinks are probably very important ‒
"natural link profile" and diversity are the keywords here. Social signals continue
to correlate positively with better rankings.
Summary of Searchmetrics SEO Ranking Factors – Rank Correlation 2013
Structure of the white paper
The study is focused, on one hand, steps involved in building a website in
practice and on the other on the relevance of influencing factors, which has
increased over time.
The study first examines the on-page area ‒ broken down into two thematically
different blocks, coding and content. This is followed by the analysis of the off-
page area in the form of backlinks and social signals. The white paper concludes
with the analysis of two factors that play a special role in this study and
therefore have to be examined separately. The final summary of the findings of
the study follows the relevant chapters on the brand factor and the role of
Wikipedia.
The chapters first present an overview of all factors influencing the area
concerned followed by the trends in the development of factors in 2013
examined in relation to the previous year. At the core of every chapter are the
remarks and conclusions on the individual factors, followed by a chapter
summary at the end of each chapter.
In addition, the impact of correlations on a combination of factors is discussed in
individual chapters.

Definitions, influencing factors, and the data
pool
What is a Ranking Factor?
Search engines use algorithms to rank websites by subject and relevance. All the
websites in a search engine index are categorized based on this structure, which
finally results in an optimal ranking for user’s search queries. In general, the
criteria for evaluating websites and performing this ranking are referred to as
Ranking Factors.
The reasons for this are obvious: Due to the exponential rise in the number of
documents on the Internet ‒ and thus in the search index ‒ it would not be
possible to rank websites without automatic algorithms despite the existence of
human quality raters. While, on the one hand, these algorithms are trivial (order
requires a pattern), they are also the best-kept secret in the Internet business,
because it is essential for search engine operators to keep the underlying factors
influencing the algorithm strictly confidential.
The reason for this is elementary rather than competition-related: If the
definition and the degree of influence of factors relevant for a good ranking were
known, it would render them at the same time irrelevant – because then they
could be manipulated.
At the beginning of the search engine era, Google would rank websites on certain
topics as relevant based on the high keyword density on these topics. However,
webmasters used this knowledge to climb up the SERP ranking by using keyword
overloaded pages that resulted in irrelevant searches1
.
This fact has not only triggered real competition between search engines and
website operators, but also the mystery surrounding the Ranking Factors. Aimed
at developing semantic search, a set of criteria was devised, which was initially
strictly technical (e.g. number of backlinks), but in the meantime also contains
less technical components (e.g. user signals).
This fact, combined with the pursuit of optimal search results, culminated in the
continual evolution of Ranking Factors, whose categorization and complexity due
to ongoing iterative update cycles under the strong influence of user signals is
subject to constant optimizations aimed at producing the most relevant search
results for users.
For website operators this means a steadily decreasing number of factors that
can be negatively manipulated, on the one hand, and the opportunity ‒ given the
increasing efforts to combat spam, irrelevance and short-termism ‒ to achieve
1
Search Engine Results Pages

high search result rankings in the long term using a sustainable business
strategy based on relevant quality factors, on the other.
Ranking Factors: Correlation ≠ causation
We are not Google. The analysis and evaluation of Ranking Factors using this
data not only has interpretation value but, in fact, represents a profound
interpretation (and thus not a mere conjecture) on the basis of facts, namely the
evaluation and categorization of website features that achieve high ranking in
search results.
We at Searchmetrics aggregate billions of data points every month and naturally
look for the answers to the question: Which factors are relevant for a good
ranking in Google's search results?
To this end, we compare website characteristics with their Google rankings and
derive a structured list of Ranking Factors. For example, if there are many pages
in the top positions of analyzed SERPs that contain a keyword in the title tag,
whereas fewer low ranking pages have that feature, then we have identified a
high correlation with good rankings. On the basis of these correlations we can
then draw conclusions about the characteristics of websites that rank well.
However, it is necessary to point out that correlation is not synonymous with
causation, and thus there is no guarantee that the relevant factors actually affect
rankings or are even used by Google as signals. For this reason, Searchmetrics
conducts studies on a regular basis which analyze, for example, the impact of
social signals under isolated conditions. As a result of this, the impact of social
signals, among other factors, on rankings has already been proven.
Nonetheless, the data speaks for itself, unlike the definitive Ranking Factors and
their degree of influence, this data is not a secret. As a result, our analyses allow
us not only to assess what factors will most likely be included in the ranking, but
also the degree of their influence. Many observations, which were previously not
based on much more than a gut feeling, can now be confirmed with the help of
the following data.
As already mentioned, there are purely technical or "hard," as well as less
technical or "soft" Ranking Factors. In order to optimize the visual impact of our
charts, we have grouped our factors ‒ even if some areas partly overlap - into
social, backlinks, on-page coding and on-page content.
As the term "Ranking Factor" and its causal elements of meaning can lead to
problematic interpretations, we prefer this term to be rather interpreted as rank
correlation coefficient in the context of our analysis and this white paper.

Influencing factors
The position and order of search results depend not only on the entered search
queries (= keywords), but also personal search history, which in turn is
determined by the behavior of the user. In addition, it depends on factors such
as: Is the user logged into Google? Does the user use Google's own Chrome
browser to surf the Internet? Where is the IP of the user located? Are cookies
containing personal information stored on the user's computer, etc.? Last but not
least, Google itself is changing the variations of these integration types all the
time.
Since the consideration of these many individual factors would make the analysis
excessively complex, this white paper focuses on the variations of the integration
types introduced by Google itself on a basis that is as neutral and national as
possible.
Data freshness
The data was first collected and aggregated in March 2013 and then again, in
June 20132
. As this is a follow-up study to the 2012 Ranking Factor analysis, the
changes of many factors compared with the previous year are particularly
interesting. They provide invaluable insights into the evolution of Ranking
Factors, which can be used to derive trends for future development and the
relevance of the criteria. The differences compared with the previous year are
considerable in some areas.
In addition, the pool of analyzed characteristics was expanded to include many
factors that allow more in-depth analyses of existing factors or shed light on new
fields.
Data pool
These analyses are based on search results for a very large keyword set of
10,000 search terms from Google UK. We have largely excluded specific
navigational keywords (as it was the case last year) from the original data pool,
which included the top 10,000 keywords according to search volume, as we did
not want to distort the analysis. Search queries are considered to be navigational
keywords if they return results, which are largely irrelevant and only return one
exact match (for example: "Facebook Login").
The first three pages of organic search results were always used as data pool.
This keyword set covers approximately 90 to 95 percent of the data pool used in
the study last year. Here, we attempted to strike a balance between two factors,
2
Second data point: after the Google update "Penguin 2.0" in the 22nd calendar week

namely the preservation of the "greatest common denominator" as an optimal
basis for comparison with last year's analysis, and the consideration of new
keywords that have emerged due to growth in search volume in the top 10,000.
The data pool at Searchmetrics is always kept up-to-date. Therefore, new,
relevant keywords are used in current studies, such as, for example "Samsung
Galaxy S4" or "iPhone 5," which did not even exist at the time of the 2012 study.
Interesting background facts
The analysis of 10,000 keywords has led to:
 30,000 SERPs (search engine result pages)
 300,000 titles, descriptions, URLs, etc.
 Approx. 150 GB of data
 Approx. 600,000 AdSense blocks
 Approx. 5,300,000,000 backlinks
 Approx. 4,150,000,000 Facebook shares
 Approx. 12,950,000,000 Facebook likes
 Approx. 600,000,000 Facebook comments
 Approx. 1,000,000,000 tweets
 Approx. 330,000,000 Google plus ones
 Approx. 14,500,000 Pins

