For about 10 years now, Google has relied on an army of human evaluators to provide feedback and critical insights on countless experiments run by the search engine. The core of the job is to rate the relevance of search results based on a userâ€™s search query and, to assign a rating, Google raters must strictly follow the instructions provided in a document known as General Guidelines.
Since 2015, Google has released to the public the full version of its search quality raterâ€™s guidelines whenever an update is available. As expected, many marketers and SEO strategists have tried to use that document as a reference for their optimization strategies. However, those guidelines were not created for that purpose — theyâ€™re too broad and not actionable for marketers.
Iâ€™ve been doing search quality rating for Google for over three years. Being ranked in the top 5% of raters for most of this time has given me some confidence that I know what Google is looking for in terms of relevance. So I decided to transform this expertise into actionable pieces of advice to the SEO and content marketing community.
My objective with this series, â€œLearn From a Google Raterâ€�, is to teach you how to approach search through the lens of a seasoned evaluator, so that you can use this knowledge to the benefit of your own content marketing strategy. The concepts and techniques you will learn here are the same that I put together in the SEA Model, a search evaluator course that I created to be used as a supplementary resource to Googleâ€™s guidelines.
Every search occurs because of a need from the user that must be fulfilled. We will call this either user intent or query intent. These terms can be used interchangeably.
Sometimes, itâ€™s very easy to infer what the user intent is, based on the query they used. For example, “I want to order pizza”. Whatâ€™s the user intent? To order pizza, most likely from nearby restaurants.
“Pizza delivery”. Whatâ€™s the user intent? To order pizza to be delivered, most likely from nearby restaurants.
In these examples, the user intent is very clear. Whenever the user intent is very clear, we call this intent the correct intent.
What about this query: “pizza”. What is the user intent?
This query is not as clear as the previous examples. It is possible that the user is looking for pizzerias nearby, pizza recipes, images of pizza, etc.
In these cases, we say that the query intent is somewhat clear. Although itâ€™s hard to tell what exactly the user is looking for, we are still able to think of several different results that could be relevant or useful to them.
When the user intent is just somewhat clear, it means that there are several possible things they could be looking for â€“ several possible intents. Some of them are more likely, some are less likely, and some very unlikely.
Approaching possible intents through these three basic levels of likelihood helps you get an idea as to whether your content stands a chance of ranking well in the SERP for certain queries.
Letâ€™s illustrate with an example.
Assume that you own a travel blog. You like to produce content that is original and of high quality — after all, your blog is known for providing interesting and in-depth information on travel topics.
Recently, you decided to launch a section about dishes that are popular for representing specific regions of the world. Your objective is to provide detailed information about the history of the dish along with the real, â€˜originalâ€™ recipe.
Although you include a recipe in these articles, you are not looking to just rank for recipes. You are targeting curious individuals who like to increase their cultural knowledge by reading your posts, and you want to provide the most interesting information available on the web regarding that specific dish you are writing about (its origins, history, variations, etc.).
You are currently looking for a representative of the Italian cuisine and you realize there are two dishes youâ€™d have a lot of interesting things to write about: pizza and tiramisu (an Italian dessert).
As a matter of habit, you go to Keyword Planner to check the expected search volume in English for the keywords “pizza” and “tiramisu”. 7.5M for pizza and 1.5M for tiramisu.
One common mistake many people make in these cases is assuming the topic pizza would attract more viewers simply because the keyword pizza is more popular. You donâ€™t want to be that person!
First and foremost, instead of just looking for the search volume of the keyword pizza, you should try to find information about more specific, intent-driven keywords, like â€œpizza recipeâ€� (165K monthly searches), â€œpizza dough recipeâ€� (246K monthly searches) â€œhistory of pizzaâ€� (10K monthly searches), etc. These are the keywords that represent the exact type of content you are looking to create.
Then, for the broader keyword pizza, a helpful technique is to convert keyword into query and assume those 7.5M people are issuing the query “pizza” to Google. Based on that information, you need to try to figure out possible types of content they would be looking for with this query, and this is when classifying intents by likelihood comes in handy.
More likely intents for the query pizza:
Less likely intents for the query pizza:
(*) intents addressed by the content we are looking to create
Very unlikely intents for the query pizza:
I like to include â€˜etc.â€™ because there could be countless possible intents associated with a query, and we shouldnâ€™t bother trying to figure out all those possible intents. What is most important here is to list those intents that come up to us naturally and try to classify the intent(s) that are going to be addressed by the content we are looking to create.
Now, letâ€™s do the same thing for the keyword “tiramisu”. We will approach it as if it were a query and classify possible intents.
â€˜More likelyâ€™ intents for the query tiramisu:
(*) intents addressed by the content we are looking to create
(**) intents that would occur on larger cities with a good variety of Italian restaurants
Please note that I classified all the intents as â€˜more likelyâ€™ intents. The reason is that none of these intents seem to really stand out from the others. They are all similarly likely.
Google exists to provide results that are relevant to the user. In other words, the search engineâ€™s purpose is to satisfy the user intent. When the query is just somewhat clear — meaning, it has many possible intents — itâ€™s reasonable to expect that Google will try to prioritize results that address intents which are â€˜more likelyâ€™.
In light of this, a smart strategy is to aim to rank for queries for which your content addresses the correct intent, if there is one, or any of the more likely intents, if there is more than one more likely intent.
If result relevance is intrinsically connected to the likelihood of the intent they are addressing, the relevance of the SERP (Search Engine Results Page) is also affected by the diversity of intents addressed by the results displayed. In other words, a helpful SERP not only prioritizes results that address most likely intents, but it also includes results that would satisfy other intents, as well. However, this depends on the amount of different results for the most likely intents.
For instance, if there are many different results to satisfy â€˜more likelyâ€™ intents, a helpful, diversified SERP would probably only prioritize results addressing â€˜more likelyâ€™ intents. In such cases, results addressing intents that are â€˜less likelyâ€™ would possibly be relegated to much lower positions, maybe on the second page or beyond.
On the other hand, if there are few different results to address more likely intents, a helpful, diversified SERP would also display results that address less likely intents in prominent positions.
From the perspective of a content strategist, I think writing about tiramisu would be a better idea. Even though the keyword pizza is much more popular, most people searching for pizza have other intents in mind and the results that are prioritized by Google will most certainly try to reflect the likelihood of those intents. There are pizzerias everywhere, and people are much more interested in eating pizza than looking for historical or more general information about it.
With tiramisu, itâ€™s a bit different. Since this dish is not as popular as pizza, many people looking for results about the Italian dessert would likely be interested in more detailed information about it, which is exactly the kind of content you are looking to provide. As I mentioned previously, â€œa smart strategy is to aim to rank for queries, and keywords, for which your content addresses the correct intent, if there is one, or any of the more likely intents, if there is more than one most likely intentâ€�.
Tiramisu would allow you to rank well for more specific, intent-driven keywords like â€œwhat is tiramisuâ€� and also for the generic term â€œtiramisuâ€�, something that wouldnâ€™t be nearly as easy with pizza. Furthermore, since your article would also include a recipe, it would be addressing a â€˜more likelyâ€™ â€œlook for recipeâ€� intent with tiramisu, but only a â€˜less likelyâ€™ â€œlook for recipeâ€� intent with pizza.
Source: New feed