On Page SEO: A (2M Keyword) Data Driven Analysis

Tim Soulo
Tim is the guy responsible for marketing and product development at Ahrefs. But most importantly he’s the chief evangelist of the company.
    Google's algorithms are evolving all the time. Back in the day you could "stuff" your page with a bunch of keywords you wanted to rank for and that alone would help to improve your rankings. This was of course very open to manipulation.

    But in 2016 we have Hummingbird, Rank Brain, semantic search and many other nifty technologies - Google has evolved far beyond just looking if you have a keyword in your Title or not.

    And besides, we've all noticed that pages often rank on the front page of Google without even having the target keyword mentioned anywhere in their content.

    So does this mean you should no longer worry about optimizing your page for a specific keyword and just let Google figure out what your page is all about?

    Well, we have studied the correlations of different on page SEO factors with Google rankings across 2 Million random keyword searches and even though correlation is not causation, the takeaways are very interesting in a lot of cases.

    What is “On Page SEO”?

    “On page SEO” refers to a set of web page optimization best practices that you can apply to the pages of your website in order to improve their ranking in search engine results.

    The majority of on page SEO advice that can be seen around the web revolves around using the exact match keyword that you want to rank for in a few "strategic" places of your page: Title, H1, meta description, content etc.

    But this kind of advice is actually outdated, because in 2016 Google is sophisticated enough to understand synonyms (and overall relevance of the page), meaning you no longer have to obsess about exact match keyword usage.

    When people who are fairly new to SEO are picking a keyword to rank for, they might look at the SERP and see that none of the top10 pages has a "perfect" on page SEO for this exact keyword:

    04-on-page-seo

    Isn't this a great opportunity to rank with their own "perfectly optimized" page?

    I'm afraid it's not.

    The fact that these pages don't have your exact match keyword in a bunch of "strategic" places, doesn't mean that they are not relevant to the search query.

    According to our data, the usage of exact match keyword in Title, URL, H1 or even within the actual content of the page doesn't have a significant correlation with Google rankings.

    If we were to study "partial match" keyword usage, synonyms and "LSI keywords" (which we're about to) - the results would most likely be different.

    But "on page SEO" doesn't end with using the keyword you want to rank for in the content of your page, right?

    It also implies quite a few "general" optimizations that should make your page better in the eyes of search engines (and searchers), such as:

    • page load speed;
    • usage of https;
    • length of your content;
    • outgoing links to quality sites;
    • etc.

    So let's take a look at the correlations of all these on page SEO factors and compare them to backlink factors:

    How "On Page SEO" Factors Correlate With Google Rank

    00-On-Page-SEO-Factors

    At this point I'd like to mention one more time that correlation is not causation.

    These correlations show you the common traits of the pages that tend to rank well, but they do not necessarily imply that these pages rank well because of these traits.

    Correlation is measured on a scale from -1 to 1 with "0" meaning "no correlation at all". And as you can see, all the most popular on page SEO factors that we studied hardly reach 0.1 mark.

    You can clearly see that on page SEO factors that revolve around using an exact match keyword in "strategic" places of your page showed a very small correlation.

    Sidenote.
    We didn't study the correlation of "partial match" keywords, synonyms and "LSI keywords" this time. But once we do - we will update this article with more data.

    Another interesting graph would be this one:

    00-backlink-factors-on-page-factors

    Such a huge difference in correlations suggests that backlinks have much more influence on your page's rankings than usage of exact match keyword in the copy.

    But even though our research data suggests that usage of exact match keyword on your page has a very low correlation with Google rankings, this doesn’t mean that you should completely refrain from using it.

    So let’s look at each on page factor one by one and discuss if you should or should not care about it.

    Sidenote.
    Our research was based on a sample size of 2M keywords, but there were times we had to reduce this number in order to study certain factors in isolation.

    For most of the below experiments we tried to exclude the influence of backlinks by focusing on SERPs where top 10 ranking pages had similar DR and UR (the standard deviation is less than 30% of their Average value).

    For each of the experiments we calculated 4 correlations:

    1. Across all keywords;
    2. High-volume keywords only (50k+ searches per month);
    3. Medium-volume keywords only (20k-50k searches per month;)
    4. Low-volume keywords only (less than 20k searches per month).

    If you have any questions about the methodology behind any of the below experiments - just post them in the comments section at the end of this article (or tweet me @timsoulo).

    Usage of Exact Match Keyword

    First of all let's look at how using the exact match keyword in a few “strategic” places of your page correlates with Google rankings.

    TL;DR: all our experiments with exact match keyword usage in different places of a page(code) showed a very small correlation with Google ranking.

    1. Keyword in Domain Name

    02-in-domain

    Back in 2012 Google rolled out an update that was meant to decrease the value of the so-called EMDs (Exact Match Domains).

    By looking at our graph it may seem that EMDs are still in the game, because there's clearly a "jump" in position #1.

    But I believe that this “jump” is caused by so-called “branded keywords”.

    For example, if you search Google for “addicting games” - you’ll notice that it’s actually a brand name for a popular gaming website, which ranks #1 for the term:

    001-addicting-games

    The correlation of 0.0877 may seem pretty high (compared to other on page SEO factors), but if we were to re-calculate it without that #1 position I'm sure it would drop a lot.

    Please also note that we didn't focus on EMDs exclusively, but counted any website that contained an exact match keyword as part of its domain name.

    Recommendation
    Can you "coin" a certain keyword by using it in your domain name and building your brand around it?

    YES, you can!

    I did just that with my WP plugin called "Content Upgrades PRO" that resides under the "contentupgradespro.com" domain name.

    And it ranks quite well for the keyword "content upgrades" - apart from the fact that it can't outrank Brian Dean's article that's not even nearly optimized for this keyword (but has tons of backlinks):

    01-Google-serp

    Is EMD a strong ranking signal? I don't think so.

    2. Keyword in URL

    03-in-URL

    Interestingly, out of all “keyword-related” on page SEO factors that we’ve studied this was the only one that showed a negative correlation.

    So does this mean that Google doesn't use "keyword in URL" as a ranking factor?

    Well, just recently John Mueller of Google has stated the following:

    I believe that is a very small ranking factor. So it is not something I'd really try to force. And it is not something I’d say it is even worth your effort to restructure your site just so you can get keywords in your URL.
    John Mueller
    John Mueller, Webmaster Trends Analyst at Google
    Recommendation
    Here at Ahrefs we’re huge advocates of the so-called “descriptive URLs”.

    Here's why:

    #1 You can copy/paste a descriptive URL in your online conversations and people will know what your page is about before even clicking on the link:

    05-descriptive-URL

    #2 This kind of link will automatically contain your target keyword as part of the anchor text:

    06-anchor-text-in-URL

    #3 Google will highlight the keyword that you're searching for in the URL of the search snippet:

    07-keyword-in-URL highlighted

    So if you’re still using URLs that look like “mywebstore.com/?product_id=7924Gh” - you should definitely rethink that format.

    But just like John Mueller has stated - I don't believe that changing your existing URLs to include a target keyword will lead to any significant increase in rankings.

    3. Keyword in Title

    04-in-title

    As you can tell by looking at this graph, the vast majority of pages that rank on the front page of Google don’t have exact match keyword in their "Title" tags.

    Funnily, even the #2 result for "on page SEO" doesn't have the exact match keyword in its "Title" tag:

    12-no-keyword-in-title

    Sidenote.
    As David has pointed out in our discussion of this experiment, the Moz's Title on my screenshot does have all three words of the target keyword in it, but not in exact match order.

    But in this research we've only studied the effect of "exact match keyword" and omitted any partial match occurrences or synonyms.

    If you scroll down that SERP you'll see that the Titles of #4 and #6 results are not "perfect" either.

    So does this mean that "on page SEO" is an easy keyword to rank for if you use "exact match keyword" targeting on your page?

    Not at all!

