In recent years the SEO has undergone epochal upheavals: among those that have marked more a difference between the old way of making SEO e link building and the new one, we could mention the change in the use of theanchor text and the shift from traditional keyword research and selection approaches (such as the long tail) to the concept of entity and topic.
Whether it's death or rather a change in the conception of keyword research, let's see where it all started - the shift from the concept of keyword to entity and topic. Our journey begins with an algorithmic update back in 2013, Hummingbird.
In September 2013, Google released an algorithmic update that was destined to change the nature of search and the way we conceptualized keywords in SEO.
With the feature called Hummingbird we can say that a shift from the concept of keywords to a "semantic seo", which focuses primarily on the search intent, context and meaning of words, rather than on the words themselves.
Already in the Searchmetric ranking factors of 2014 there was a progressive decrease in the weight of the keyword in relation to url, title and heading tags:
But with Hummingbird, the way we think about keywords changes.
As you can see in the following image, if the typical old style SEO question is: how do I rank for this keyword?, the new SEO focuses on how to give the best answer to the question expressed by the user in his query:
In a way, with Hummingbird, Google acts as a medium, as we will see, between our optimized pages and the needs of the searcher, trying to correctly interpret the user's query and provide results (ours) that answer as much as possible to that question.
But to be a medium, Google needs to understand the real meaning of a keyword, more than its meaning and as we will see in a moment to do so it uses co-occurrences, entities and topics, to which we dedicate the first part of this guide to semantic SEO.
On the other hand, SEOs must create content that is optimized on the right answer to the question expressed by the querywith the help of software and techniques that we will see in the second part of this guide.
Let's start with the first part, trying to understand the concepts of co occurrence, entity and search intent.
Co Occurrences & Entities
After Hummingbird, we could say that the keyword does not play a solo role, but as part of an entity, that is, a set of terms that occur together frequently in a given topic.
In other words, in a given context, there are some topics that co-occur frequently with the keyword we have examined, some co occurrences:
An example of Co occurrences and entities: "Polo"
Let's take an example.
Let's assume an ambiguous term that can lend itself to different linguistic uses such as "Polo".
"Pole" can mean:
- the model of the car;
- the garment;
- the geographical location.
How do we, and how does Google probably figure out which "Pole" we're talking about?
From the context of the sentence: we might say that it approximates the meaning of the word from the associations that appear in the text.
For example, we could have 4 different contexts, for example:
- "Polo is a team sport in which two formations of four players, riding horses and equipped with bamboo sticks, face off with the goal of sending a wooden ball through two poles. The team that scores the most points wins" (from Wikipedia)
- "I've just bought a new Polo - it gets 10 miles to the gallon, standard car radio and alloy wheels."
- "I always go to the office in a polo shirt and jeans: the shirt is too tight on me, let alone a tie!"
- "At the North Pole, the first expedition was in 1908 and legend has it that it is the home of Santa Claus."
Clear, right? From the context, we were able to figure out the meaning of "Polo."
But there's more: in each different context, different terms are needed.
So if, for example, I talk about the Polo Volkwagen, in the pages related to the topic it is easy that terms like Volkswagen, petrol, litre, car radio, rims...very different in the case of the Polo sport, where we will typically talk about players, horses, bamboo, ball and points.
An example of Co occurrences and entities: "Panda"
Here is another example related to the term "Panda", taken from thearticle by Andrea Minini on semantic optimization: depending on the co occurrences in the texts, we could assign different entities to the term Panda, depending on whether it is:
- of the animal;
- of the software;
- of Google's algorithmic update;
- of the Fiat automobile;
To recap: a keyword is always placed within a context and in this it is natural that they recur, or rather co-occorranno of terms, which form a whole, a specific entity.
This concept is also the basis of how Google works: as pointedly noted by Enrico Altavilla in his ebook "SEO Mythology"., the operation of Google is not limited merely to "search" in its index exactly the query typed, but the keywords entered go through a "revision engine", where the query is revised and extended with new or simplified concepts and words.