The correlation coefficients at a glance
As was already the case with the previous analyses on the subject of "Ranking
Factors", the presentation of the correlation between Google's search results and
the various influencing factors in this white paper is also based on Spearman
correlation. In comparison to 2012, significantly more factors have been taken
into account ‒ the most important are shown in the following bar chart. Further
factors will be discussed in the respective chapters. The cluster color coding
introduced in the previous year was used again:
Figure 1: Searchmetrics ‒ SEO Ranking Factors – Rank Correlation 2013, Google UK

Legend: The correlation coefficient is plotted on the x-axis, which measures its
size. The longer/shorter the bar in the positive part of the x-axis, the
higher/lower the correlation between that particular factor and a good Google
ranking. Our analysis has shown that factors with a correlation factor of zero do
not have any measurable correlation with Google's results. Values in the negative
range can be positively interpreted with the opposite statement.

1. ON-PAGE CODING
The entire on-page area absolutely belongs to the basics, especially with respect
to the technical parameters of a website and the factors affecting a good search
engine ranking. It has been the case for some time now that fulfilling certain on-
page criteria does not have (any longer) a positive effect on the website’s
ranking. On the other hand, if a website does not meet these criteria, this will
turn out to be very bad for its ranking.
Figure 2: On-page coding ‒ Rank Correlation 2013
Nearly all websites ranked among the top 30 are compatible with the
components mentioned later in this chapter. In the case of binary features, in
particular, it is hardly imaginable for them to have a positive influence in the
presence of factors vs. negative influence in the absence of factors, when in the
group of examined elements, the share of elements lacking the analyzed features
is so low ‒ i.e.: If almost all analyzed websites meet a specific criterion, a high,
positive rank correlation coefficient cannot be logically calculated.

Instead, it appears that on-page coding factors, in particular, represent in many
ways the basic prerequisite for a high search result ranking.
In contrast to our results from the previous year, this is no longer related to a
domain name or the URL on a keyword basis.
Two of the newly introduced factors, however, clearly show a certain level of
triviality: "URL length" and "URL is not a subdomain". In principle, these two
features, which are very closely intertwined, tell us that shorter URLs and
domains tend to rank higher on Google. This can be logically explained by the
fact that on average across all rankings, the homepage of a domain performs
better than individual subdomains or other subpages of this domain.
1.1 On-page coding: Developments since last year
The following detailed view shows the development of the correlation of
individual factors since 2012. Features that were included in the study for the
first time and for which there is no basis for comparison have been grayed out.
Figure 3: On-page coding 2013 ‒ Changes since 2012

The above chart clearly shows that all but two on-page coding factors display a
markedly higher correlation with good rankings than was the case last year.
While the relevance of keywords on the page itself, as well as the relevance of
the keyword position in the title (the closer to the front, the better) have
increased considerably, factors, which used to have a pronounced positive effect
on rankings such as keywords in the domain name or the URL, have lost
significance.
1.2 Keyword domains are no longer a Ranking Factor
In 2012, the situation was different:
"The power of keyword domains has been known for years and is still
clearly visible in our analysis."3
Those days are now over ‒ the once positive effect of keywords in the URL and,
in particular, in the domain name has decreased considerably. As a result, the
quality of the keyword in the URL/domain as a Ranking Factor is no longer
relevant. The following chart clearly shows the sharp drop in correlations for
these two features within one year.
Figure 4: Correlation ‒ Keyword in Domain / URL
The summary chart, which compares the Ranking Factors in 2012 vs. 20134
also
shows that these two factors are the biggest losers emerging from this analysis
in comparison to the previous year.
This is mainly due to two algorithm changes by Google that were made during
2012:
3
Google Ranking Factors 2012, Searchmetrics (Chapter 5)
4
See Chart 43

The Penguin update, which Google rolled out first in Europe (DE, UK) in
March/April and shortly afterwards in the US, was aimed mainly at combating
web spam, i.e. websites with low-quality backlinks5
, i.e. domains with links from
low-quality sources "link farms" / link networks and – in the broadest sense –
"unnatural" (forced) link building.
Accordingly, there was a massive decrease in the relevance of "hard" keyword
links, i.e. backlinks with nothing but the keyword in the anchor text. As a result,
the losers of this web spam update tend to drop more sharply in the rankings on
a pure keyword level, which, in turn, affects the whole domain through the
presence of main keywords in the domain name. For example:
Figure 5: Screenshot ‒ Losers at keyword level for an example domain after Penguin update
In this screenshot from “Keyword rankings: Losers of the week" from the
Searchmetrics SuiteTM
, we can clearly see a rather drastic loss of rankings at the
keyword level in the "Trend" column, which naturally has an impact on the
overall performance of the whole website. The domain, from which the data was
5
See Chapter 3, Backlinks

collected, is one of the websites, which is about to recover from the impact of the
update with the help of new link strategies.
In 2012, Google released two minor updates to Penguin, which was followed by a
major update of the algorithm in May 2013. This was launched as Penguin 2.0
and is the second generation of this "over-optimization penalty". This algorithm
change occurred after the initial data collection was completed, which is why
most of the data had to be updated again. While the effect on the correlation was
minor, there was a shift in several results that are only visible when large
amounts of data are analyzed (see Wikipedia). Google has already announced
further Penguin updates.
The negative developments of the factors "keyword in domain" and "keyword in
URL" are also due to the so-called "exact match domain update" (EMD Update).
The EMD Update was rolled out by Google in September 2012 and has currently
only been implemented in the US – it has not yet been iterated.
In contrast to the Penguin update, which is mainly link-based and works at a
keyword level and rather indirectly via the off-page detour in the form of the
backlinking source (or its significance as well as the quality of its links), the EMD
Update algorithms, while also running at a keyword level, work directly on-page
and are URL-based and thus focus directly on keyword domains.
Interestingly, Google only seems to have rolled out the EMD update in the US.
No algorithm change has been recorded outside google.com to date. The reason
may be perfectly obvious: In other countries ‒ for example, in Germany and also
in the UK or France ‒ the ranking of most of the suspicious domains has already
been successfully degraded in the course of the Penguin update, and therefore,
another algorithm change may not appear necessary in this case.
However, one question still remains: Why is the correlation for the factor
"keyword in domain" still positive? Here, too, the answer is not far away:
brands.
The special role of brands in the Google ranking already stood out in last year's
study. Search engines do not seem to apply the same standards to brands - and
their websites - as they do to other domains. The brand factor will play a bigger
role in the course of this study.
The fact is that brands – or the related websites ‒ are often quite naturally linked
with the name of the brand in the link text alone. What is negative for non-
brands, is simply normal for brands. Google is aware of this and treats brands
accordingly by not rating this factor as negative in the case of well-known
brands, which then rank in the top positions. As a result, there are domains for
which the feature "keyword in domain" appears to be positive based on rankings
(although other factors, i.e. brand factors ‒ play a key role here), which is the
reason behind the still marginally positive correlation of these features.

The following section discusses several factors, which could prove that it is not
particularly easy for Google to rank the field behind the brands as the tried and
tested on-page methods appear to play again a more significant role here.