    Recommendation
    Search for any keyword in Google and you'll notice that it is no longer highlighted in the Titles of the search snippets as it used to be:

    08-titles-highlighed

    That might be a hint that Google is de-valuing the importance of having an exact match keyword in the Title.

    Looks like they want searchers to actually read Titles before deciding which one to click, rather than quickly skim through to find where the exact match keyword is highlighted.

    So do I recommend putting your target keyword in the Title of your page?

    I'd say it depends.

    You can often craft a very catchy Title without a keyword in it and still rank well:

    09-titles-dont-matter

    But will using an exact match keyword in the "Title" tag of your page make it rank better?

    Probably not!

    Unless of course your page doesn't have any other content on it and the Title is the only clue that Google can get about it.

    Like in this experiment by Distilled, where targeting a different keyword in the Titles and H1s of their pages actually improved search traffic.

    But I'll get back to commenting on this experiment in the section about H1 tag.

    4. Keyword in the beginning of the Title

    05-beginning-of-title-tag

    I wanted to study this ranking factor in isolation, so we only looked at the SERPs where keyword was present in the Titles of all top10 pages.

    And that experiment resulted in a very small positive correlation.

    Recommendation
    The correlation numbers behind this experiment suggest that Google might actually be using this as one of their ranking factors (a very-very weak one).

    But I believe that using an exact match keyword at the beginning of your Title doesn't give you any significant advantage over pages that don't.

    So if the Titles of the top 10 pages in a SERP don't start with a target keyword - you should not consider this as an easy ranking opportunity.

    5. Keyword in Meta Description

    First of all we studied how many SERPs in our data sample had the "meta description" tag filled in:

    06-PERCENT-meta-description

    And it looks like the vast majority of pages ranking in Google top 10 don't have meta descriptions.

    Which kind of implies that filling in this tag is not crucial for your rankings.

    Then we took the segment of SERPs where all 10 "meta description" tags were filled in and studied if using exact match keyword within this tag correlated with rankings:

    06-meta description

    And again, over 50% of all meta descriptions don't have a target keyword in them and the correlation with Google ranking is very close to 0.

    Recommendation
    Should you fill in the "meta description" tag for every page on your website?

    I'd say it's recommended, but not crucial for ranking well.

    In most cases Google will omit your "meta description" anyway and just pull a short excerpt from the content of your page (that it thinks is the most relevant to a user's search query):

    13-meta-description

    And yet, as much as I hate filling in "meta descriptions", they just have to be polished for a few core pages of your website:

    14-lame-description

    All in all, I don't believe that using an exact match keyword in your "meta description" tag will improve the rankings of your page.

    So you probably shouldn't take this factor into account when analyzing your chances to rank for a certain keyword.

    6. Keyword in Headline (H1 tag)

    07-headline

    Looks like 85% of pages that rank in Google top10 don't have the keyword in their H1 tags (based on our data sample).

    I guess that's mainly due to improper use of H1 tags in the layout of web pages - our data has shown that over 70% of pages that rank in Google top10 don't have H1 tags on them.

    Sadly most web developers are still not familiar with the basics of SEO.

    Recommendation
    Do I recommend you to use your target keyword in the Headline of your page?

    YES, I do!

    Here's why:

    #1 Having a proper headline improves the user experience and the "scanability" of your page.This way visitors can quickly guess that they are in the right place and dive deeper into the page, which translates into all sorts of "behavioural ranking factors".

    #2 By framing your headline around a specific keyword or phrase you can influence how people link to you:

    15-title-backlink-framing

    And as you already know, the anchor text of your backlinks (and most likely the surrounding text too) does have a certain impact on the keywords that your page will rank for.

    But do I think that using an exact match keyword in the H1 tag of your page is a strong enough ranking signal in itself?

    NO, I don't think so.

    Unless of course the H1 tag is the only unique content there is on your page, and the keyword you're targeting is very-very long tail.

    Like in this experiment by the guys from Distilled that I've mentioned earlier:

    16-h1-tag-hotels

    Do I recommend using a keyword that you want to rank for in your H1 tag?

    YES, I do. (for the reasons mentioned above)

    7. Keyword in Subheading (H2 tag)

    08-subheading

    This time over 93% of pages that rank in Google top 10 don't have a target keyword in their H2 tag (based on our data sample).

    So the recommendation will be very straightforward.

    Recommendation
    I don't think that using exact match keyword in the H2 tag of your page will lead to any significant boost in rankings.

    So when writing subheadings I advise you to think about your readers, not the search engines.

    8. Keyword in Content

    09-percent-in-content

    Our research has shown that almost 75% of pages that rank in Google top 10 don’t have even a single mention of an exact match keyword in their content (based on our data sample).

    And the correlation of this on page factor with Google rankings is very close to 0.

    Initially we also wanted to study the correlation of the good old "keyword density" too, but after seeing these numbers it no longer made sense to do it.

    Recommendation
    Google called RankBrain their "third most important factor in the ranking algorithm along with links and content".

    This is clearly yet another hint for us SEOs to stop being obsessed about how many times we should use the exact match keyword on a page and focus on creating resources that perfectly address user intent behind a certain search query.

    9. Keyword in the first 100 words of a page

    11-in-first-100-words

    To study the correlation of this on page factor we only looked at the SERPs where all top 10 results had at least one keyword occurrence in content.

    And in about 80% of cases we saw it in the first 100 words.

    Though the correlation appeared to be almost non-existent.

    Recommendation
    My recommendation here is no different to the previous one.

    When writing content for your page you should first and foremost focus on helping searchers, rather than worry about where in the content of your page this exact match keyword should appear.

    10. Keyword in image "alt" tag

    10-alt-tag

    This is yet another case where the vast majority (70%+) of pages that rank in Google top 10 don’t have the target keyword in the “alt” tags of their images. And actually over 50% of all pages that we saw in Google top10 didn't have their alt tags filled in.

    But please be advised that we’ve studied regular search results.

    For Google Image Search the results as well as the correlation would most likely be very different.

    Recommendation
    "Alt" tag is something that you should use to describe what's on the image.

    That's because this text will later be used if your image didn’t load or your visitor is using a screen reader.

    Most likely a descriptive "alt" tag will help your image rank higher in Google Image Search (we didn't study that though).

    But flooding your "alt" tags with keywords that you want to rank for is unlikely to move the needle.

    And this wraps up the section of on page SEO factors that imply using your exact match keyword on a page.

    Up next are…

    General On Page SEO factors

    In this section we’re going to look at the on page SEO factors that are independent from the keyword that you want to rank for.

    In some cases they actually showed a much better correlation with Google ranking than "keyword-dependent" factors.

    But they still look very weak when compared to backlink factors.

    1. Age of the page

    01-age-of-the-page

    Out of all the on page SEO factors that we’ve studied, this one showed the best correlation with Google rankings.

    Sidenote.
    please don't interpret this as "older pages had more time to gain more backlinks", because like I said earlier, all on page factors were studied across SERPs with similar amount of backlinks so we coud remove their influence.

    We also studied that factor from another angle - by looking at the percentage of pages that were less than 1 year old.

    And here’s what we’ve got:

    11-less-than-1-year-old

    Recommendation
    The data above (as well as our personal experience) suggests that newly published content takes quite some time to get to the first page of Google.

    So if you need to grow your organic search traffic fast, consider investing your time and resources into the old pages of your site that already rank for a bunch of keywords.

    With a little effort you should be able to push them up in the search results.

    Here's how:

    1. Update the content on these pages, make it up to date;
    2. Make sure your page matches "searcher's intent" better than what's already in top10;
    3. Change the "last updated" date on that page;
    4. Re-promote it as if it was a new page;
    5. Build new links.

    2. Using HTTPS

    02-https

    Looks like over 80% of pages that rank #1 in Google are not secure (based on our data sample).

    But at the same time you can clearly see that the percentage of secure pages in positions 1-3 is visibly higher than in positions 4-10.

    So should you consider switching to https?