The query is then reviewed to find the most relevant documents: the advice is then to guess the terms that Google will use to extend it, adding synonyms and related keywords.
This is not enough: the same term, besides belonging to different contexts, can also be understood with different research intentions.
Traditionally, we talk about 3 research intents:
- transactional intent;
- informational intent;
- navigational intent;
Google, per term, must strive to interpret the query's search intent.or the term the user types and provide a consistent result in SERPs, or results pages.
What does search intent have to do with SEO?
We should strive to write content that is optimized not only for the keyword (old school SEO), not only for its set of related concepts (entities) but also based on the search intent of the user searching for a given keyword.
To clarify the concept I take an example taken from the beautiful article by Andrea Minini on SEO entity:
In this case we have the keyword computer.
In an approach of on page optimization old schoolthis could be the modus operandi: insert the word "computer" in the sensitive parts of the html code.
In an entity-conscious approach, rather than just focusing on the keyword on the hotspots, I'm going to create a text that takes into account the user's search intent, which is different depending on whether I post "computers" because I sell them or because I want to talk about them.
In case my query is informative, i.e. I'm looking for a computer because I want to know more about how it's made, where it was born, my entity will be composed of co-occurrences such as: minicomputer, CPU, history, bit and maybe DOS (if you remember it we're old!) and IBM.
If instead the query is transactional, i.e. commercial, I will find an entity with co-occurrences of terms such as features, price, GB, all terms that would not make much sense in a document that talks about the history of the computer.
To recapitulate, Google analyzes the text, trying to infer the semantic context from the presence or absence of certain entities, i.e. groups of keywords and concepts; therefore, if it finds "polo" next to "sport" "ball" and "horse" it will understand that we are talking about the entity Polo Sport, and not the entity Polo Volkswagen.
From this modus operandi consequent 2 important practical applications for organic positioning:
- By the presence or absence of certain co-occurrences, Google will understand if your content is spam or really informative on the topic and this could affect its ranking. From this point of view, searches that correlate the amount of text written with good positioning could simply mean that longer text has a better chance of hitting co-occurrences and thus rendering the entity fully.
- if you want to sell computers, and "history" and "DOS" are not part of the search intent of your interest, you will have to choose different entities. If you don't, maybe you'll position yourself, too bad that those looking for computers won't want to buy it, but learn more about it and that could negatively affect your conversion rate.
5+5 strategies for a "new" SEO optimization
Should the old keyword research and selection be considered outdated?
We could say no: about the old long-tail keyword search, you could say that Google has simply incorporated the searches of long tail in documents that may not contain the keywords in a dry way, but simply satisfy a certain search intent.
And we might speculate, as proposed recently by Rand Fishkin of Moz in an emblematically titled article: Can SEOs Stop Worrying About Keywords and Just Focus on Topics?a joint use of the old school method of keyword research and the new school based on entities and topics.
Let's see how.
Try to understand ontologies (lateral)
What are lateral ontologies?
Ways of meaning something without saying it.
For example, Francesco Margherita gave an effective and at the same time nice example of lateral ontologies by calling himself "bald guy with glasses who does SEO".
Why use side ontologies?
Why People often do research but don't know the exact term to look for what they have in mind.
For example, an entrepreneur looking for a way to be first on Google, could use a phrase like "read something to be first on Google": here Google could return results with guides, books and a seo course, despite the fact that none of the dry queries appeared in the search:
How can we leverage lateral ontologies?
Trying in turn to put ourselves in our user's shoes and understand what linguistic uses this system generatesfor example with tools like Quora or Faqfox that we will see later.
Care prominence and proximity
Remember keyword proximity and keyword prominence?
