1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

[How To] Keyword Density in 2014

Discussion in 'White Hat SEO' started by seocrab, Jan 30, 2014.

  1. seocrab

    seocrab Senior Member

    Joined:
    May 2, 2013
    Messages:
    969
    Likes Received:
    825
    Occupation:
    seo freelancer
    Location:
    UK
    I saw a post today about a site ranking with 8% keyword density so I collected some data to clear things up a bit:

    Keyword Density in 2014


    What does Google say about Keyword Density?

    They advise against "Loading pages with irrelevant words" (aka keyword stuffing).

    [Source:https://support.google.com/webmasters/answer/66358]

    In 2011 Matt C***s explained that keyword value is measured on a normative distribution curve; mention a word too few or too many times and it won't register. AKA there is an optimum density value!
    [Source: http://www.youtube.com/watch?v=Rk4qgQdp2UA]

    At the end of 2013, the Hummingbird Update placed emphasis on "Conversational Search", i.e. extracting meaning from a whole sentence, not just a few words. So how does that affect keyword density in 2014?

    What do the SERPs show us?

    Let's look at some of the top-ranking sites for the popular term "payday loans". This is a good phrase to test out because Google has already applied extra spam filters to the niche.
    [Source: http://searchengineland.com/google-pay-day-loan-algorithm-google-search-algorithm-update-to-target-spammy-queries-162941]

    Google.com Top 5 Results for "payday loans" (note: I am in the UK so results may differ)

    #1 Repeats 36, Word Count 4033, Density 1.69%
    #2 Repeats 97, Word Count 3632, Density 5.41%
    #3 Repeats 15, Word Count 1689, Density 1.78%
    #4 Repeats 15, Word Count 1474, Density 2.04%
    #5 Repeats 66, Word Count 11,771, Density 1.12%

    Result #2 stands out with a keyword density of over 5%.

    The above figures take into account TEXT DENSITY.

    I'll crunch the numbers again but this time taking into account all content (text, alt tags, links, meta data etc)

    #1 Density 2.36%
    #2 Density 3.25%
    #3 Density 2.69%
    #4 Density 1.57%
    #5 Density 1.35%

    Now Result #2 has a more reasonable keyword density of 3.25%

    Tentative conclusion: Google evaluates total keyword density over just text density.

    Based on this tiny sample, a total density between 2 and 3% seems preferable.

    Now onto a completely different niche:

    Google.com Top 6 Results for "learn to swim"

    total keyword density

    #1 Density 3.28%
    #2 Density 0.24%
    #3 Density 0.04%
    #4 Density 0%
    #5 Density 0%
    #6 Density 0.73%

    Aside from result #1 (a YouTube video) all pages have a total keyword density well below 1%, and two pages never mention the phrase at all.

    Why is the keyword density so low?

    While the density for the exact phrase "learn to swim" is low or non-existent, the Hummingbird effect can be seen at work here.
    [Source: http://www.searchenginepeople.com/blog/detailed-analysis-of-semantic-search-and-its-role-in-hummingbird-algorithm.html]

    Results are automatically picked (and highlighted) for the phrase "learning to swim" too:

    [​IMG]

    Results also show for the word "swimming".

    The most commonly found words on the top-ranking pages are:

    swimming, water, pool, training, swimmer, swim

    and phrases:

    how to swim, in the pool, in the water, learned to swim

    Between the Top 6 results, the phrases "in the water" and "how to swim" occurred much more frequently than the exact phrase "learn to swim".

    This is evidence of Latent Semantic Indexing (LSI) at work.

    How does Latent Semantic Indexing work?

    A massive volume of texts are scanned and a mathematical algorithm then calculates patterns and trends between words. This is how we know interesting facts like "the" is the most used word in the English language.

    Google use LSI as one of their many search algorithms. Using the above example, Google's LSI formula has determined that articles about "learn to swim" must also explain how to get "in the water" and teach us "how to swim".