    Recommendation
    Back in 2014 Google officially stated that they will use HTTPS as a ranking signal.

    And our own research has confirmed that "https" does correlate with Google rankings (a little bit).

    This is why I do recommend you to make your website secure.

    Especially in case you’re collecting any kind of personal data from your visitors.

    To be honest it's very unlikely that a switch to "https" will lead to any significant boost in your rankings, but at least visitors will notice “https” in your web address and perceive your website as more trustworthy (try Let's Encrypt).

    3. Page load time

    03-page-speed

    Last year we saw some rumors that Google might have stopped considering page speed as a ranking factor.

    And according to the results of our research, the correlation of this on page factor with higher position in Google is very small.

    Sidenote.
    We’ve only studied the load speed of the base HTML code of the page. Ahrefs crawler doesn’t download any images, css, js, etc.
    Recommendation
    I believe that if your site is utterly slow (to the point when it starts to irritate your visitors) it would be totally fair if Google would give a (very small) bit of preference to sites that are faster than yours.

    But if your site loads “fast enough” to keep your visitors happy, optimizing it even further and cutting down on a few dozen milliseconds might not be the best use of your time and resources.

    4. Length of content

    04-long-form-content

    This on page factor showed the second best correlation across all on-page SEO factors that we’ve studied (the first one is "Age of the page").

    The conventional wisdom says that "long form content" (the one that tends to attract tons of backlinks and social shares) starts roughly at 1500 words.

    But notice how our median value for #1 result only shows 800 words - is that "long form content"?

    I think Rand Fishkin has the point here, by saying that “Great Content ≠ Long-Form Content”.

    So my recommendation would be pretty much in line with this idea.

    Recommendation
    Even though our data has shown a relatively high correlation of content length VS Google ranking, I believe that purposely making your content longer with a goal of ranking higher is a terrible idea.

    In the words of Antoine de Saint Exupéry:

    Perfection is attained, not when there is nothing more to add, but when there is nothing more to take away.
    Antoine de Saint Exupéry
    Antoine de Saint Exupéry, French writer

    So don't make your pages longer just for the sake of it.

    Do the opposite actually - give more value with less words.

    5. URL Length

    12-characters-in-URL

    We've studied the part of the URL that starts where the domain name ends:

    • domain.com/short-url
    • domain.com/this-url-is-a-bit-longer-than-the-last-one

    And as you can tell from the graph above, shorter URLs tend to rank better.

    So then we studied if there was any correlation with number of "folders" in URL:

    06-folders-in-URL

    Here's how we were counting them:

    • http://domain.com (Folders = 1)
    • http://domain.com/folder1 (Folders = 2)
    • http://domain.com/folder1/folder2 (Folders = 3)

    And the results of this experiment are pretty much in line with the previous one.

    Somehow URLs with less folders tend to rank better.

    Sidenote.
    after seeing this data a few people asked us to study how "# of clicks" from Homepage correlates with Google rank. We've already did that and the results will be published soon here at Ahrefs blog.
    Recommendation
    Even though we did see a small correlation here with preference towards shorter URLs, I don't believe this is a significant enough ranking factor.

    I do recommend you to keep your URLs short, but mostly for user experience reasons and not because this will help you rank higher.

    6. Linking out to authority sites

    07-linking-to-DR70

    We’ve all heard Google say that outbound links are not a ranking factor, right?

    But at the same time we’ve seen some nice proof that they actually are.

    And our own experiment resulted in a small positive correlation of pages that link to DR70+ sites VS Google ranking.

    Does this mean this helps you rank better?

    Recommendation
    Even though we did see a positive correlation here, it is still very insignificant.

    Probably in some cases when Google is lacking other ranking signals, it will have no choice but to take this into account.

    But I don't believe that there's any way for you to use this tiny ranking factor to your advantage.

    7. Broken links

    12-broken-links

    Broken links are considered a "bad user experience", which Google might punish you for.

    Well, according to our data only 2% of pages that rank in Google top 10 have broken links.

    Recommendation
    Obviously I don’t recommend you to have broken links on your website.

    I don't believe that Google will penalize you for having broken links, but I'm pretty sure that your visitors will get frustrated once they happen to click one.

    8. Social Shares

    Ever since our beloved social networking sites started counting the number of shares that a page has generated, the SEO community got filled with rumours of whether or not Google is using these numbers as one of its ranking factors.

    So we decided to study how many of the pages that rank on the front page of Google have at least 1 social share on any of the major social networks.

    And here’s what we’ve got:

    13-social-shares

    Looks like roughly 70% of pages that are seen in Google top 10 have zero shares (based on our data sample).

    And yet there’s still a very small positive correlation between having at least one share and ranking higher.

    So then we dug a little deeper and studied the correlation between the # of social shares on each individual social network with Google rank.

    And here’s what we’ve got:

    10-social-correlations

    The number of social shares correlated with Google rank quite decently (compared to other on page factors that we've studied).

    But let’s not forget that correlation ≠ causation and it could easily be the other way around: people share these pages more often, because they rank high in Google.

    Recommendation
    It’s very hard to prove that the number of social shares has any influence on rankings.

    Our data (as well as our guts) do tell us that this could really be a nice ranking factor, but at the same time I do admit that it is too easy to fake.

    Relevance

    The relevance of your page to a search query is something that we didn't study at this time, but we do believe it plays a much more significant role than simply using an "exact match keyword" on your page.

    Back to my example of SERP for the keyword "guest posting":

    guest-posting-serp

    Clearly Google understands that "guest posting" and "guest blogging" is pretty much the same thing and that is why when you google for "guest posting", "guest blogging" and "guest writing" - you'll see pretty much the same pages occupy top10 results (but in slightly different order).

    Relevance beats backlinks.

    Here's a cool case that was brought up while discussing "on page SEO VS backlinks" the trenches.

    Search Google for "chocolate lab" (Total search volume: 74,000):

    18-chocolate-lab

    How come pages with 19 and 26 referring domains outrank pages with 278 and 1100 referring domains?

    Because they are more relevant!

    The first two pages rank so high with less backlinks because they are talking specifically about "chocolate labs".

    I mean both these articles are entirely dedicated to THIS specific dog breed of THIS specific colour.

    What about the other two articles?

    These are general pages about labrador retriever dog breed that mention "chocolate labs" fleetingly.

    Google only puts them on it's front page because there's not much relevant content that would talk about "chocolate labs" specifically.

    And these two pages have too many backlinks to ignore them, even though they are too general.

    I like to call this kind of scenario a "lack of relevant content in SERP".

    But how does Google know if your article is relevant to a search query or not?

    And, most importantly, how can you make sure you're sending the right "signals" to Google?

    This article has some good answers: "More than Keywords: 7 Concepts of Advanced On-Page SEO"

    Exact match keyword optimization works when there's a lack of content and backlinks.

    Let's go back to that experiment by Distilled that I've mentioned earlier.

    Please review a few important details behind this experiment:

    #1 They were re-optimizing ~10,000 category pages.

    Obviously, 99% of these pages don't have any backlinks pointing at them.

    You can easily check that in Ahrefs:

    19-best-by-links

    And the pages that they are competing with in Google SERP are pretty much in the same situation.

    (everyone knows the challenges of building backlinks to tens of thousands of category pages, right?)

    So that's a "no backlinks" battle.

    #2 All the keywords that they are targeting are "long tail":

    20-long-lail-keywords

    There's no secret that long tail keywords are (generally) much-much less competitive than keywords with higher search volume.

    So that's a "long tail keywords" battle.

    #3 These category pages don't have any relevant content that Google could "read" and "understand".

    Title, URL and H1 is pretty much all clues that Google has about these pages:

    16-h1-tag-hotels

    So that is also a "no relevant content" battle.

    And under these circumstances (no links, no content, long tail keywords) - it doesn't surprise me that Google will refer to appearance of exact match keyword in Title, URL and H1 tags.

    Back to you

    This is it! We've shared all we got.

    And more than anything we would love to hear what you guys think about it.