1.3 Presence of keywords in the description, title and H1
remains important
Even if the "keyword" factor ‒ as will also be shown later ‒ does have a
considerable negative effect in the case of over-optimization on many levels,
there are some on-page areas, in which the presence of keywords does not show
an above-average correlation with good rankings ‒ because almost all top 30
websites ‒ except for a few brands ‒ have these features. In terms of on-page
optimization, keywords remain very important.
The mere presence of a description, H1 and H2 is also important ‒ irrespective of
the keyword. Most of these factors even show a higher correlation with better
rankings than 2012.
For reasons of clarity, the content factor discussed in the next chapter is included
in the following chart, which focuses on these areas.
Figure 6: Correlations ‒ Existence of on-page keywords / description, H1 and H2
One cannot deny this development’s certain logic. Google is a machine that
categorizes data by its relevance to specific search queries. While Google is well
on its way to its ideal "semantic search", the process of analyzing content (i.e.
words/text) according to semantics (i.e. meaning) probably represents the
highest achievement in decoding.
The days of "keyword density" are clearly gone. However, with the exception of
the brand factor (which once again has a role in this context), it is ‒ and it has
always been ‒ worthwhile to incorporate keywords on-page in the description,
the title ‒ and also in the H1 title. These factors are comparable with features of
a book, for example. If the title and the description are not attractive, no one will

read it. It is difficult to imagine a book without a description, chapters or
headings.
The fact that not only its mere existence, but also the position of a keyword
seems to play a role in these lateral areas will be discussed in the section after
the digression on the relevance of the so-called WDF*P*IDF formula.
1.4 WDF*P*IDF
As already mentioned, "keyword density" as an isolated factor has lost its
relevance. In recent times, however, the influence of the so-called WDF*P*IDF
formula has been often discussed, which describes the principle of similarity of a
query to the page content.
This approach focuses on the correlation of two factors, namely the "number of
instances of a keyword in a document in relation to all other subjects "within-
document-frequency" (=WDF) and the "number of instances of a keyword in all
other relevant documents in the underlying database or inverse-document-
frequency (in this case: the Google index)" (=IDF)6
.
In short: The relevance of a term to a document, and in the second stage, the
relevance of a term to a search query using a keyword, should, according to this
approach, be weighted with this term/keyword in the index not only in relation to
the individual website, but more importantly in relation to all other websites.
The correlation for this factor was also calculated using our data pool for the top
30. Only the content of the website has been taken into account in this case. On
the y-axis of the following chart, the average similarity of keywords related to
the document in the respective search result ranking is presented on a scale
from 0 = "least similar” to 1 = "identical."
6
http://www.karlkratz.de/onlinemarketing-blog/seo-keyword-density/

Figure 7: Average by ranking ‒ Similarity
There is almost no correlation ‒ the average is 0.09. On average, the results
ranked in the first position seem to exhibit a greater similarity to the search
query.
This result suggests that taking the WDF*P*IDF formula for a term, for which an
optimized website is already ranked, into account does not have a positive
impact on the ranking. Even if the URLs are included in the search results, the
ranking for that specific term will not be improved on that basis alone. Instead,
the website will more likely improve its search engine performance thanks to the
additional use of synonyms and subject-related, relevant terms by getting
ranked for other keywords as well.
The relevance of a document in relation to the search query as well as the
relevance of all other documents to this term, therefore, does not correlate
positively with better rankings, because this factor also appears to be among the
trivial prerequisites for a website to be ranked.
In addition to term frequency, keyword proximity and semantic indexing is more
important for search engines.

1.5 Keywords in the title should be placed as close to the
front as possible
The fact that the presence of keywords in the website title as well as in the
description shows good correlation, which has even increased compared with the
previous year, has already been proven. Moreover, the position of the keywords
in the title itself once again seems to play a greater role in ranking.
The following chart shows the Spearman correlations for the values "position of
the keyword" and in relation to the position of the word as well as the characters
‒ according to the formula the values are to be interpreted as the closer to the
front, the better:
Figure 8: Correlations ‒ Position of the keyword in the title
The positive value for both correlations, which are also closely related, has even
increased since last year. Accordingly, many websites at the top of the search
results have keywords placed as close to the front of the title as possible.
1.6 Page load times as an important performance factor
“Site speed” is a highly-technical feature which was not included in last year's
analysis. This factor was included in the pool of correlation factors for 2013 in the
expectation that it would show a very positive correlation ‒ but as the summary
chart shows, the correlation is only marginally positive.
This may be related to the fact that fast page load times appear to be an utterly
trivial precondition of achieving decent rankings on Google similar to the on-page
phenomenon mentioned at the beginning of this chapter.

The truth of the matter is that Google measures site speed both directly ‒ when
crawling websites ‒ and indirectly ‒ via user signals such as CTR7
and bounce
rates. Therefore, it is highly likely also a ranking criterion for Google.
The detailed view clearly shows that all the top 30 websites have, in fact, very
fast load times and the websites that rank in the first positions load by up to
one-tenth of a second faster on average.
Figure 9: Average by ranking ‒ Page load time
In this context, it should be noted that the factor "site speed" alone only exhibits
a marginally positive correlation with good rankings. This is likely due to the fact
that the top 30 websites tend to have comparably fast page load times.
1.7 DIGRESSION: Authorship integration does not (yet)
have an effect
A technical measure, which appears to gain in importance both on-page and off-
page ‒ namely on the social-media level ‒ is authorship integration. Therefore,
we will start this section with a short digression on this subject.
7
Click Through Rate

Google has made it possible for quite some time to connect URLs with user
profiles on Google's own social network Google+ by using the rel="author" tag.
This is intended, to assign content in a network clearly to an author, and thus to
verify the authorship to some extent. The implementation of this measure by
website operators means an immense user assistance in Google's mission to
categorize the vast amount of information available on the Internet ‒ and
ultimately in the Google index.
Finally, by having authors identify themselves, Google can immediately draw
conclusions about the author's popularity and relevance for certain search
queries with the help of information about the author received through the
verification process from the social chart (connections to other users, activity
rate, recognition, status, etc.), as well as relations to other publications and
activities of the profile owner in connection with metrics of the website hosting
the content itself, as well as user signals from the networks and click behavior
from SERPs.
On this basis, the factor "URL has authorship integration" was expected to
correlate well with good search result ranking for this URL. This, however, is not
the case ‒ at least not to a measurable extent. Other, more in-depth studies on
the effect of author integration on rankings conducted by Searchmetrics have so
far not delivered any truly positive or negative results.
However, even if there is no direct proof for the positive influence of authorship
on the performance of search engines, Google's "authorship integration" will
likely become a very relevant factor in the categorization and evaluation of
content in specialist areas and niche subjects. When people stand by their
content with their name and their reputation, which they may have built over a
long period of time, this represents one of the most trusted factors of content
verification for search engines, which is also very difficult to manipulate. At the
same time, the proportion of spam has fallen in SERPs.
The following chart shows the distribution of URLs with author integration in
SERPs. The y-axis shows the percentage of the analyzed URLs with authorship
integration, which rank at the respective positions (x-axis).