    It's a form of artificial intelligence that looks set to get more sophisticated based on the $Billions Google is spending on acquiring AI technology. [Source:http://techcrunch.com/2014/01/26/google-deepmind/]

    LSI Further Reading: http://www.seobook.com/lsi/lsa_definition.htm

    What does this mean for Keyword Density in 2014?

    Here are some informed suggestions for creating top-ranking content and good on-page SEO in 2014:

    TIP #1:Evaluate total keyword density (not just visible text).

    HOW? Paste your site in one of the tools given below in the Resources list.

    WHAT PERCENTAGE? 1-3% still seems like a safe bet, but perform some competition analysis to double-check. Ultimately, it will depend on your keyword because the principles of LSI determine that your optimum keyword density is related to the density of other sites that use your keyword!

    TIP #2:Use long sentences, especially when including your "keyword".

    HOW? See the difference:

    BAD: Used cars are cheaper than new cars. People on a budget can benefit from used cars.
    GOOD: There are over 250 million used cars registered in the US, making the second-hand car market one of the largest in the country.

    TIP #3:Find and include terms related to your niche.

    HOW? There is a link to a free LSI generator tool in the Resources list. It also pays to analyse (again, with the tools given below) the top 5-10 results manually for extra ideas.

    Resources:

    Calculate text density using SEOquake (firefox plugin):
    http://www.seoquake.com/

    Calculate total keyword density:
    http://tools.seobook.com/general/keyword-density/

    LSI Generator:
    http://lsikeywords.com/
     
    • Thanks Thanks x 57
  2. ranga

    ranga Regular Member

    Joined:
    Aug 9, 2009
    Messages:
    268
    Likes Received:
    85
    Tentative conclusion: Google evaluates total keyword density over just text density.

    This conclusion is based on one bias, that Google evaluates Keyword Density for ranking purposes and that for a website to be ranking no 2 its keyword density should be in the Google range.

    Its like assuming a result and to conform to that result, what dataset is possible must be the answer. What if the keyword density range for this keyword would range from 2% to 12%, since nobody is over 4%, we would never know (this is after believing what Google says and I don't believe 80% of that)
     
  3. cataratas

    cataratas Junior Member

    Joined:
    Apr 13, 2011
    Messages:
    166
    Likes Received:
    75
    Everyone overthinks keyword density and whilst I did pick up something interesting in the OP about LSI, this is essentially overthinking maxed out.

    All as you need to do is write naturally. It's as simple as that. No more, No less. Density and LSI occurs naturally then as well in a highly effective manner.
     
  4. seocrab

    seocrab Senior Member

    Joined:
    May 2, 2013
    Messages:
    969
    Likes Received:
    825
    Occupation:
    seo freelancer
    Location:
    UK
    UPDATE:

    How do 'Readability Scores' affect Google rankings?

    'Readability' refers to how easy a text is to, well, read.

    Readability has 2 strands:

    1) How words are displayed
    2) How words are organised

    When we display text on our websites, we choose whether it's large or bold or in comic sans etc. There must be a preferred practice when it comes to pleasing Google. The better web readability project seems like a good place to start:

    [Source: https://code.google.com/p/better-web-readability-project/ ]

    Now onto how words are organised: there are many different algorithms that measure readability, but the Flesch Reading Ease Formula is most commonly used.

    Here's the algorithm:

    Code:
    Calculate the average number of words you use per sentence.
        Calculate the average number of syllables per word.
        Multiply the average number of syllables per word multiplied by 84.6 and subtract it from the average number of words multiplied by 1.015.
        Subtract the result from 206.835.
        Algorithm: 206.835 - (1.015 * average_words_sentence) - (84.6 * average_syllables_word)
    The types of books we might have learned to read with go like this:

    The above selection gets a Flesch Reading Ease score of 108.8, meaning that pretty much anyone can read it. As text gets more complicated, the score lowers.

    Here's an extract from a scientific journal:

    The above text gets a Flesch score of 24.4.

    Most websites are going to fall somewhere between these 2 values, but is this information useful for SEO?

    Firstly, I will search for broad topics and see if, as I suspect, the readability scores are high for the top results.