    I know there are a lot of very experienced SEOs among our readers and I'm super excited to hear what you have to say about our research and the takeaways from it.

    Also, note that the takeaways are based on analyzing the (very big) data and my own personal experience in SEO.

    Ultimately however, they are my own opinions and I am happy to discuss where I might be wrong, or where the data may be open to other interpretations.

    Do you have any interesting data or case studies that contradict our research? I'll be happy to take a look!

    See you in the comments!

    Tim Soulo
    Tim is the guy responsible for marketing and product development at Ahrefs. But most importantly he’s the chief evangelist of the company.

    Get notified of new articles

    51,255 marketers are already subscribed to Ahrefs blog. Leave your email to get our weekly newsletter.

    • As someone who has always been a big advocate of technical SEO I must admit that some of the takeaways in this post don’t sit comfortably with me. With that being said — data is data.

      What is clear here, and what makes perfect sense considering Google’s move towards better understanding meaning of content + user experience etc is that the traditional signals are becoming less and less relevant. It is also clear that Google still relies heavily on link popularity to rank web pages, which despite what they say from time to time, I guess we all knew anyway.

      I agree with Tim, that in terms of simply adding some keywords in strategic places, and expect to rank because of perfect optimisation — those days are over.

      However:

      I have anecdotal evidence of numerous times I have conducted a technical audit on a client’s site, implemented some small changes and seen improvements in rankings/search traffic. Granted, many times this is in Ecommerce where keywords are more long tail (as Tim intimated).

      My personal belief is that all things being equal, the small stuff makes a difference. So if you are sitting at position 5, have roughly the same links, content depth, and domain authority as the site at position 4 — then perfectly optimising a page may jump you up a spot.

      But only when the other metrics are similar (which I guess is rare).

      I also believe that on-page can get you an invite to the wedding, just not a seat at the table for dinner. So perfect on-page (without over-optimising) can get you to page 2/3, but is probably not going to be enough to crack page 1. Which ultimately, is where we all want to be.

      So, as Tim said I want to run some experiments on this.

      My first thoughts are to:

      1. Perfectly optimise some existing pages (and see if they go up)
      2. De-optimise some existing pages (and see if they go down)

      But I’d love to get some feedback on other ways we can test this. And if anyone has a sacrificial lamb page/site that we can use in the test (we’ll be using some pages on Ahrefs, but it would be cool to use some other sites too) then please drop me an email, or leave a comment.

      So, at this stage… consider me sitting on the fence on this one. Sorry, I hate to do that, but I’m currently an on-page SEO agnostic.

      Looking forward to your comments!

      • As per our Skype discussion, I myself find it real hard to believe that putting keyword in Title, H1, Content, etc no longer has any significant effect on rankings.

        But like you said — “data is data”.

        And data suggests that “On Page SEO” plays a super minor role in ranking.

        So I’m looking forward to our experiment…

        And I’m looking forward to any arguments that other SEOs will suggest in this comment thread 🙂

        • yup 🙂

        • Miloš Milosavljević

          But are you taking into account signals that encompass user experience once someone clicks on a search result. Bounce Rate, Time on Page, Time on Site, Pages per Session. Because this is the on site data that will show the impact of your destination page content and surrounding (domain level) content relative to the search result (intent), much more than optimising keywords and other tech elements on the page does. The real issue is to what extent do metrics that signal user experience and activity on a web page and web site that derive from a click on a search result affect rankings. This is what you should be measuring.

          • don’t these only kick-in if you get to the Google top10 first and start getting these clicks?

            • No — organic clicks are just one of many traffic channels where Google measures the on-page behavioral signals. We’ve seen an indirect relationship with PPC bidding and organic rankings. The same applies with traffic from social channels, like crowd sourcing. Rand Fishkin has done several tests that baffles Google for a little while and rankings shoot up, then back down again after the traffic settles down. Let’s not forget that optimising your page is important for users not only bots, and improving internal metrics like bounce rates, time on site/page and number of pages per session helps massively with SEO.

            • search traffic is VERY different from any other form of traffic in terms of visitor behaviour

              I’ve seen cases where Google would put some page to top10 for a while — probably to test how people will react…

              but as for tracking behaviour from other traffic sources.. not sure about that.. got any good read or patents on this one?

      • I agree with David, also term on “data is data”. As of he & many SEOs have been seeing positive impact through their Onpage SEO

      • SEO is pretty much like trading the markets (I do both!). In both cases, if “going technical” and “applying technical due diligence” were enough, everybody would win and rank at the top. If something can be reduced to easily repeated / imitated actions and technology, then everybody will eventually do exactly it. This is how the huge, pseudo-automated huge copywriting companies of old operated. Google found out this and nuked the hell of them.

        Like for markets, once somebody finds the honey pot, the market maker (Google in this case) shuffles stuff around until the honey pot becomes less of a gold mine.

        Just give up.

        Stop working 90% of the time trying to “game” the system, implement just common sense SEO basics and put all that effort into delivering quality content.

        • thanks for the comment, Dario. I also do think that SEOs should care less about targeting a specific keyword and care more about producing the kind of content that would benefit searchers

    • ViperChill

      Hey Tim,

      I’m never a fan of these type of headlines but you have written a really great article here. 

      I can’t imagine (well, OK I kind of can) how long it must have taken to put together.

      I’m pretty jealous that you guys have access to so much data so that you can put these kind of reports together.

      There are a few examples I don’t really agree with — Moz definitely have on-page SEO in their title tag, it’s just not together — but the data is fascinating.

      Would love to see you guys doing more articles like this with the massive amounts of data you have available.

      Thanks for sharing this. No doubt it’s going to get the SEO world talking 🙂

      I’m going to go and give it another read later to take it all in.

      • ViperChill

        I forgot to add, one thing that we probably don’t give Google enough credit for is how they’re able to discern queries could mean the same thing but worded differently

        For instance your example of guest posting returning results for guest blogging. I would argue that on-page SEO is still at play here, but besides that, it’s more a testimony to Google’s engineers.

    • Felix

      I am seeing you position 2 for — seo link tool. Not positon 1 ( ಠ_ಠ)–Ψ

      • I guess it was my “personalized” search 😉

        • Felix

          😉

        • Not to be an ass, but are you saying you allowed your personalized search history to affect all these results? Ie. didn’t use an incognito/private browsing session?

          • Felix

            No, this is just a query he did on his computer. Most likely in incognito so he could get a screenshot for the blog post. Their crawler wont display the css

            • But there are numerous deductions in this post based on Google SERP screenshots. Those are what I’m referring to (not that anyone’s individual search, incognito or not, represent “the official rankings” so I guess it’s maybe a moot point).

            • Felix

              Personally, if they have done the search correctly there would be no issue in display screenshots this way.

              Fresh proxy US. Incognito search on google.com accurate result.

              I believe this is how they did it. But Tim saying it a was personalized search says otherwise.

            • most of the screenshots in this post were indeed made in incognito mode using US proxy (and I trimmed ads in photoshop, to make them smaller in height).. 

              but this one doesn’t talk about ranking.. it talks about what is displayed in ahrefs.com meta description, so I didn’t bother doing it in incognito mode

              and obviously all data for 2M+ keywords was collected by our crawler, not by myself. So it can’t be “personalized”.

    • Damarco Lampkin

      Nice post! This was my first read but I’ll have to re-read with out skimming. Sorry the title of this post got me so excited I read it during my morning workout at the gym.

      I’ll leave an updated comment after the re-read.

    • Andy Drinkwater — iNET SEO

      For some time now, I have been saying that basic old on-page SEO doesn’t work as the primary ranking factor any more. It makes sense that Google would move away from some of these elements because they are too easy to game — why would Google rank a page just because it had a keyword in the title?

      Today, SEO for my clients focuses on content, LSI keywords / synonyms (had great results here), backlinks and making sure the site health is good (technical audit to confirm this).