Figure 10: Average by ranking ‒ Presence of authorship integration
As can be seen, most URLs with authorship integration rank on the first search
result page further below, on the second SERP or towards the middle/end of the
third SERP. Websites with this feature most often rank approximately between
the positions 8 and 16, or 25 and 27 respectively. However, it is much rarer for
websites with authorship integration to rank in the top two positions.
In this context, it is, however, important to note that only around 50 percent of
analyzed keywords could be included in the calculation of this correlation, which
is simply due to the fact that authorship integration is not yet that widespread.
However, it is noticeable that snippets accompanied by images with the help of
authorship integration achieve significantly better click-through rates than
snippets without an author photo ‒ whereby the click-through rates are affected
by specific image factors (brightness, background color, attractiveness, etc.).
1.8 Summary
The implementation of on-page coding factors is largely trivial and thus largely a
basic prerequisite of achieving top rankings on Google. A well-coded website
structure and performance belong to the absolute SEO basics.
With the updates over the past 15 months, Google has sharply discounted the
factor "keyword in domain" and thus keyword domains as ranking factors ‒ the

on-page presence of the keyword in the title and the description, however,
continues to be very important ‒ as well as the presence of description, H1 and
H2. In addition, fast pages load times, as well as the relevance of a document for
a keyword in relation to all other documents in the index that are relevant for
this keyword, are among the basic requirements for good rankings.

2. ON-PAGE CONTENT
Similar to the coding of their own websites, webmasters can influence the
content area, as well. However, these two areas differ in the case of search
engines to the extent that technical factors are naturally much easier to crawl for
the Google algorithm and thus to categorize8
.
Content features, however, are often complex and move within a scale with
respect to which it has to be determined ‒ from Google's perspective ‒ which
value range should be classified as "good" or relevant. After all, it is the content
and not the technical implementation that Google wants to evaluate and rate
according to its quality and relevance to return a search result.
Even in the case of content features, the number of factors analyzed has
increased significantly compared with the previous year.
8
This is partly due to the fact that most of the factors are binary (presence: yes/no).

Figure 11: On-page content ‒ Rank Correlation 2013
In principle, the content factors can be broken down into three groups: content
quality, linking and advertising.
Except for three values from these areas, all content factors correlate positively
with good rankings. In the overview of the year-to-year trends, the increase in
relevance of website content becomes apparent.
2.1 On-page content: Developments since last year
The following chart shows the development of the content features that were
included in this study in the previous year.

Figure 12: On-page content 2013 ‒ Changes since 2012
It is noticeable that virtually all factors appear to have been upgraded ‒ some
quite markedly. Particularly impressive is the development of the factor "word
count".
The integration of advertising on the website does not seem to be as negative as
in the previous year.
The only factor showing a (slightly) negative correlation, next to “keyword in H1”
(which is a trivial factor), is "title character length." Nevertheless, this can be
again interpreted positively: the shorter the page title of a document, the better
the ranking.

2.2 Higher number of words in the text is ‒ in contrast to
the previous year ‒ clearly important
While in 2012, the feature "word count" did still correlate negatively with good
rankings, this has changed considerably in 2013.
Figure 13: Correlations ‒ Content quality
The detailed chart shows that pages that rank in top positions have a higher
word count than websites positioned at the lower end of SERPs. The y-axis shows
the average word count for websites for the respective ranks on Google (x-axis).

Figure 14: Average by ranking ‒ Word count in text
The presence of something described in this study as "brand factor" is clearly
noticeable in the first two positions. The brand factor, which is present in
virtually all rank correlation coefficients, will be subject to closer scrutiny in
Chapter 5.
In a "normal case" one would expect a continuously falling curve from position 1
(highest value) down to position 30 (lowest value). However, this is not the case
because of high ranking of brand websites, which, as already mentioned, occupy
a special position. Brand websites seem to have lower word counts on average
than search results on the first SERP. Consequently, pages ranked 1st have with
407 words on average 120 words less than pages ranked 2nd (529 words).
2.3 HTML length is important but only to a certain extent
While the length of the HTML code of a website is a technical factor in some
respects, the factor "length" in this context depends to a large extent on the type
of content, and it will therefore be examined in more detail again in this section.
The majority of websites ranked among the top 30 SERPs tend to have a certain
‒ and not inconsiderable ‒ (minimum) length of HTML code. On average, the
analysis of 300,000 HTML documents has resulted in the following correlation of
HTML length and ranking (y-axis: average length of HTML code in characters):

Figure 15: Average by ranking ‒ HTML code length in characters
Again, the brand factor figures into the top ranks. However, we can observe the
curve starting to drop beginning at position 3, which suggests that better ranked
documents usually have a longer HTML code than pages in lower positions.
Looking at this chart, it could be assumed that the more characters the HTML
code has, the better the ranking. However, this is not the case ‒ at least not
without limitation. It should be noted that the optimum is exceeded once a
certain limit of approximately 80,000 characters has been reached. While the
effect of a bigger number is not necessarily negative, the factor will not improve
the ranking results from this point onwards.
2.4 Text length is relevant only up to a certain point
The curve of the coefficient "text length" is nearly identical to the detailed chart
of the factor "HTML length". This is logical, because the units are directly
interlinked.

Figure 16: Average by ranking ‒ Text length in characters
Accordingly, the text length is also "capped" with respect to its positive
correlation with good rankings. Up to this limit ‒ and excluding the brand factor ‒
one can say, however, that websites with more text achieve a better ranking.
2.5 Using media to enrich content pays off
Images and videos seem not only to make the text more attractive for users, but
also for Google. As shown in detail in chart 13, the correlation for the number of
images on the website (the more, the better) was already positive in the
previous year, and it has increased even further since then.
The following chart shows the entire data pool for image files, i.e. all files with an
image tag on the website.

Figure 17: Average by ranking ‒ Image count
Again, it is noticeable that the higher the number of images on a website, the
better the ranking (exception again: the brand factor). It should be noted in this
context that this calculation is not limited to images that are directly embedded
in the content. In other words, images in the content as well as image files on
the website in general, seem to correlate well with a good website ranking.
2.6 Internal links: Number, structure and the text of
links optimize link juice
The internal link structure of a domain name is an important driver behind the
search engine performance of the domain. Factors such as the "number of links"
and "link text" are crucial for the optimal distribution of so-called link juice9
.
However, not only the number of links pointing to the ranked URL is relevant,
but also the link structure on the URL itself.
The number of internal links in the content of the analyzed websites is shown
below:
9
Every link has a value. This value can be optimally distributed according to the principle
of inheritance using a good internal link structure according to the hierarchical structure
of a domain. The term link juice is meant metaphorically here.

