Last week, when the SEO world was distracted by the fact that Google has blocked reporting on keywords from users who are looking for an account, leaked official document of Google “Google's official guidelines for quality raters”. After reading this 125 page document for assessors, we can summarize the following conclusion...
There are no secrets, sorry
If you are looking for «SEO Secrets", you'll be disappointed with this post. It is also an internal document of the company, and Google will not be happy because of this leak. You will find a practical guide to the philosophy of Google's quality. The key point for SEO – is understanding how to think Google. If you're trying to catch up with an algorithm, you will always respond to changes after they occur. I am writing this post, as honestly believe that understanding of the philosophy of the Google’s quality – is fundamentally "white SEO".
Who are the assessors?
Assessors who monitor the quality of Google search – are people who watch over that the algorithm does what it has put in. Data from the assessors are not only serves as a quality control issue of an existing page, but also helps to improve the potential changes in the algorithm. When you know that Google has tested about 13,000 changes of the algorithm over the last year, than you realize how important this work is.
This particular document focuses on monitoring the quality issue page (SERP) of search engine for some queries. Mostly, the assessors look over websites which are displayed in the SERP for the specified requirements and evaluate each result based on relevancy. Assessors also mark the sites which they consider are spammed. And again, the philosophy of Google is not always reflected in search algorithms. Algorithm – is an attempt to encode the quality of code and rules, so this attempt will always be imperfect. For example, the document is not tells about the number of backlinks, unique domains, C-block links, etc. These are all the quantitative indicators.
Here are the 16 guesses about human understanding of Google's quality
1. Relevance – is continuous
I think the main revelation of this document in the broad sense of what Google company has a rather complex and subtle opinion. Assessors are instructed to assess the relevance of five categories of documents (pages): “Vital”, “Useful”, “Relevant” “Relevant”, “Slightly Relevant” “Off-ropic”. Of course, there is always a certain amount of subjective evaluation, but Google provides many examples and detailed instructions.
2. Relevancy and spam are independent
Relevance is a score, but spam – is a signal. So, according to Google site may be helpful, but spammed. Or the site may not be relevant, but without spam. I think we can see that philosophy in the algorithm. Content is relevant or not relevant, but spam is a tactic or intention.
3. The most interesting rules of intentions.
Some queries are ambiguous - for example, «Apple», can mean many things, no matter the context. Google instructs assessors in most cases, to use the prevailing interpretation. It’s interesting the prevailing interpretation is usually concerns of large brands. For example, the dominant interpretation of the word «apple» will be Apple Computers, and to request «kayak» - is travel site Kayak.com.
Other interpretations (such as «apple» as fruit, «kayak» as mode of transport) will automatically fall in the ranking of the relevance, if there is a dominant interpretation. I think that the notion of the prevailing interpretation makes sense, and it may be necessary for an assessor, but it is also very subjective. In some cases, I disagree with the examples of Google, and I feel that the prevailing interpretation is not fair lowers good sites. Many people want to buy the iPad, when doing a query «apple», but a site that specializes in online sales of organic apples is also highly relevant in ambiguous request, in my opinion.
4. Some results are «vital»
"Vital" rating of relevance – is a special case. Any official entity - a company, actor / actress, politician, etc., may be an important result. In most cases, it is the official home page. Only the dominant interpretation may be important - "Apple Vacations" will never become a vital result for «apple». I think - it is the fuse for testing the algorithm. If the "vital" results do not appear on the subjective needs, many people have questions to the issuance of Google, even if the SEO on such requests are not being used.
Social profiles can also be vital if these profiles are for individual or small groups of people. So, politicians, actresses and rock groups can have a lot of "vital" pages (their home page, a page in Facebook, Twitter profile, for example). It is interesting that Google gives specific instructions that the profiles of social networks for businesses, companies can’t be "vital."
5. General queries are never «vital»
Obviously, Walmart.com is an important result for a query «walmart», but Couches.com is not vital result for a query «couches». The exact occurrence in the domain name does not do automatically something vital, and some requests in its meaning are common/general.
6. 3 kind of queries
The purpose of the request may be classified according to Google – "Do", information "Know" or navigation "Go". This model is Do / Know / Go often found in the manual and is very convenient framework for understanding the search in general. Relevance is determined by the target - if the request is focused on an action in its pure form ("to buy a computer"), then only the result oriented to the action will be highly relevant.
7. Usefulness is higher of (than) relevancy
The broad concept for interpretation, but Google says that "useful" pages (out of 5 rating categories below the relevance "vital") should be more than just relevant - they must also be highly satisfying a request, credible, multimedia and fresh. It is left to the discretion of the assessor and there is not site corresponding to all these criteria. Relevance of the document is not always enough to be in the TOP.
8. Relevance means the language of accordance
If the search query does not match completely with the objective language query, then in most cases the result will be low relevant. Similarly, if the request involves or implies a particular country, but the result does not coincide with that country, then the result won’t be relevant.
9. Local target can be automatic
Even if the request is common, it can have a local plan. Google gives an example of «ice rink» - «ice rink» query should give the result of local search, and not just local results should be marked as "not subject" or "useless." This applies to the region in search. Again, waiting for Google's increasing importance of local search more and more.
10. Specificity of the landing page plays a role
A good landing page will be specific to request. Detailed product page, for example, the best match for the low request for a particular position. On the other hand, if the request is greater, then the greater resource may be more relevant. For example, to the query "chicken recipes" will be more relevant page of all prescriptions, rather than a page with a recipe for chicken.
11. Spelling errors are estimated by intention
If the request is clearly erroneous, the relevance of the results should be based on which respond user likes more. Previously, targeting of failed requests was one of the SEO tactics, but I think we're seeing more and more, Google will automatically give searcher correct result. It looks like Google becomes more aggressive in determining the purpose of the request, and even directing the user to a dominant purpose.
12. Copied content may be relevant
It may come as a surprise, after a panda in the world, but Google has formally recognized that the copied content is not low quality automatically. If the content is well organized, useful, and is not only placed in order to obtain views.
13. Some queries do not need to be defined
Dictionary or encyclopedia of the page is useful only if the request is usually worth a definition or more information. If most people understand the value of the requested words - Google does not consider (for example «bank») dictionary and encyclopedia useful. Of course we are speaking about Wikipedia.
14. Advertising without value or spam
One quote stood out in the document: "If the page exists only to make money, this page is spam." Now, some business owners may object, saying that almost all sites are for making money in some form. When Google says "only to make money," they apparently mean earning money with no valuable content. It's okay to earn money from advertising placed on its site, if you have valuable content. If you created the portal only to make money - you're a spammer.
15. Google.com is very relevant resource
By the standards of Google, the search field is empty without the results shown "not subject" or "useless." The irony, isn’t it? Joking aside, the document does suggest that the internal search results page may be relevant and useful in some cases.
16. Google assessors use Firefox
I said no secrets, but I think it's just a little insider information. Google assessors are instructed to use the Google Firefox, along with the web developer add-on. Do as you know.