      I can’t wait to see the case study produced — no time for you to be chatting here 😉

      Andy

      • Thanks, @andydrinkwaterinetseo:disqus 🙂

    • Keeping in mind that #1 rankingfactor is Links, #2 raningfactor is content does mean that there must be a correlation with onpage seo. No onpage optimization, no “good content”.

    • I think you make an excellent point in the “chocolate labs” example — SEOs should look for a “lack of relevant search results” as an opportunity for page 1 rankings vs. exploring the superficial elements like title, meta, headers…but this requires a little more legwork to see whether the content that is ranking lacks relevance. This way, while page 1 SERPs may seem dominated by high ranking domains with substantial backlinks, you’ll better know if there is an opportunity to create and rank highly relevant content. Oh, and I hope you get to page one for “On Page SEO” now 😉

      • > Oh, and I hope you get to page one for “On Page SEO” now 😉
        DONE! 😉

    • Hi Tim, first off, epic freakin’ content. This is a topic that is always top of mind in the SEO world. So thanks for tapping into your VLDS and providing us with intel that we could not get on our own. 

      My experience with this data correlates with what you have posted quite well..when the Domain and Page Authority is > 50 and the Referring Domains (RD) are > 1,000, with 150:1 RD to Backlink ratio. 

      There is no argument in the community that off-page SEO factors carry a much stronger signal for ranking. At the DA/PA et al levels, I referenced, on-page elements, based on my limited data, simply becomes noise, and its correlation to ranking factor disappears. 

      However, when the Domain Authority + Page Authority drop below 40 and 20 respectively, the C.P.C. (average) increases above $ 15.00, and the keyword volume < 200 / keyword, then I see the on-page elements you have analyzed having a real, and tangible impact. 

      I see this occurring in the legal, health, construction, financial, insurance, and home remodeling industries where you have high value, and low volume geo-targeted keywords. 

      Individually these sites and markets are tiny compared to your DoX parameters. However, collectively, this market (low volume, high value traffic) is huge. 

      I would consider a 2nd test where you filter the results by putting limits (a) on the Domain and Page Authority, (b) referring domains, © backlinks, and (d) KW volume, and see if your current on-page SEO correlation results change or stay the same. I am not sure if your data has any C.P.C. data to incorporate, but if it does, this would be another important variable to include as a 3rd test. 

      Are the on-page ranking factors being lost in the DA+PA+RD+BL noise? Don’t know. You have the data. Thanks again for an epic post! 

      P.S. I have not read the Distilled report you referenced.

      • thanks for the comment, Joe!

        just wanted to note that DA and PA are metrics of Moz and we didn’t study them in this research.

        We’ve only studied DR and UR, which are Ahrefs’ own metrics 😉

        • Hey Tim,

          My mistake…the values I referenced in my comment are DR + UR units.

    • Rony Daniel

      @timsoulo:disqus What a wealth of information here. Have you seen any SERPS that contradict this data?

      • Not really. “Exact match keyword targeting” doesn’t seem to give any advantage anymore. Topical relevance is the key

    • Deacyde do hurtin

      What a great article Tim, it really drives home what I’ve been seeing in SERPs from competitors as well as myself. 

      Backlinks reign supreme in a world obsessed about popularity, and of course not just any backlinks, the head nods from the ” Cool Kids “. 

      I’ve wondered though, will this have an overall effect in the quality of SERPs over the years?

      On-page SEO is becoming like W3C guidelines, mostly ignored. 

      Thanks for such a great post.

    • Stephen Floyd

      Nice link bait! another great article that proves that correlation does not equal causation.…

      • exactly

      • Matt Smiles

        They would have spent a lot of time on that, not exactly going to give that away for free (or at all).

    • Thank you very much for share..

    • Steve Bee

      RE: #8 Keyword in Content

      Without keywords, how would Google know what the page is about?

      • Jacob Maslow

        Tim was referring to relevant content but not a true match to your search term. 

        So google is taking the user query and finding relevant content even if that content doesnt have the exact keyword you searched for but has similar keywords. 

        For example, if I search “does linkbuilding work”, that phrase is in this post but google can tell based on what is on this post and by looking at the pages linking to this post.

    • Greg Morrison

      Cool data!

      But as Glen said above, guest posting and guest blogging are essentially being treated as the same thing…and Google loves when you have use multiple variations. For example the kissmetrics site does have guest blogging in the title and URL…but they also have “guest posting” 23 times in their content.

      And URLs are still highlighted (even the LSI variations like guest blogging for guest posting).

      One thing I highly disagree on is the power of EMDs…they are extremely powerful. Of course there’s certain niches they just won’t work on (specifically certain financial and health niches). Google wants the high authority sites (trusted sources) ranking for information that can seriously effect someones health or financial status (and I’m sure other things). 

      I remember when I was ranking for big financial terms specifically in the precious metals niche, and people were making HUGE financial decisions based off my affiliate style review sites, where I am no where near an expert in this (so I totally agree and understand what they’re doing here).

      But EMDs are absolutely crushing it still…that’s got to be the biggest thing I disagree with here.

      Also one thing that can be throwing off data is that a lot of URLs that were analyzed were inner URLs…we know keyword anchor text going to the homepage can greatly influence how the inner pages rank.

      In the end I agree and have always agreed with the overall “moral of the story” that links are greater than on page.….and they always have been! 

      On page is way easier to manipulate than off page, which is why it will always likely be inferior to off page.

      Definitely some good data though, especially the damn age of pages…how annoying that can be 😉

    • I love technical SEO and when I clicked the title I thought this was just a clickbait article.

      I honestly feel that Google AI is learning is becoming more semantic and topical. Its no longer about just keywords in headings, titles or anywhere else.

      I was a little disappointed that internal linking or anchor text was not talked about much here, as well as semantics, synonyms or word variations that show relevance but still debunking on-page.

      Super jealous of the amount of data and found this study very interesting. The correlation with google plus, folders, url size and content was insightful. I’m happy you guys spoke about this.

      Great article and I look forward to the next one.

      🙂

      • hey @mattgdavison:disqus , we’re hoping to do a follow up study mentioning internal linking, anchor texts and all the rest of the factors 😉

    • Nirmala Santhakumar

      Awesome post Tim, you have nailed the topic.

      Its good to know the power of backlinks and in-depth content. I agree with your viewpoints on optimizing the post to support On-Page SEO.

      Thanks for contributiing this awesome piece of content for us.

    • You have successfully proved that “On Page SEO” is dead. Okay, I am here not to debate with you. I have checked all of your proof and those are correct. So, there is no way to deny you. But, there is a lot of marketers who depends on ON PAGE SEO. So, to do a favor to the marketers, please explain in details that what should we do to rank up our site. And one more thing, If my target keyword is “On Page SEO”, but if I do not optimize it on page, then what should be the process to target this keyword in my content? Thanks Dear TIM.

    • Hey Tim

      Great piece. I pretty much agree with what you are saying here and that there is no need for daft old school practices of cramming a page with a specific keyword (even at the expense of connective words for readability etc). Unfortunately we still see a lot of page titles optimised like:

      Keyword 1 | Keyword 2 | Keyword 1 Variation | Keyword 2 Variation 

      So, it’s kind of not surprising that these are not exactly high quality pages. I think the whole philosophy that you can just apply some on page SEO magic is damaging in and of itself — the on page SEO should be thought of as something that you sprinkle over at the end. 

      My thinking is that it is more about topics — there are lots of ways to ask the same question. Search terms also tend to be abbreviated versions of questions. So, you kind of have to consider synonyms and intent rather than just search terms. Historically sure, but in our experience we are way beyond that now. 

      I just searched for:
      Hotels in Rome 

      Page title of the #1 site (TripAdvisor)
      The 10 Best Rome Hotels — TripAdvisor

      So, not the same exact keyword but the same intent behind the search. Variations on that theme also take us to the same site:

      best hotels rome
      rome hotels
      hotels rome
      etc

      Ultimately the best result is the best result and it does not matter how I ask the question — relevancy and authority is key here. 