    Google.com Top 5 Results for "buy bananas"
    (ignoring YouTube/image results)

    #1 70.70
    #2 67.88
    #3 70.19
    #4 62.30
    #5 49.97

    Google.com Top 5 Results for "how to make a cake"

    #1 64.3
    #2 85.5
    #3 78.1
    #4 69.3
    #5 73.9

    Now for a more technical subject:

    Google.com Top 5 Results for "install new ram in pc"

    #1 76
    #2 71.5
    #3 53.45
    #4 56.05
    #5 53.63

    and something in the medical niche:

    Google.com Top 5 Results for "vitamin health benefits"

    #1 42.41
    #2 43.58
    #3 46.80
    #4 41.19
    #5 37.85

    It appears that some niches require more authoritative content with lower Flesch scores. For example, the health niche is less 'readable' than the cookery niche.

    Conclusion:

    For SEO, we need content that matches Google's niche and keyword expectations.

    We can figure out what G wants by researching top-ranking sites for our niche/keywords.


    Free Resources

    Readability test tool: http://read-able.com
    And another: http://juicystudio.com/services/readability.php

    You can check the readability of your domain by following these steps:

    1) Google "site:yourdomain.com" (without quotes)
    2) Select "Search Tool"
    3) Click "All Results" and change it to "Reading Level"

    You will now see an overview of your site's readability.

    [​IMG]
     
    • Thanks Thanks x 9
    Last edited: Jan 30, 2014
  5. seocrab

    seocrab Senior Member

    Joined:
    May 2, 2013
    Messages:
    969
    Likes Received:
    825
    Occupation:
    seo freelancer
    Location:
    UK
    Yes, if you're gifted with a strong grasp of the English language, sure! But not everyone knows how to write so eloquently and the natural style of some writers will be penalised because of a set of centralised standards.
     
    • Thanks Thanks x 1
  6. Serpable

    Serpable Jr. VIP Jr. VIP Premium Member

    Joined:
    Jun 13, 2013
    Messages:
    496
    Likes Received:
    85
    Occupation:
    Giraffe Riding
    Location:
    Uk
    Home Page:
    Okay great thankfully i always keep mine low.

    @ SeoCrab - With the changes toward conversation search what percentage do you aim for with you anchor text?? Its obviously been getting lower and lower over time and i would imagine that will continue. What do you class as the current safe ground?? Ive heard very different numbers.

    Edit * Cant go around calling him seo crap even if it is a mistype..
     
    • Thanks Thanks x 1
    Last edited: Jan 30, 2014
  7. ch8878

    ch8878 Elite Member

    Joined:
    Mar 21, 2009
    Messages:
    2,242
    Likes Received:
    428
    Gender:
    Male
    Occupation:
    Gamer
    Location:
    Youtube
    Home Page:
    LSI generator can help you add ideas to what to include into your articles.

    Thanks for sharing!
     
    • Thanks Thanks x 1
  8. matrix79

    matrix79 Junior Member

    Joined:
    Jan 20, 2013
    Messages:
    123
    Likes Received:
    27
    I wrote naturally for a while now, but it was more like shooting in the dark. Some of my content ranks and some doesn't. I tried to analyze SERPs myself and results are always very confusing. At times it seems like you almost should use only synonyms of your keyword and never mention the actual keyword to rank for it.
     
  9. PrinceVisi

    PrinceVisi Elite Member

    Joined:
    Jan 11, 2012
    Messages:
    1,916
    Likes Received:
    1,009
    Occupation:
    BusinessMan
    Location:
    Tropoja

    Although, I have my own way of doing things, I like the the work you have done.

    Good job and thanks for sharing.
     
  10. seocrab

    seocrab Senior Member

    Joined:
    May 2, 2013
    Messages:
    969
    Likes Received:
    825
    Occupation:
    seo freelancer
    Location:
    UK
    Be careful with synonyms; they are not actually that beneficial when it comes to LSI:

    Synonym = another word that means the same, i.e. warm = hot = boiling

    Related Terms = a word/phrase that often appears in the same context, i.e. warm = sunshine = beach
     
    • Thanks Thanks x 1
  11. seocrab

    seocrab Senior Member

    Joined:
    May 2, 2013
    Messages:
    969
    Likes Received:
    825
    Occupation:
    seo freelancer
    Location:
    UK
    SeoCrap - dude!

