How Google ranks content credibility: 6 possible metrics

Adapted from an article Russ Jones (Russ Jones), researcher to find in System 1 and Moz.

Every idea in SEO can be considered as a set of metrics that we use to improve page ranking. This means that if we talk about such a thing as "the credibility of the content," it is almost certain there are factors that can be analyzed and improved to obtain higher positions in search results.

If Google measures the credibility of content, what indicators it can use for this purpose? According to what criteria SEO-experts should measure EAT its content? This will be discussed in the article.

We will talk about what might be a "content credibility" and how can we measure it.

, Google may use these techniques for measuring credibility, but maybe not. But we hope that we can develop the concept of metrics that can be used to assess the credibility of the content.

The theme of the credibility of content surrounded by controversy. So, the Google quite clearly stated that expertise, credibility and reliability (EAT) are a very important part of the manual for assessors. At the same time, what part of the algorithm helps to identify the type of content, the company did not specify.

We know that the search engine does not use the data from the assessors, for learning algorithm. In this case, Google can use them to cancel the changes in the search algorithm, which do not meet his standards.

How to measure the credibility of the content?

So, how can we measure the credibility of the content? Ultimately, any idea of which we speak as part of search engine optimization should be limited to a particular set of measurements, which in some form is used to rank pages.

Sometimes it's just enough to feel it. However, authoritative content - this is a difficult concept. And assess whether the content is authoritative or not, it is quite difficult.

In fact, for the assessors Guide leads fairly clear examples of what can be regarded as highly authoritative content. For example, in the "News" section is referred to such criteria as the presence of the article's author of the Pulitzer Prize.

It is unlikely that among the authors of a lot of people, so it is unclear how such examples can help us to determine the credibility of the content.

Therefore, below we will look at a number of ideas that can help get a better understanding of this concept.

1. ALBERT

So, then we'll talk about what could be a potential metric authoritative content. And we start with the concept of how ALBERT.

We have already talked about how to use the Google search algorithm BERT system for natural language understanding. With regard ALBERT, it is "lightweight BERT» - a similar model used by Google, which is designed to answer questions.

This may not seem very important. answer technology questions have been used for quite some time. For example, in the work of selected snippets (featured snippets). However ALBERT translated this work to a new level.

Currently ALBERT and variations of this model took the first top-5 positions and 9 of the top 10 positions in the ranking of answers to questions, and all of them work better than people. This is a clear dominance. And this technology is very effective in the field.

How does this relate to the content authoritativeness? If you think about the document, any content that we produce, our goal is - to answer the questions that our users want answers. Therefore, any topic with which we begin - for example, Data Science, - implies the existence of a series of questions, the answers to which users want to receive.

So, people are following questions may be of interest:

  • Who is the expert in the analysis and processing of data (Data Scientist)?
  • How many of these professionals earn?
  • What you need to know to apply for this position?

And this is where systems such as ALBERT, can be extremely useful for the measurement of the content authority. As you can see, one of the measures authoritative content is, how well he responds to questions related to a specific topic.

From this, one can imagine how Google will evaluate all pages that are ranked on request [data science]. Search engine knows the top 10 questions related to this topic, and look what publication is best to respond to them. Hypothetically.

If Google can do it, it would be a great metric to determine whether solid, meaningful, valuable, useful and credible is the content.

Therefore, we believe that the ALBERT algorithm really has great potential. But let's see, what else can be used to determine the credibility of the content.

2. Data Density

Another possible way - an analysis of the information density. When writing the article, the authors - especially if they are not familiar with the subject - often use a lot of extra words, so-called "water."

At the same time we know that the number of words on the page does not affect the content item.

Thus, one way to find out is whether the content really valuable or not, or at least whether it can provide important information - is to use a natural language processing program (NLP) to retrieve information.

  • ReVerb + OpenIE

One of the most popular open source NLP-technology started as a project called ReVerb, who later teamed up with Open IE Project.

How it works: you can provide the system with the content fragment, and it will extract from it all the factual allegations.

Thus, the information density will be a set of facts which can be extracted from a document against a number of words.

If we had this metric, we could quite easily filter out content written just for the sake of volume, and to find content that contains important information.

Just imagine an article in Wikipedia - how dense is the information in it as compared to the content that most of us creates.

3. Content Style

content style can also be quite a simple metric. For example, if the article has embedded quotes, as is done on Wikipedia, and relevant links are at the bottom of the page - the same scientific articles, it may also indicate the authority.

The same applies to the use of lists of facts or a table of contents, as does the Wikipedia, etc.

All this is fairly simple techniques, which, if you think about most often used by those sites that are more trustworthy.

If this is true, then they can be a clue for Google, that the content that you produce, is authoritative.

4. The quality of the text

Apart from the above parameters for assessing quite complex, there are simple enough. Such as the quality of the text.

Is there anything in the text spelling and grammatical errors? Have you ever looked at the readability? Have you noticed that your content is written so difficult that no one can understand? Or is it just that it simply does not seem solid and authoritative?

Also, there are such things as the length of sentences and the use of certain words.

Perhaps, instead of analyzing all of the sites that mention Data Science, Google looks only at .edu domains or only on published scientific articles, and then compares those formulations that are used in them.

This would be a fairly simple way for Google to identify the authoritative content and one that is not.

5. Media Styles

Another thing that we notice is the media styles. Only in the case of this metric a bit more difficult to understand how it can use Google. At the same time it is easy to measure and Optimizer.

  • annotated charts

I think that we should move away from the graphics and move more to the use of graphic libraries open source. This will enable Google to provide the facts and figures in the source code.

  • Unique visual content

Proper visualization - this is what we can see exactly. Unique visual content is also referred to in the Guidelines for Google assessors.

  • Availability

Accessibility is also important. This option does not make authoritative content, but indicates the extent to which a person concerned about the information that is contained on the site.

6. Klikbeyt headers, qualification of the author and Google Scholar

Finally, we look at several parameters that need to talk.

  • Klikbeyt headers

The presence of this type of header indicates that the content is not authoritative. This is also stated in the Guidelines for Assessors. It is therefore important to avoid the use of this technique.

  • author Qualifications

Another thing that is mentioned in the Guidelines for Assessors - qualifications of the author.

If the author has some certificates, academic degrees, etc., then it helps Google understand how he holds that topic is talking about.

  • google Scholar

Another aspect that also deserves attention - this is Google Scholar. How much Google earns on this project? Probably not very many. Why, then, companies need a huge base of academic information data, if there is no advertising on these pages? Perhaps these materials are needed to assess the content - whether it is consistent with scientific data on the topic.

instead of a conclusion

Finally, we again want to note that everything described in the article - it's just a suggestion. What exactly does Google use to assess the credibility, is still unknown.

But this does not mean that we should ignore it. We hope that the ideas we have described will help you understand how you can improve the content on their sites.

Source: Moz.com