      It’s highly likely that 90% of the search queries out there deliver traffic to 10% of the sites. These hugely authoritative sites are not so reliant on on-page optimisation and certainly not these old fashioned on page hacks. Relevance to the search query though is still absolutely critical and authority does the rest. 

      All that said there is also a big portion of the web that does not have much in the way of authority. Once you get down into the small / local business world of sites that are not doing content marketing then these traditional optimisation techniques like page speed and sensible on page optimisation can make all the difference. This is much like the long tail situation you describe where relevance and sensible on-page optimisation is critical. I guess the small / local business world is the long tail of the web! 

      On page optimisation is not dead — it is just different. It’s more a case of just being the best answer for a given query and applying some sensible optimisation (that is not just an exact match keyword in the page title & h1). 

      Great post though. Hopefully this is a nice full stop on some of the brainless SEO tactics of years gone by. 

      Cheers

      • amazing comment, @marcusbowlerhat:disqus !!

        That’s exactly what I was trying to say with my article!

    • I wouldn’t mind never hearing, “content is king.”

    • Another entry poking seat. We agree 100% with the theses spiked in the article. SEO is changing, and we hope for the better 🙂

    • On-page is dead? Adjusted keyword density, Title Tags, and edited urls. That’s it. I do this every day of my life, so Tim, what say you?

      • Jacob Maslow

        Dentist is a local search so the competition level is different. proximity is a huge factor. 

        Like it was pointed out in Claim # 3 (near bottom of a very long post) on page can make a big difference if there isnt much backlinks.

        When you are competing on a national level, there is a lot more sites that are competing with you who have plenty of relevant content and links.

        • And how many of those sites that rank for those competitive terms have bad on page SEO for those terms? I am not talking about exact matches, which is a really dumb thing to even consider these days, but having their on-page on point? 

          Look at any of the examples and they have all basically got their on-page right, which is why the only thing to separate them is the relative strength of their backlink profile.

          So the results have shown what happens when you control for on-page but for what is said about on-page not being a factor to be correct you need to control for the backlink profile, ie show that sites with similar backlink profiles but differing on-page do not show a similar ranking correlation.

      • Matt Smiles

        Yeah, let’s not talk about keyword density…

    • Great study. IMO every “on page” SEO element has a user experience component which almost always outweighs any perceived SEO value. On page SEO will always be mission critical, but is becoming far less of a ranking factor. Maybe it needs a new acronym. Just OPO? Then SEO becomes purely link graph and social engagement?

      • I agree Mike, UX (and completely satisfying search query) > any individual on-page factor. However, that being said, I still believe it’s best to follow best practices and give Google as much help as you can.

    • Spartan Tanvir

      Great article for the SEO world!! Really great job, Tim… waiting for your ‘follow up case study”
      But what about infographics, using of modifiers, using multimedia contents? Can you plz tell about these factors ?
      also waiting to hear what Brian Dean says about this ‘killer article” 🙂

    • Maybe onpage SEO as we know it is “dead” in the sense that ranking sheerly based on whatever you put into your title, meta description, and content in the form of keywords is just not enough to move the needle (any SEO worth his salt would concede to that much…I mean, Google’s measuring 200+ different ranking factors, is it not?), but what if Google’s simply adjusting for results that just prove to be more topically relevant (which you admitted) and more favorable to users? Wouldn’t that expand the definition of what onpage SEO is beyond just placing your keywords in a few key places and calling it a day?

      • hey @maximillian_heth:disqus
        Exactly so!
        I already admitted that I toggled the definition of “on page SEO” to be able to say it was dead 🙂
        Obviously topically relevant content is a huge ranking factor.

        • Yes, you did. To be fair though, I did publish my comment before you changed the title, hence the context. Perhaps we should start calling on-page SEO “on-topic” SEO (pun intended) instead. =)

    • Hey Tim I have to say your skillful use of wording to get me (and most others I’d say) to click through and then to keep us reading was pretty well played, blogging is definitely a skill I must work on myself, something I can’t help but notice though about the post and I don’t think anyone else mentioned it, is either very interesting and part of a broader study which will be in your follow up post, about “ON PAGE SEO” or ???. Would love if you could hint in anyway as to if I’m making any sense to you, if not hopefully the update post will make more sense to me, very interesting data to have cheers.

    • OK a couple of things that fly in the face of your conclusions

      1) For low competition keywords you can rank with few or no backlinks.

      2) If you have a page about car leasing and throw a load of links about children’s clothing at it, it will not rank for either car leasing or children’s clothing even if those links are from highly authoritative sources.

      The on-page factors tell Google what your page has the credentials to rank for. If this were not true the sites with the most backlinks would rank for everything. So while I do agree that the on-page factors are not the thing that drives your rankings in higher competition keywords, but quite simply, if you don’t have them, you don’t rank at all for those keywords. 

      From my experience it seems that Google devalues links that are not seen as “relevant” to your content, which is why fixing on-page issues can see a page increase the rankings as a few others have pointed out — you are aligning your on-page with your off-page SEO essentially.

      We also know that with too much optimization you also get penalized and we also know that Google understands synonyms — kid’s clothes will also rank for children’s clothes for example — and other LSI stuff. This is not 2011, so the idea that your on-page has to match a keyword exactly seems a little…well, naive at best.

      You have found data to support your arguments, but we all know to be a valid hypothesis you also need to show the lack of evidence to refute your claims too.

      The problem I see with this article is that you have essentially only looked at pages that have already passed Google’s on-page SEO filter to allow them anywhere near the top of the SERPs for their relevant search terms. 

      I would be really interested to see some comparative data of pages with similar backlink/anchor text profiles and see how differently they rank and then compare their on-page SEO. If you can find no correlation there, then you might be on to something.

      • Jacob Maslow

        As to your first point, Tim covered that in “claim # 3”. That was towards the bottom of a really long post. 

        As to your second point, the essence of the article is that you dont have to worry about H1 tags, Titles and the like. Google can recognize what your content is about and rank you for relevant terms even if it isnt in your title tags. 

        Tim does mention how important relevance is.

        • It just doesnt make sense. It is basically aying the only thing that matters is backlink relevency but if that was the case pages would only rank for them and not what is on the page. Its clearly a mix of both still. Nothing can disprove that. This article is biased as they have something to sell.

          • hey @GaryCarlyle:disqus , did you read the part about “relevance beating backlinks”?

            if you see that top10 results are not very relevant to search query — you can rank there with little to no backlinks, if your page would be more relevant then what’s already there

            BUT

            if top10 results are perfectly relevant to a search query, no on-page SEO trick will help you outrank them. You’ll have to work on your backlinks

            agreed?

        • Says the opposite in the title and does nothing to actually test it though — he only tests the positive hypothesis, not the negative

          • yep.. the Title was this article was made up purely for controversy. That backfired, so I re-wrote the post and avoided controversial claims. Hopefully now my message and the thought process behind it is fairly straightforward

        • thanks a lot @jacobmaslow:disqus
          Way too many people judged the article purely by it’s headline, without actually reading it from start and to the end.
          I do admit that I oversimplified the definition of “on page seo”, but I never said that “relevance” doesn’t work.

    • Boni

      Every one that has dealt with SEO has “felt” this article in his skin. Way to go.

    • This is (and excuse my French) bullshit.. It’s a clickbait title based on the entire wrong idea.

      Firstly, OnPage SEO isn’t purely about optimizing a page for a keyword, because Google also class synonyms as a bonus — Google “automotive” and a bunch of “car” related terms will pop up…

      Secondly, OnPage can also relate to the structure of your site, your interlinking work, other relevant pages on your site (topical relevance of a site), your sitemap setup, formatting/mark up and so, so many other factors.

      I understand you won’t clicks, but the title alone could damage a lot of good SEO agencies.

      • Well said — look at any of the examples and they have all basically got their on-page right, which is why the only thing to separate them is the relative strength of their backlink profile.