Figure 18: Average by ranking ‒ Number of internal links
The chart clearly shows that the pages in the top search result positions have a
comparably greater number of internal links than pages that rank further back.
It is also noticeable that brand websites ranked at the very top have a lower
number of links than websites ranked in the next positions on average.
The average number of links will naturally increase in the case of websites with
strong menu structures and many internal links in the footer. However, not only
the number of internal links, but also internally-linked keywords as a link text
play a role in the optimal distribution of the link juice. The current rule of thumb
suggests to use "hard" links internally (= with keyword/s) and "soft" links
externally (= generic / stop words / more words in the anchor, brand links, etc.).
This leads to an interesting question: Does a well-ranked website contain internal
links to another website for a keyword for which it is ranked? If this were the
case, it would imply, in principle that the website that receives the link would
have to be more relevant for that keyword than the website from which the link
originates. The ranked website would quasi pass on its link juice and presumably
lose its ranking sooner or later.
It is therefore interesting to examine whether websites that perform well use the
opposite principle and use the keyword to link to itself. This question is the basis
of the analytical approach presented in the following chart:

Figure 19: Average by ranking ‒ Number of internal links pointing to the same website
This chart shows the correlation between the average number of self-referential
internal links (= links that refer to themselves) of websites, which are plotted on
the y-axis, with the respective Google ranking, which is plotted on the x-axis.
From the pool of websites that have at least one internal link, eight out of ten
URLs ranked in the 1st place have one link to themselves on average.
This number is naturally influenced by several websites, which have more than
one self-referential link. In the case of some domains and associated URLs, this
phenomenon is purely technical ‒ for example, on Twitter, where each subpage
links back to the profile itself with a tweet in the form of a permanently linked
address of the user (@profile name).
When examining the specific results, it becomes apparent that those top-ranking
websites (including many shopping websites) that refer to themselves using a
keyword tend to do this in the site-wide menu or breadcrumb link.
It should be noted that using a keyword for which one website is supposed to be
ranked to establish a link to another website should be avoided. However, many
websites that rank in the top positions create another internal link to themselves
using the keyword ‒ mainly on a structurally-technical basis.

2.7 Moderate integration of AdSense and other Adlinks is
less negative than before
In 2012 there were clearly negative correlations with good rankings with respect
to the integration of advertising ‒even for AdSense, Google's own product. This
means that last year, well-ranked websites had less advertising than those
ranked lower. In 2013, these correlations are close to neutral.
Figure 20: Correlations ‒ Ad integration
This chart shows that even well-ranked websites are now integrating advertising,
and the number of ads is rising. The following charts illustrate this in greater
detail.
Legend: The y-axis of the charts 21 and 22 is scaled from 0 to 1, and multiplied
by 100 it returns the percentage of websites that rank in a particular position for
which the respective factor applies (y-axis caption).

Figure 21: Average by ranking ‒ AdSense
Figure 22: Average by ranking ‒ Adlinks total (including AdSense)

The main message of these two charts is that especially among the top 3-4 URLs
in terms of search results, there are on average fewer pages with at least one ad
integration (AdSense or another form of Adlinks) than is the case with URLs
ranked in the following places. Lower ranking tend to have more advertising than
the the top 3.
Incidentally, the results for charts 21/22 are not identical. This is simply due to
the fact that the vast majority of websites use AdSense rather than another form
of Adlink. According to this, most of the top 30 websites use the advertising
integration from Google itself ‒ in fact, the integration of other Adlinks such as
AdSense shows a slightly higher correlation with good rankings than in the
previous year.
Figure 23: Average by ranking ‒ Number of AdSense ads
Even in relation to the number of AdSense blocks, well-ranked websites are
positioned below the rest of the top 30. The top ranked URLs in terms of search
results feature less advertising integration than URLs in the second half of the
first SERP. From the second SERP on, the number of AdSense integrations
remains nearly unchanged on average, or it even drops slightly.

2.8 Summary
The quality of content was already very important in the previous year, which
has become even more the case in 2013. Well-ranked URLs tend, to a certain
extent, to have more text and even more multimedia content than was the case
in 2012. In addition, the moderate involvement of advertising ‒ even outside of
AdSense ‒ does not seem to have as negative of an effect on rankings as before.
While keywords in H1 or H2 titles do not seem to be crucial for rankings, they
are almost trivial. Well-correlated feature combinations were not identified.
A good internal link structure is also an important quality factor. It seems to
have a positive effect, when the most relevant URL of a domain for a particular
keyword uses this keyword once again to create a link to itself. On average, the
majority of top-ranking URLs contain one self-referential link.

3. BACKLINKS
The quality factors in the area of backlinks are probably the most important SEO
metrics on the off-page level. Google's success is based not least on the criterion
of "PageRank", the metrics of which have always been backlink-heavy. Links
from other websites, therefore, are historically among the most important
indications of relevance in the form of recommendations while taking the main
subject of the website into account. Any possibilities of manipulation that existed
in the past have been continually minimized by the aforementioned iterative
algorithm changes performed by the search engines.
As a result, the ranking of a website is determined not only by the number of
backlinks, but also the quality of the backlink structure, which, in turn, is subject
to various factors. In order to reflect the diversity and complexity of the backlink
cluster, the factor pool for the analysis of backlinks was expanded to include
several parameters, including its own metric of SEO visibility.
Figure 24: Backlinks ‒ Rank Correlation 2013
The Penguin and the EMD update in the US have had a decisive impact on the
development of correlations in the course of the year10
. Matt Cutts, Google's
"Head of Webspam" described the importance and the extent of the latest
10
http://searchengineland.com/penguin-2-0-losers-porn-sites-game-sites-big-brands-
like-dish-com-the-salvation-army-160744/

Penguin update shortly after its roll-out on his blog11
. In comparison to the
previous year, this factor has changed as follows:
3.1 Backlinks: Developments since last year
The following chart shows the development of four factors compared to the
previous year.
Figure 25: Backlinks 2013 ‒ Changes since 2012
Websites that rank in top search result positions contain an even higher number
of backlinks than was the case last year. Moreover, the number of links as shown
in the summary chart exhibits one of the highest correlations with good rankings,
which becomes the highest once the correlations of the social area are taken out
of the equation.
The websites are not only required to have a greater number of links, but the
relevance of the quality of links has also increased.
As a result, well-ranked websites tend to have a significantly higher proportion of
both “nofollow” links as well as backlinks with a stop word, while the proportion
of back links with a keyword has been markedly reduced. The correlations
between these quality factors have shifted quite considerably. As anticipated for
some time, links that have been "artificially" keyword optimized appear to have
been discounted in favor of a "natural" link profile.
11
http://www.mattcutts.com/blog/penguin-2-0-rolled-out-today/

3.2 The number of backlinks remains immensely
important
The chart shows in detail the massive difference between websites ranked in the
first place and further in the back:
Figure 26: Average by ranking ‒ Number of backlinks
This chart shows the number of links in relation to the ranking of the linked
websites. While websites ranked 1st have around 6,038 backlinks12
on average,
this figure drops for websites ranked 2nd to 2,221 backlinks. Websites ranked in
the 30th position only have just about 70 backlinks on average.
3.3 Link diversity is always relevant
The positive trends with respect to link quality factors, which have displayed an
even higher correlation with good rankings in 2013 than was the case in 2012,
suggest that a more differentiated link building strategy is required in order to
achieve the goal of optimized rankings.
12
Excluding Wikipedia URLs. When Wikipedia URLs are included, the average decreases
by around 30%

Figure 27: Average by ranking ‒ number of words in anchor text
The "soft" factors on the backlink level seem to be gaining in relevance. Well-
ranked websites not only have a significantly higher proportion of multi-word
anchor texts in the backlink profile, but also feature “nofollow” links, which is
demonstrated in the following chart (y-axis: the proportion in % is obtained by
multiplying the result by 100):

Figure 28: Average by ranking ‒ Percentage of nofollow backlinks
The relevant correlations with link diversity are clearly pronounced, and the
upward trend is expected to continue. In conclusion, this means that the factor
"keyword in anchor text" ‒ or the "keyword as anchor text" ‒ has lost its
relevance. Indeed, a look at the correlation of this factor shows a clear result.
3.4 "Hard" keyword anchor text has lost its relevance
The days of "hard" keyword links appear to be numbered. The positive
correlation for this factor across the top 30 has decreased significantly during the
year and is now barely in the positive category. The downward trend is expected
to continue.
The detailed view shows quite clearly that this factor does not (any longer) have
a well-formed correlation. The shape of the curve remains virtually unchanged.