    I think there's no exact number we can quote. My tiny research sample shows wildly different results. If you want to be safe 2% seems ok but you have to do research into your own keyword to get a better idea.
     
    • Thanks Thanks x 1
  12. Heatz

    Heatz Jr. VIP Jr. VIP Premium Member

    Joined:
    May 6, 2011
    Messages:
    175
    Likes Received:
    150
    The Hummingbird effect you mentioned on the "learn to swim" are great examples and proofs about the way Google analyses and ranks naturally written content better over forced keyword articles.

    We just need to analyze the back-link profiles for each of the sites you mentioned to see if you can really rank with no backlinks on these cases.

    Anyway, thanks for sharing a really complete and well grounded post man.
     
    • Thanks Thanks x 1
  13. FutureProofSeo

    FutureProofSeo Jr. VIP Jr. VIP Premium Member UnGagged Attendee

    Joined:
    Jul 3, 2013
    Messages:
    921
    Likes Received:
    454
    Gender:
    Male
    Occupation:
    Google Warrior
    Location:
    UK
    Home Page:
    Haha that is a blatent freudian slip I think crab man ;)

    Nice article, time to test my niches I think!

    I also have a feeling that even within a niche, sub-niches there are different levels of KW density on page 1, I have a few examples of this where following the same format for a page I have in no1 spot, for another page just doesn't quite hit the mark (very annoying).

    FPS
     
    • Thanks Thanks x 1
  14. Serpable

    Serpable Jr. VIP Jr. VIP Premium Member

    Joined:
    Jun 13, 2013
    Messages:
    496
    Likes Received:
    85
    Occupation:
    Giraffe Riding
    Location:
    Uk
    Home Page:

    Ahh balls you go out of your way to help and i call you seo crap haha my bad... i suck. Im going to sit in the corner now im so ashamed. :07:
     
    • Thanks Thanks x 1
    Last edited: Jan 30, 2014
  15. totbweb

    totbweb Newbie

    Joined:
    May 9, 2013
    Messages:
    16
    Likes Received:
    3
    Thanks for sharing your thoughts. I think you're on point with looking after payday loans because they've mentioned it. Based on the way the convo is going, there is a way to auto scan text to determine a reading level too. I never really used it, but ms word has options to use the Flesch Reading Ease test. Would it be worthwhile to notice a pattern there too? Words reading level tests calculate:
     
  16. Aty

    Aty Jr. VIP Jr. VIP

    Joined:
    Jan 27, 2011
    Messages:
    5,411
    Likes Received:
    3,699
    Home Page:
  17. Reyone

    Reyone Elite Member

    Joined:
    Sep 30, 2012
    Messages:
    1,577
    Likes Received:
    1,277
    Occupation:
    Hunter
    Man, that's a fucking nice post. Thanks!
     
    • Thanks Thanks x 1
  18. Lalalaenhund

    Lalalaenhund Power Member

    Joined:
    Sep 14, 2010
    Messages:
    738
    Likes Received:
    418
    Location:
    errywhere
    Great post, thanks SeoCrap :D
     
    • Thanks Thanks x 1
  19. arnosv

    arnosv Regular Member

    Joined:
    Jul 21, 2011
    Messages:
    322
    Likes Received:
    103
    Interesting post Seocrab, thanks for this!
     
  20. IceHD

    IceHD Jr. VIP Jr. VIP Premium Member

    Joined:
    May 8, 2013
    Messages:
    488
    Likes Received:
    199
    Occupation:
    SEO, SEM, HTML5&CSS3 Coder
    Location:
    Not Romania
    I've been readung about this for some time, but never pulled data before.
    Thanks for taking you're time and comple it.

    I'm going to my bat cave and start doing some research.

    Thanks.
     
    • Thanks Thanks x 1