        Now if they can show that sites with similar backlink profiles but differing on-page do not show a similar ranking correlation I might take it seriously

      • I partially agree with you. The problem here is each factor has their proportion contributing to rank a website (both of on and offpage) and we can’t separate each of them so it makes this analysis somewhat inaccurate.

    • Juha Sompinmäki

      That’s pretty damn epic research and hard to go against as most of us do not have access to similar data. I would also not go as far as saying “seo is dead” but certainly the on-page factors have become really minuscule on a grand scale of things. Backlinks are backlinks, but if you would take that away, it’s certainly interesting to wonder how Google “understands” the page content and rank the results.

      • Tim’s main message was that no single factor is going to be enough to give you a competitive advantage when compared to the “big” issues. That being said, I still recommend following best practices and giving Google as much to go on as possible.

    • Joe Kizlauskas

      I can 100% agree and 100% disagree with this post. 

      RankBrain is still learning, so when it’s in niches whereby it has a good understanding then the above if true. 

      I am killing it in certain niches with good old fashioned On Page SEO

      It’s an algorithm not a human. It can only guess at searching intent whereas as a human we know the intent far better than google. 

      Common sense is not something you can engineer.

      So my advise is study the niche. See if RankBrain is working its magic. Try and find those niches where it isn’t.

      • That’s good advice Joe and I would agree that there are some niches (generally less competitive ones) where on-page factors still seem to have more influence.

        • Joe Kizlauskas

          One strategy is to scoop up the pennies. 

          Go for one KW that had 1500 searches a month. Harder to rank and full of high authority sites. 

          Or 10 articles that have 150 searches a month. 

      • hey @joekizlauskas:disqus
        > You are talking about one aspect of On Page SEO.
        Agreed! The study was mostly focused on “exact match keyword targeting”.

        • Joe Kizlauskas

          I can guarantee that if you did a webpage on how to play the game Mario. 

          Then I came along and created a webpage to target the following KW 

          MARIO WALKTHROUGH WITHOUT LOSINGLIFE

          Did that title in my URL created a H1 title and dropped that phrase in the the text I would outrank you. 

    • Edwin

      Well, I must admit this is what actually happen in 2016. Based my experience, my website was pushed to rank 2 since the last month, no matter what I did, I can’t go to the first rank because of that site. Then I check that page onpage, guess what, no keyword in url at all, no keyword in the title, even no backlinks at all. wtf I use PBN, expired web 2.0 and guest post as well. Ok then I examine a little further. What I got is the site is using https and the page has 1000 words along with 9 exact keywords in the content. So, I think very long content along with https will give good results nowadays. Correct me if I’m wrong guys.

      • > very long content along with https will give good results 

        Long content is not the answer. Definitive, “best” content for a query is the answer — whether that is 500 words or 5,000. By their nature, some queries (such as “on-page seo”) take a good few thousand words to properly explain, hence the impression that longer content = better. But it depends on the query.

    • Rafał Grochala

      Many SEO correlation “studies” are bad at interpretation part. You’ve done it almost right. I don’t think that correlation equal to zero, especially in such high noise environment, can show that there is no cause and effect relationship. In fact there is a relationship for example with keywords in content and position. Also keyword density is influential factor. We see it in our (SEO company) internal experiments, where we work on isolated search keywords that are made-up and never been in the internet (0 results), with fresh domains isolated from content/theme or other factors, and of course with control groups and many repeat cycles. Results clearly show that domains with many keywords in content outrank domains with less keywords or no keywords. I can hypothesize that correlation doesn’t show this because of noise from other unisolated factors (title, h1, backlinks hidden from ahrefs etc.). Maybe it’s also about exact and partial match division. In our experience and experiments we see that generally Google threats equally both matches.

      And question about study design — why correlation beetwen 1–2-3–4-5–6-7–8-9–10 positions and not beetwen 1:3–4:6–7:10–11:13–14:16–17:20 or 1:5–6:10–11:15–16:20–21:25–26:30? In top10 alone variation is quite low, and so correlations are very low (nearly equal to zero).

    • Just the right article every SEO is looking for as it tells everything about on page optimization and how it becomes less of a ranking factor nowadays as Google has been updating every now and then and it is actually affecting the SEO world. 

      Over all, this is indeed a killer article!

      Cyrel Nicolas of http://resultdrivenseo.com.au

    • I am glad you changed the title Tim. I have always believed that when you make it easy for the search engines to do its work and without being too manipulative, it will reward you with better rankings.

    • Usama Tahir

      Hello sir,
      Such a nice information but kindly tell me i am working on Gaming site can i rank my keyword with out any article by just doing OFF-PAGE SEO ???

      • I would still advise you follow best practices and give Google as much of an indication as possible as to the topic of your page. That being said, content quality, or more specifically solid user experience probably trumps any individual on-page measure.

        • Usama Tahir

          I’m not under standing sorry to say but can you explain more????

    • but sometimes, without any seo has blog in first page in serp

      • ☆THĒ_ĒMÏLYGAMÏ☆

        Ty for following me ^^

    • @timsoulo:disqus Did you consider you might have got it backwards? I see a pretty clear correlation between using these factors and ranking in position 1 in pretty much every graph?

      • @john_e_lincoln:disqus if you’re talking about the headline that says “on page SEO is dead” — that was definitely made up 🙂

        if you’re talking about the importance of exact match keyword in Title/H1/content/etc — the correlations are super small to rely on this as your competitive advantage

        so I’d say “exact match keyword targeting is dead”

        • What I was saying is that if you look at your graphs the ones in the number one position are using the techniques that you’re arguing against.

          Actually do agree with you though that I don’t think that you need exact match.

          I think the one thing that made this kind of tough for me to swallow was saying that the sides had similar backlink profiles. The main reason being that nobody knows the real quality of a backlink except for Google. External links are so important that a couple small differences there could be causing all of these alterations in search results opposed to the on-page factors that you’re pointing out here.

    • Zakaria Desai

      Three thumbs up. On page seo for 2016 and beyond explained beautifully 

    • We SEO’ers are thé top-mechanics in a formula1/nescar –race. Thinking,creating, optimizing,testing before the driver takea it seat and let him/her go into the race with millions to get first over the finishline. Over and over again. We — the mechanics — working on On-Page and the Driver on Off-page, or must i say Pace. At at a certain moment we have to let the driver go making his rounds and trust him/her to do its utmost best and not crashing in any way. That being said: 

      We need to make the fundamentals of a site topclass. We need to make the driver understand how the track is running with the time, knowledge, experience and determination (and money?) we have. To get more gain and less pain we work with our clients on a themed-bases-and-authority-based seo. First the low-hanging-fruit (with quite conversion-promising) longtailers and if we feel the right pace on the track (google) we are starting to work on the dream-keywords and create as much as authority as possible. But only then we feel some pace. 

      We still believe that Google is a “stupid” machine with high-intellectual talents but Google has made his own playing-field (track) so stay between the borders. Look at ur competition (not top10) and try first to get in top100 ehile the longtailers makes u smile.

      Trully a great piece of content and would l9be to read the next where a correlations within a correlation is extracted. “If we mix factor 1 and 2 and 3” what happens s than. Regards mark eurlings Opmax Online Marketing and more.

    • SEO has change after a time. Thanks for sharing a latest seo trends

    • Ito Martinez Alonso

      Why exact match keyword in URL is bad for seo?? I mean in the first image

      • It’s certainly not bad for seo, just doesn’t necessarily give much advantage

        • Ito Martinez Alonso

          I understand! Thank you!

    • Chris Frampton

      Really enjoying the article and discussions Tim. Just a quick question regarding the number of folders and how it impacts, obviously there will be more folders if you have your tracking implemented in the URLs. Is the potential cost of having more folders outweighed by the benefits of the analytics?