Figure 29: Average by ranking ‒ percentage of backlinks with keyword
Nevertheless, it should be noted that in absolute terms, backlinks with a keyword
clearly dominate the backlink portfolio of well-ranked websites (the proportion in
% plotted on the y-axis is obtained by multiplying the results by 100). On
average 37 percent of backlinks of websites ranked among the top 30 contain
keywords in the anchor text.
This is not surprising ‒ after all, having keywords in the link text is part of a
natural backlink profile ‒ in particular in the form of brand links. Crucial in this
context is, however, the distribution of individual link parameters within the
backlink profile as well as the trend, which is sharply downward.
This is also noticeable with regard to the proportion of backlinks with stop words,
which has actually not only increased significantly for websites ranked at the top
of search results but also shows a particularly strong correlation:

Figure 30: Average by ranking ‒ Percentage of backlinks with stop word
The URLs in the first four places have a higher proportion of links with stop words
than websites that rank below them. From Google’s point of view, it seems to be
simply "unnatural” when individual URLs are always linked with the same
keyword or anchor text.
At the very least it is highly unusual that a large number of different webmasters
(authors, etc.) would independently link a website with exactly the same anchor
text. In any case, Google tends to suspect manipulation ‒ and probably quite
justifiably ‒ in the case of link profiles, where the proportion of hard keyword
links exceeds a certain tolerance level. For this reason, Google has been
depreciating this factor. It is, therefore, clearly more important to increase the
number of links ‒ in particular, brand links, links with stop words, “nofollow”
links ‒ and at the same time to decrease the number of keyword links.
3.5 Backlinks from URLs with high SEO Visibility are
significantly more valuable
In introducing this subsection, it is necessary to explain the term "SEO Visibility"
first. This metric, which is also referred to as "visibility" and has its origins in SEO
analysis software tools, is calculated from a whole range of various factors
related to keyword rankings of a fixed keyword pool and used similarly to factor

traffic (which is not measurable from the outside by non-webmasters) as a basic
factor for the quality categorization of websites for comparison purposes. In
addition to keyword ranking, Searchmetrics also takes into consideration search
volume, information about keyword characteristics (e.g. navigational or
informational search) as well as the layout of the search result. In short: The
higher the SEO Visibility of a website, the more traffic it likely generates13
.
The correlation of the factor of "SEO Visibility of the backlinking URL", reveals
that the traffic of URL B ‒ in terms of its relevance as a source for a link to URL A
‒ is crucial for the ranking of URL A.
In short: Backlinks from websites with high SEO Visibility seem to be much more
valuable than links from websites with low SEO Visibility.
Figure 31: Average by ranking ‒ Average SEO visibility of the link source
Again, top ranking URLs clearly stand out with their backlink profile from other
URLs that rank further down. While websites ranked in the 1st place tend to have
links from URLs with an average SEO Visibility of over 8,63614
‒ in the 2nd place
this drops to an average visibility of around 5,510 and in the 30th place to only
around 664.
13
The SEO Visibility of domains is calculated for 20 countries and is available for free at
http://suite.searchmetrics.com/uk/research
14
Excluding Wikipedia URLs. With Wikipedia URLs for pos. 1: around 11,000 SEO Visibility

3.6 Summary
Backlinks continue to be one of the most critical SEO metrics. Over the years,
very little has changed to diminish the importance of the factor "number of
backlinks". Websites with more links rank higher in search engines. However, the
factors related to this metric have evolved over time. As a result, it is no longer
only the quantity of backlinks which is crucial, but also increasingly their quality.
Google has significantly discounted several keyword factors ‒ especially during
the last year. Naturally, a certain proportion ‒ and not necessarily a small one ‒
of so called "hard" anchor text or "keyword link" is still considered to be "natural"
to a certain extent. Not least through its Penguin updates and iterations, Google
has changed the tolerance limits decisively.
As a result, well-ranked sites feature a balanced mix of “nofollow” links, multi-
word anchor texts, link text with a stop word share and keyword links. The
quality of the link source also plays a crucial role. Google's algorithms and the
group's own quality raters are getting better at unmasking link farms. Links from
websites with high SEO Visibility are much more valuable. The backlink profile is
nothing more than a collection of very complex quality factors.

4. SOCIAL SIGNALS
On average, social signals are the domain with the highest correlations. At the
beginning of this chapter, we want to reiterate that correlation is not
synonymous with causation.
More detailed analyses of the actual impact of social signals on the indexing and
ranking of URLs conducted in isolation from other factors were part of several
additional studies, which will be discussed in more detail later in this chapter.
Nonetheless, the following fact remains completely disconnected from causal
aspects: Websites that rank in the top positions on Google usually have a large
number of social signals. In other words: well-ranked URLs have many shares,
likes, comments, plus ones and tweets.
Figure 32: Social signals ‒ Rank Correlation 2013
In addition, the strong correlation of the entire social sector is supported by the
fact that the number of social signals seems to drop in line with the ranking of
websites in SERPs – the worse the ranking, the lower the number of social
signals. This is particularly evident in the detailed charts shown in the following
subchapters. However, we will focus our attention first on the development of
correlation of social signals since last year.

4.1 Social Signals: Developments since last year
The following chart shows that social signals that already had a very high
correlation with good rankings in 2012 still feature frequently and strongly in
URLs well-positioned in the SERPs in 2013.
Figure 33: Social signals 2013 ‒ Changes since 2012
Facebook likes as well as tweets are even showing slightly higher correlation
than in the previous year.
A new addition to the factor pool are signals from Google's own network
Google+, which also shows a strong correlation with good rankings, as well as
social signals from the aggressively promoted photo community Pinterest.
4.2 Social signals continue to show strong correlation
with good rankings
The correlation of social metrics with better rankings is impressively high. This
has already been the case last year, and it is true again in 2013.
Recommendations and related signals from the social chart have a strong
correlation with good rankings on Google.
The detailed charts for individual networks have distinct curve shapes, which are
used to calculate the strong correlation.

Figure 34: Average by ranking ‒ Facebook signals in total
As an example, we show here a correlation of the summary of all Facebook
metrics (shares, likes and comments) for the factor "Facebook total" which has
‒ as an aggregate of all factors ‒ one of the highest correlations and by far the
highest average result on the y-axis when compared to the Google rank (x-axis)
within the social area.
The average number of tweets from the Twitter network ‒ which is more of an
information aggregator than a social network ‒ is in relation to the Google rank
of the relevant URL already well below the figures for Facebook's signals. Even
here, the correlation with rankings is clearly visible.

Figure 35: Average by ranking ‒ Twitter signals
Pins ‒ i.e. signals from the Pinterest network, which allows members to "pin"
content, mainly in the form of images to their so-called pinboards ‒ are
comparatively rare as a social signal per website and ranking. However, even
here the distribution reflects a fairly good correlation.