These old school SEO factors find new life in semantic SEO, as according to certain studies, the placement of keywords in the text, their frequency and distance are factors to be taken into account when optimizing semantically:
As you can see here we have a joint use of the old on page optimization based on title, heading tags...and semantic seo.
Use internal links
In this new approach the internal links, both to pages of our site as to external sites, acquire new relevance, because Linking to related sources can help thematize content around a certain topic:
Identifies a latent concept
The identification of a latent concept is a technique that Francesco Margherita mentioned.
In essence it consists ofidentify, within the topic of topics related to a keyword, a concept, a problem felt around that topic but not explicitly expressed, as in the case of Francis "thrift" relating to getting out of debt, and associate co-citations with it to strengthen its connection in the eyes of Google.
Expand the semantic field
What does semantic field expansion mean? Covering related but not fundamental, relevant but not relevant topics in your article so that Google, in the absence of more in-depth resources on the topic, can also position you for the secondary related topic.
Let's take an example, taken from the article on theexpansion of the semantic field by Andrea Minini:
In this case I, if in the article related to "how to join Facebook", I treat in passing the relevant but not relevant topic "how to unsubscribe from Facebook", I will benefit in the positioning of the latter, if it is highly searched.
The strength of the system lies in the association, within the same content, of topics A + B.
Other 5 suggestions
- write for usersnot for Google: remember we said at the beginning that doing SEO after Hummingbird means answering the user's question? Well, the old saying in SEO applies more than ever: write for the user who searches, instead of Google;
- use different synonymsi: the era of keyword repetition is long gone. To find different synonyms you can use the tools below;
- write a long textThere are several studies that correlate text length with ranking. The reasons may be various, what interests us here is that a long text is more likely to include co-citations of concepts that are part of that entity, thus making the text even more relevant in the eyes of Google;
- become an expert on the subject: obviously length is not enough, you need knowledge of the topic to know what are the primary, secondary and related themes to the topic. On this you can help yourself with the tools we see below;
- you give the best answer to the question: try to figure out what the question implied by the query is and write a text that gives it an answer.
10 tools for creating search-optimized texts
Okay, we've seen what co-occurrences are, entities, and some strategies for semantic optimization.
But what tools can help us?
Let's look at some of them.
- Editorial Assistant: The editorial assistant of SeoZoom allows the extraction of concepts in a SERP related to a keyword, allowing you to choose what to include in the heading tags;
- Discover keywords: allows you to discover keywords related to an entry;
- Text Analysis Page: SeoZoom analyze the distribution of keywords within the page you are analyzing, with relative keyword occurrences;
- Keyword relevance analysis: SeoZoom analyzes the average and suggested keyword density for each domain in the TOP 10 of a query.
Here are 9 other useful tools:
- Related Searches: Google's related searches can give us insight into topics related to a certain keyword;
- Faqfox: is a tool that finds discussions and questions around a certain keyword in forums and the like;
- KeywordTool.ioA tool to find Google instant suggestions in an organized and quick way;
- Ntopic helps you find terms relevant to your topic;
- Dandelion.euan Italian entity extraction tool;
- Count co occurrencesa tool by Andrea Minini for the analysis of the co-occurrences of a keyword;
- Alchemylanguage: a tool that analyzes entities and relationships within an online document;
- Wikistalker an innovative tool that illustrates semantic relationships in Wikipedia articles;
- Quora: The well-known question social network can help you find related and hidden questions about a topic;
I admit: if you made it to the end of this article you are a hero :-). It's been a difficult journey at times, requiring effort in changing the way we've done SEO for years.
Let's not throw the baby out with the bathwater.As the saying goes, even in this case the "old" on-page optimization and keyword research should not be thrown away, but integrated with a new way of thinking about keywords, not as individual entities to be optimized on a page, but as part of sets.
If it can be complex, on the other hand you have the possibility to position yourself with long tail keywords that you would not even have imagined, without even having inserted them.
What do you think? Do you use any of these techniques or tools? Let's talk about them in the comments!