      • I think the # of folders in the url probably relates to the depth down the hierarchy. I don’t see that there would be a specific disadvantage to using folders in url (save for the fact that lots of them will look ugly)

      • Hey @disqus_2QItpPvAgL:disqus , I didn’t see Google indexing URLs with tracking and displaying them in it’s SERP. Got any examples?

    • Stacie Woodside

      As an SEO professional, I mostly agree with all of this, however, you didn’t go deep enough to diffentiate the algorithm for local vs. national. I would be eager to see how some of that data may (or may not) change.

    • Ótimas informações, vou altear o conteúdo de alguns artigos meus, porque tenho muitas palavras chaves nas tags.

    • Hello Guys,

      This article is enough but as we know that Google ranking depend on above 200+ conditions. So I would like to say that the keyword ranking depend on business category. Here all the example are related to seo study but if we compare to e-commerce website and tech support website, real estate website, tour and travel website and some other business then Google prefer top 10 ranking only old domain, huge back links, high authority, Branded website, Popular websites etc. And we all know these all are not job only 1 or 2 years. We have to invest time and and promote the business not only seo but we have to promote the website by PPC, TV Ads, new paper Ads, poster ads, B2B Ads and many more then we achieve top ranking in Google for business website. 

      Summary- I think. If visitors increase on website any ways then we could see the website on top 10. And now it’s your job how to increase visitors on website.You will have to do- well optimize, promotions, branding, and keep patience for business website. Thanks.

      If we want to promote blog then we have to write in detail and proper information on exact subject.

    • Akash kale

      Yes i can say one thing pretty clear here if you are writing content for user then one way or the other you will able to rank, i have experience Rank-brain quite a bit these day on my blog i.e http://www.techlessions.com.If i do decent on page seo with quality content and use of LSI keywords then i rank little better then stuff keywords.But The one factor which i found make pretty big difference in ranking i.e interlinking of page among your best post on the blog if you do that and relevent off page you willl bound to rank on top.

    • Ito Martinez Alonso

      Hello again! A doubt that i have is:
      — The antiquity of a domain is relevant due to it should have backlinks already, or simply for the time that you bought in the past?
      — I bought an old expired domain with only 2 backlinks, but register in 2002 for another user. Would i have this advantage for antiquity?

      Thank you in advance!

    • First of all thanks for sharing this valuable information with us! Learning from 2M kw is defiantly something worth doing. And I actually feel good now because I know I don’t need to optimize a page like crazy, because getting a good link to the page will generally help me much more.

      • thanks, Ilan! And the whole point is that you can only get a good link to your page if the content on that page is worth a good link 🙂

    • hey @disqus_yBANN5S3zd:disqus , the title of this article was fixed.. so hopefully there’s nothing left to disagree on 😉

    • Hi Tim,

      You say that websites are not using Keywords in the title tag or meta description that are ranking on the first page. Take the phrase “keyword research” as a example. Both the Moz guide and Brian Dean’s page have the keyword in the title, url and in the meta description.

    • Excellently informative article on the fundamentals of on page SEO Thank for sharing this important content. I got a good knowledge of on page.

    • Awesome post!

      But this post didn’t got to any conclusion for Best On-page practice. On the way reading down the post it sounds to me that #ahrefs want to get ranked in 1st position on Google for “on page seo”. 

      I was expecting for more , depth answer for on-page optimization. 

      Neverless it’s always good to read blog by #ahrefs

    • Matthew Kong

      A fantastic in depth guide to on page SEO and where its at now! Especially enjoyed the graphics demonstrating the correlation between your optimisation and your results.

    • My comment is in regards to using an exact match keyword for searcher intent and to meet searcher expectations. 

      While I get where you’re coming from here and that using exact match keywords is in no way a guarantee that you’ll even get close to ranking high on the search engines, I don’t see a problem with using, targeting and optimising for an exact match keyword IF it’s relevant and natural and is something that you would say in everyday conversation, plus better meets the searcher’s expectations. E.g. an exact match keyword like “food photography props” in my opinion, is natural. It’s natural in the sense that it’s a term used naturally in a sentence, conversation and/or written text without compromising readability. 

      I recently worked with a client where we identified a relevant keyword for his business that he has the potential to get traffic to his biz. But when we checked the top 10 search results, none of them had the exact match keyword in the title tag at all (which probably helps to validate your study). However, if I was a searcher or consumer legitimately looking for information on this topic using this exact match keyword (which has very high search demand BTW), then wouldn’t we better off using this exact match keyword to give the searchers exactly what they’re looking for and to better meet their needs? Sure we probably won’t rank on page 1 of Google in the short time just with onpage SEO alone and would need some relevant backlinks but if we eventually do rank for this keyword (which in the long term I am confident we can), then I’m confident our page would be the best one that meets the users needs because we created content that solely answers and goes deep on this topic. 

      With this particular keyword, we noticed that some of the other pages ranking for it don’t really have much relevance with this keyword at all and it appears that some of the pages ranking were not solely created to specifically provide an answer based on this keyword.

      • Totally agree that backlinks still have much more correlation and effect on ranking. On page factor is something you can control thus will have less effect on ranking than an actual back link. Very detailed blog post thanks again!

    • Great post about latest SEO trends. Actually i was not aware of Semantic. Thank you for mentioned about Semantic. When it comes to on-page SEO placing Keywords Meta tags is the almost out dated SEO. I mean to say it’s not only on page SEO. When ever it may be Unique and quality content on the webpage can really increase the ranking of a website (or) Webpage.

      • Matt Smiles

        How were you not aware of semantic search? Do you work in SEO?

        • I know what is Semantic search, but i don’t know this is what Semantic search.

    • Shiwangi Singh

      Hey it is nice blog thanks for information.

    • Jonathan Alderson

      As a minor point, it’s not an “alt tag”, it’s an alt attribute (i.e., the alternative text is an attribute of the image tag).

    • looks great! Thanks!
      Will double read it at home!

    • NCode Technologies

      A fantastic piece of content. It represent all the aspect of on-page SEO which effect your website ranking. The best part of this article is the way your have mentioned the recommendations in noticeable manner. It’s the best part of the blog as it directly has an advice which we can follow. I was under impression that Exact Match Keyword in Title, H1 and content plays a major roll in ranking the page. Thanks Tim for clarifying the topic.

    • does google update the SERP daily. i get different result on same day at different timing.

    • Aleksandr Lilik

      Great article. I have a question about “Folders”. Our company competitors usually have 2–3 folders for the urls they rank, do you think we should restructure our urls that have 3–5 folders and should we expect actual improvements from this?

    • Backlinks remain an extremely important Google ranking factor. We found the number of domains linking to a page correlated with rankings more than any other factor.

    • This was a very useful post. Thank you for creating. I’m particularly thankful that you focused on long term as opposed to fly by night strategies. It’s great to meet individuals teaching sound knowledge! I was sure to share it as well!

    • seotrick

      great post

    • Great tips! I will definitely use them going forward. It is so true that on page SEO is vital to get right and great to see such an in depth look at it!

    • Niranjan Pande

      thanks for sharing with us

    • Hello,

      Thanks for sharing these great techniques. Will definitely help in making a post perfect.

    • Ronald Dod

      Amazing article!

    • These are very important insights and detailed explanation you have given about keywords selection. It’s a very informative post..and a helping one.

    • Carl Garside

      Thank you for taking the time to do this research and sharing your results. As many other people on the comments have suggested, believing On-Page has a low importance in ranking is partially correct and at the same time incorrect. 

      Not everything is created equally and unless a good engineer comes here and provides evidence to what is and what is not important for ranking, i’m simply going to go with the masses (Yes, I am just a small sheep, in a very big field). 

      Never the less, Thank you so much & it was an interesting read.

    • Hakan Erdur

      Hi,

      Nice informative article, first of all thank you for your efforts.

      I am curious about the importance of keyword densities. Do keyword densities have an impact on SEO? Should we balance the density? like %1.2–1.5 for 1 word, %1 for 2 words, and %0.7 for 3 words?

      Thanks