Figure 36: Average by ranking ‒ Pinterest signals
In summary, the following can be said about social media factors:
Typical for these curves is the e-function shape, which is characterized by a
sharp drop between the ranks 1 to 2 followed by a much shallower slope as the
curve slowly converges to the x-axis. URLs that rank at the top of search results
have by far the greatest number of social signals across all the analyzed
networks.
4.3 Google+ new in the analysis ‒ with very high
correlation
Of particular interest in the analysis of Ranking Factors for the search engine
Google is clearly the correlation of signals from the company's own Google+
social network, which has seen an enormous growth in users and activity over
the past few months.
Due to insufficient data, it was not possible to make any meaningful statements
about this network and its signals as a Ranking Factor last year. In the UK,
Google+ signals are now just behind tweets in absolute terms.

Figure 37: Average by ranking ‒ Google+ signals
Even on a qualitative level, Google+ signals seem to correlate well with top
search result rankings. In 2013, plus ones show the highest correlation with good
rankings.
4.4 Digression: Google+ will overtake Facebook at the
beginning of 2016
At this point we want to offer another digression. This time we will focus on
signals and activity of the two by far largest social networks on the Internet.
The catalyst for this analysis is the remarkable development of Google's network
Google+, which manifests itself not only in rapid membership growth, but also in
the already examined strong correlation with search results, as well as the high
number of plus ones for URLs at the top of Google's rankings.
While last year no meaningful conclusions could be drawn with regard to the
correlation of plus ones due to lack of data in 2013, it has recorded the highest
rank correlation coefficient for this metric.
Searchmetrics has also carried out several detailed studies with respect to the
performance and activity of the two networks ‒ Facebook and Google+ ‒ and the
interpretation of the respective group strategy behind this accelerated growth ‒

which will not be presented as part of this study, but will be briefly introduced in
the summary of this chapter.
We will, however, briefly introduce one aspect of this study, which focuses on the
evolution of networks and their competitive character. This is a forecast, which
addresses the following question:
When will Google+ overtake Facebook, the current leader in social media in
terms of the number of social signals?
Even the choice of wording of this question hits towards the result ‒ we are not
asking "if" but "when."
As a brief introduction, the following can be said about data pools: The starting
point for the calculations were the last available monthly percentage growth
rates of both networks on average over a period of six months, limited to
members' activity in the form of "shares" (Facebook) vs. "plus ones" (Google+).
Plotted on a chart, this analysis looks as follows:
Figure 38: Percentage growth and absolute number of Facebook shares vs. plus ones
The growth forecasts for the coming months and years for both networks are
based on this data. The purpose is to estimate the point in time when Facebook
will be overtaken by Google+ in absolute terms with respect to Facebook shares
per month and plus ones, respectively.

Figure 39: Facebook shares vs. plus ones ‒ growth forecasts in %
Conclusion: According to the projections based on data collected so far, Google +
will overtake Facebook in February 2016 based on a comparison of the number
of plus ones vs. Facebook shares.
4.5 Summary
Of all the factors, social signals are the strongest in terms of correlation with
better rankings. The short to medium-term outlook is clearly positive. These
positive results are spreading from Facebook, Google+, Twitter and Pinterest to
all networks.
Nevertheless, the question remains as to the extent in which social signals affect
rankings directly. As already mentioned, the analyses to date have mainly
documented the fact that better ranking URLs have a greater number of social
signals than those with a worse ranking.

Several detailed analyses were conducted last year with regard to the causal
connection between social signals and good rankings.
For this purpose, two texts15
on two cognate subjects (four articles in total) that
were identical in content were written and published at the same time on the
same domain under as neutral conditions as possible (no sitemap, no comments
function, no internal / external links from or to the URL). Next, social signals
were specifically targeted to the article - under strict network-specific separation.
This means, that two articles only received Facebook signals, whereas the copies
of this text were exclusively shared on Google+.
While observing the strict separation and compliance with the neutral
environment, we evaluated the rankings for specific keywords, which have been
determined in advance and were identical for the respective articles.
The results of these social signal studies included: Indexing only through social
signals, without presence of any internal or external links, is possible; a
significantly smaller number of plus ones is in the position to produce faster
indexing or better ranking for an identical set of keywords in a direct
comparison; with a repeated increase of shares or plus ones after a period of
stagnation, the ranking increases again, even if this has previously also
stagnated or even declined.
Without having to reproduce the case studies here in greater detail, we can
conclude that it appears that social signals can influence the ranking of URLs
even in isolation.
15
No duplicate content

5. THE BRAND FACTOR
A phenomenon that pervades the entire study, and has had a strong impact on
the results in nearly all correlations and detailed charts is the brand factor.
Brands rank in the 1st position, even if they do not meet certain criteria, or if
they do not meet them to a sufficient extent. Conclusion: From Google's
perspective, brands have a special role and are strongly preferred by being
ranked first in SERPs.
5.1 High impact of the brand factor on SERPs
To illustrate the "brand factor" phenomenon we provide again several meaningful
example charts, which highlight any distinctive areas in the shape of the curve
with respect to the top search result rankings.
Figure 40: Average by ranking ‒ H1 title tag
The special position of brands is particularly noticeable in on-page coding. It
follows from chart 40 that it is much rarer for brand websites to have a H1 title
than it is the case for all other websites in the top 30.

On average, about 55 to 60 percent of webpages out of the top 30 do have a
keyword in the description. Only regarding the top 3 ranking pages, the average
is significantly lower.
5.2 The top ranks are reserved for brands
While the requirements placed on non-brands to be ranked on the first SERP are
consequently, very high, the top search result positions for certain search
queries appear to be reserved for brands per se, without the fulfillment or non-
fulfillment of certain Ranking Factors being of any relevance.
This failure to implement some important factors, which for non-brands is a basic
precondition to be ranked at all, concerns mainly the on-page area – brand
websites use i.e. much less frequently keywords in the title or description, offer
on average less content and have a lower number of internal links.
On the positive side, brand websites usually feature a better than average
backlink profile. Link sources with high SEO Visibility tend to establish links to
brands more frequently. In addition, brands receive significantly more social
signals. Consequently, factors from the off-page area seem to have to a
considerable impact on the special role of brands.

6. THE SPECIAL ROLE OF WIKIPEDIA
Wikipedia not only has a major impact on ranking of search results but also the
correlations calculated for this study. The crowdsourcing project enjoys an
extremely high level of trust and high ratings for many keywords (especially
"informational keywords"16
).
To generalize, one could say that Wikipedia typically ranks in the 2nd
position17
‒
for example, behind the respective brand or the specific URL for the search term
itself. However, if there is no significant brand that would match the search
query, Wikipedia often ranks directly in the 1st position. This is often the case
with aforementioned "informational keywords."
6.1 Wikipedia results affect correlations
The actual extent of the Wikipedia influence is highlighted by the following
correlations which have been calculated specifically mainly for Wikipedia in
addition to the existing correlations.
Figure 411: Wikipedia correlation
This chart illustrates the somewhat trivial conclusion that the feature "Wikipedia
URL" or "Wikipedia URL-EN" indeed appears to be a Ranking Factor.
16
Keywords that are googled by users seeking information, e.g. a definition
17
Taking into account the search volume of a keyword and thus the query frequency

The other two correlations arise from the original intention to analyze the
correlation between top-level domains (TLDs) and search result rankings, which,
particularly in the case of the “TLD.org” was strongly influenced by other
confounding factors and thus no meaningful conclusion could be derived.
At the same time, the individual correlations for “TLD.org”, including or excluding
Wikipedia, are influenced by a major disruptive factor called Wikipedia, which if
added to the mix makes the “TLD.org” factor correlate much better with good
rankings than it would be the case without.
The extent of the Wikipedia influence on the search results can be derived from
the following detailed correlation for the feature "Wikipedia URL".
Figure 422: Average ranking ‒ Wikipedia URL
This chart shows that Wikipedia URLs tend to have a high ranking at
google.co.uk. Generally, they rank 1st, more often 2nd and rarely 3rd. The data
also showed slight clustering at rank 17 and 27 for a long time.
This unusual data clustering of Wikipedia's results was until recently even more
pronounced.

