Latest / Semantic SEO Podcast / Algorithmic Authorship in Semantic SEO: Balancing Factual and Opinionated Content with Koray Tuğberk Gübür
Transcript
- James Dooley: Algorithmic authorship. Today I'm joined with Koray, the absolute legend when it comes to semantic SEO. Thank you. So let's jump straight in. What even is algorithmic authorship? Koray: Okay, it's actually a concept that is completing the visual semantics as well. But the main rule of the algorithmic authorship is rather than writing randomly, we write in the way that algorithms actually can understand, process and also relate the text that you are writing better to the query embeddings. Basically, usually when we write the content, there are four main things there. Either you write factual content, opinionated content, structured or unstructured content. The two of them, which is structured and unstructured, these are format related. Factual and the opinionated content, these are about actually formality of the content. Some keywords or the queries require us to use opinionated content as unstructured. Some of them require us to have structured with the facts and sometimes the vice versa. Basically the majority of the work is about finding a balance and combination between these things. Let's say we talk about dating over, uh, dating for over 60 years old seniors. In that case, when you ask a question like, let's say, how to find seniors over 60 years old. In that case you can directly give a textual structured instructional list and then to the bottom part you can also give an opinionated sentence and you can change the background colour as well. In terms of visual semantics, it means they won't be cutting each other the moment that you use a different background with a different parent HTML. Meaning that actually we are using factual structured and also unstructured opinionated content together. Because in Google's AI, basically AI Overview and AI-motivated Google pattern as well, we see that they use different language models for the same content. And if you are able to use different language models in your content too, you increase your language score overall. I can also give one more simple example. There is a research that I suggest to check, "Are Large Language Models Reliable Rankers?" In that research paper, they have a, a simple test. They basically make a user to search for the best coffee machine and they get the results directly and they try to understand that after the first sentence, what attribute you should be focusing on to increase the relevance better. In that case, they find that if they focus on the material and if the material is the steel, the paragraph is ranking better in the LLMs and also the Google. This is actually what we call micro semantics, meaning that actually micro, micro changes in the words we do to increase the relevance. So it's a kind of relevance configuration of the content by using the words and the format of the text. James Dooley: So on algorithmic authorship, right, you there, you mentioned opinionated versus factual kind of content. If someone's looking to do a full semantic content network or they're building out a topical map and then they're doing the content briefs, should they be looking to have all the site being just opinionated or should they have all factual? Or can they have a mix of opinionated and factual on the same website? Koray: That's a great question actually. In, in Thursday during my workshop, I will be having an example of this too. Uh, let's say I can, I can give a site name too, JóSzaki.hu. There are actually, we have four type of different pages for four type of different queries. It's a local, uh, service provider search engine actually, like for plumbers, electricians, etc., in Hungary. And let's say you ask for how do I solve XYZ error on AC. The moment that you ask the question like, "How do I solve?" Google is going to be actually parsing the query as experience related. Meaning actually I'll be using my subdomain as a forum to answer that. And forum indexing is very tolerant for Google. You can write nonsense to a forum, still Google will be indexing it because that's just an opinion. It doesn't have to make sense at all. That's the good side of opinion content. Quality is very related or, uh, it's not basically objective. It's very subjective. And when it comes to the question of, let's say, the AC installation prices, we directly use actually our local directory of the service providers. If you're asking for, let's say, what is the price of AC installation in X country, we have another page type that we merge our directory with actually a paragraph because usually this question gives a featured snippet and we try to get the featured snippet in the above-the-fold area. Then we also provide a directory with also some subsections as well or combining this according to type of the AC or type of the, let's say, the city district, etc. James Dooley: And then if someone was to, let's say, try to write a factual piece of content that for something that you think should be an opinionated, would that just take you straight out of the bucket of being you're not actually meeting the search intent? Koray: That is true. Search intent and also, uh, again, it's also related to Google's fight against a little bit, let's say, scaled content with AI because that's why they are keep promoting the social media platforms or forums as well. Because usually the forums have people's opinions for other people. And if you are talking about, let's say, the allergy symptoms, normally it's a very factual topic that you should be very careful while writing it. But also some of the allergy symptoms, they're not about facts because how you or a certain allergic symptom might be changing from a person to person. Let's say you are a pregnant person with certain age and there is not that much fact that people are looking there. They are looking for actually somebody who is experiencing it. In that case, usually we create actually a kinds of, uh, review section or submit your opinion type of section, and we use that opinionated content together with the factual content. If you go to eharmony again and search for, let's say, the best dating ideas in New York City, there we have written all the dating ideas according to the different types of couples, like Black couple, Latino couple, etc. Then as a user, you can vote the ideas. And if a certain idea gets more vote, we move it higher on the page automatically. We also verbalise the listicle differently as well. Meaning that actually users are able to put their opinion and we change the content based on their, their opinion on the live, uh, directly too. James Dooley: And then with regards that you mentioned, like, different subfolders, like this sits in the blog section, this sits in the forum section, and you mentioned there about kind of like user-generated content or opinionated content. Reddit is working better than ever. Facebook groups is working better than ever. Could you ever... Would you ever on a page that was a factual page of, like, say, how to install an air conditioning unit, like you mentioned, where it's factual, could you in the micro content then have an opinion based on the same page of it being factual at the top? Koray: Yeah. It is actually about what we call Query Deserves a Page because not every keyword or the query deserves to have a separate page. Meaning that actually we wouldn't open four different pages every time in some of the certain type of ACs or certain cities if there is not enough level of search demand or if the entities in the keywords are always the same. We don't open a new page. Instead, we create a complex page that is having all these components together in them. Actually, the senior dating over 60 from eharmony is a good example for this. You can examine it because there at the bottom you will see a Q&A component that a relation expert, relationship expert, is answering questions of people. At the top of the page, we have more factual section. But in the further top, we directly have a visual engaging component to find your match as well. So you can merge it according to the search demand and according to the importance of the keyword. James Dooley: And then with regards to algorithmic authorship, someone like myself, I'm not very advanced when it comes down to semantic SEO. So, is there anything that I should be asking you that the listeners might want to know about algorithmic authorship that they should be implementing that they're probably not doing? Koray: The biggest win will be coming from the beginning of the content, as always. Because in the visual semantics, as we explained, the centerpiece annotation is nearly everything for the initial ranking. In the same way, the first near, like, six, seven sentences or the word groups that we are using on the page will be the most important to exceed the threshold there. And using the right predicate with the right level of definition, again, will be important. And technically, if we are trying to rank for a certain question, like let's say, "What is the milk allergy?" you can directly go to Google, search for "the milk allergy is" in the quotation marks and get all the "is" definitions. That is all the ranking pages. And try to create a created and richer and unique version of all these definitive or definitional sentences, which will be again helpful, be helpful for you actually to increase your relevance while providing actually uniqueness too. This is why you can actually increase your, let's say, the initial rankings further. And you can also use, uh, machine automation for this too because let's say there are 40 types of allergies and you can make this 40 searches on the Google. You can get all the already used word groups and definitions and these attributes, and you can create a better initial sentence group for your centerpiece annotation, also from authorship point of view too, to increase your overall rankings initially. James Dooley: And I bet that nicely presented with visual semantics would be brilliant. Anyone who's not yet seen it, check out the link in the description where I go through visual semantics in depth with Koray. Koray, it's been an absolute pleasure. Thank you. Koray: Here always.