6.2 Rank 17 and 27 "reserved" for Wikipedia results
Not long time ago, the line charts for the individual factors presented in this
study showed in many cases substantial breakouts at rank 17 and 27. This is
where Wikipedia's results ranked for a long time with very high distribution18
.
This regularity was so strongly pronounced that it suggests that it may have
been implemented in the algorithm deliberately.
However, after being in place for a long time, this distribution has shifted
markedly in the 22nd calendar week due to Penguin 2.0 Update even if not
completely dissolved. Nevertheless, the share of keywords to which this
distribution applies has decreased significantly in the Searchmetrics database.
This would suggest that Google is currently working on its algorithm.
This lead to the fact that the data used in this study was aggregated again in
order to take account of the latest developments. The charts in this report are,
therefore, based on the current dataset19
.
It is noticeable from the 22nd calendar that the Wikipedia results - with a few
exceptions ‒ tend to rank in the 1st, 2nd, or 3rd position or in the 17th or 27th
position. Before the Penguin update, the situation was different. Prior to the 22nd
calendar week, there were still combinations of rank 1-3 and 17, 27 for the same
keyword in relation to Wikipedia. However, this is now hardly the case.
Meanwhile, Google seems to be moving any Wikipedia results that rank further
down to the front and as a result there are often several Wikipedia URLs ranked
in the first positions, which are, nevertheless, thematically related to the search
query. At the same time, the positions 17 and 27 are often taken by Wikipedia
subsidiaries, such as the collection of quotes Wikiquote20
.
The question remains as to what purpose Google has been ranking Wikipedia
URLs with such a striking consistency in these parts of SERPs. The fact is that to
improve its performance, save processing time and relieve the server, Google
caches its SERPs for certain search queries in advance. A possible explanation for
the striking ranking distribution would be that Google "parks" Wikipedia results
for cache purposes in ranks 17 or 27 to catapult it to the top of search results
when e.g. a related, but slightly adapted search query arises. As already
mentioned, this clustering has decreased significantly in the meantime.
18
This has also been the case at rank 37, 47 ff. However, this study focuses on the top
30.
19
June 2013
20
http://www.wikiquote.org/

Summary of Searchmetrics SEO Ranking
Factors – Rank Correlation 2013
To conclude this study, we provide an overview of all rank correlation coefficients
of Google UK, as well as their development since the previous year, 2012.
Figure 433: SEO Ranking Factors – Rank Correlation 2013 – Changes since 2012
It is noticeable that social signals correlate well with better rankings and content
quality as well as the number and diversity of backlinks seem to have a huge
impact on search result rankings.

Summary of most important developments
in Ranking Factors – Rank Correlation 2013:
The most important findings of this study are summarized again below:
Keywords are losing in importance ‒ not only with respect to technical on-page
factors such as domain names or URLs, but also backlinks in the form of link
texts. The days of "hard keyword optimization" of links are well and truly over. In
other words, the aforementioned features have lost relevance as a Ranking
Factor - Google has simply devalued their once positive influence. However, this
does not mean that the basic requirements relating to the existence of keywords
and related terms on websites have also become obsolete. It remains one of the
basic principles to place relevant keywords on-page, even if not necessarily in
the domain or the URL but preferably in the title and description. Google remains
a machine, even if the semantics of its algorithm are continuously being
optimized. It is also important for a website to contain a description, H1, and H2.
Brands are the exception to the rule. The decoupling and separation of the
"brand" factor from the "keyword" factor has been put into practice very well by
Google. The majority of top positions in SERPs are occupied by brands and thus
websites that mostly do not meet some of the basic principles (often on-page
factors).
The entire range of examined social signals correlated well with better rankings.
Well-ranked URLs have an abundance of likes, shares, plus ones and tweets.
URLs ranked in the 1st position stand out through their large average number of
signals. Detailed studies conducted by Searchmetrics suggest that under certain
circumstances, social signals appear to have an impact on indexing and ranking
of URLs even in isolation. From Google's perspective, social signals function as a
reliable signal of good-quality content, especially since the search engine is
sitting directly at the source of the social chart in the form of its own network
Google+, which is doing very well and is likely to be doing even better in the
future.
As already mentioned, several traditional coding factors have gained again in
importance. On the content level, nearly all factors correlate significantly better
than it was the case in the previous year. It appears that more words in text as
well as the integration of additional media in the website both have a positive
effect on ranking, while a moderate integration of advertising no longer seems to
be an obstacle to good rankings.
In a nutshell: Google favors quality, substance, and ultimately, relevance of the
respective search result and always takes into account secondary factors such as
brand websites and exceptions, such as Wikipedia.

About Searchmetrics
Searchmetrics is a pioneer and the world's leading provider of search and social
analytics software for companies and online agencies.
Searchmetrics offers a unique server infrastructure and software solution which
its customers and partners can use to develop intelligent, digital marketing
strategies to increase their visibility and success in the digital world. With the
world's largest pool of proprietary data and external analyses and information,
the Searchmetrics SuiteTM
uniquely combines various factors, which make it
possible to construct accurate forecasts and projections. Furthermore, users can
use the software to aggregate and analyze large amounts of data on the ranking
of websites, search keywords and the activity of competitors, as well as gain
insight into the visibility of websites on search engines in 124 countries, 155
cities and on the largest social media portals. Inbound marketing plays a decisive
role in this, as excellent visibility on search engines is a key priority for
companies. Only companies that use intelligent and integrated content strategies
across all online channels to present themselves will be found and noticed by
potential customers, enabling them to increase their ROI. In addition to the
Suite, Searchmetrics offers Searchmetrics Essentials, a tool set for getting a
quick start in search engine and social network optimization to gain market
insights through detailed analysis.
The company has offices in Berlin, London, Paris and New York and sells its SaaS
solution through a network of partners and its online store. The fast-growing
Berlin-based company, in which the Holtzbrinck publishing group and well-known
venture capitalists Neuhaus Partners and Iris Capital hold an interest, currently
has over 100 employees and an ever-growing customer base. T-Online, eBay,
Siemens, TUI and Kaspersky are some of the well-known customers who already
rely on Searchmetrics to achieve excellent search engine rankings and to
implement optimal strategies for the digital world.
If you have any questions about the Searchmetrics SuiteTM
or if you would like to
receive a quote:
For more information, visit our website at: www.searchmetrics.com
Contact us now!

Searchmetrics Northern Europe
Atrium Court
The Ring
Bracknell RG12 1BW
Phone: +44 1344 393 056
whitepaper@searchmetrics.com

Searchmetrics seo-ranking-factors-2013-uk

More Related Content

What's hot

Viewers also liked

Similar to Searchmetrics seo-ranking-factors-2013-uk

More from Daniel Howard

Recently uploaded

Searchmetrics seo-ranking-factors-2013-uk