Jason Barnard

Jason Barnard

Founder and CEO Kalicube підприємець і письменник.

Kalicube – компанія програмного забезпечення, яка допомагає бізнес-лідерам будувати свою стратегію цифрового пошуку на перспективу.

  • Допомагає бізнес-лідерам розуміти пошукові алгоритми на основі штучного інтелекту
  • Допомагає клієнтам контролювати наратив бренду, підвищувати авторитет і домінувати в Google у своїй галузі
  • Автор книги “Основи брендової пошукової видачі для бізнесу” та ведучий подкасту “Брендований пошук (і не тільки)


Виступи спікера

img НаЗахід 21

Google's Three Main Knowledge Algorithms: How They Work and What you Can do to Take Advantage

Відео конференції

In his speech, Jason Barnard talks about the knowledge graph, how to work with it, how to get there, and explains the main knowledge algorithms.

According to Jason, understanding the algorithms will help us work better on Google in terms of knowledge. Overall, the trick is to understand three types of algorithms:

  • Knowledge extraction algorithm
  • Knowledge graph/vault algorithm
  • Knowledge panel algorithm

By breaking it all down, everything becomes really simple.

The full knowledge panel is driven by the knowledge algorithm. So how can you make that happen? How can you get Google to represent you the right way, be it you, any other person, the entity, the company, the podcast, the music group, the music album, even the product? You get to educate Google by working the right way with its knowledge algorithm.

Let’s start off by imagining that Google is a child that wants to understand the world through the web but is perplexed because the internet is absolute chaos. We think, as human beings, that we organize things pretty well on our websites.

That might be true. Individual websites are particularly well organized. But when you take all the different websites built by different people, with their unique perspectives, ideas, and tech stacks, things get baffling instantly.

We also tend to think that the web is well organized because we look at it through Google and Bing search, and their role is to organize the web for us, so we get a false idea of how neat the internet is.

So this child is trying to use the internet to understand the world. It’s thirsty for knowledge, and the trick to mastering the knowledge algorithms is really simple: educate Google. This is the term Jason uses all the time, and it gives you control and power, and it makes Google seem less scary.

So you teach it through its knowledge algorithms. Jason’s got a whole talk about teaching Google the specifics to build a knowledge panel, and you can find those online as well as a lot of resources on kalicube.com

None of the algorithms function in isolation. Every time you do anything, it affects everything else. Therefore, it is essential to think strategically. Remember that this is not magic! Even though no one knows exactly what causes different things to happen, it is important to use data analysis and past experience.

Knowledge Extraction Algorithm

Google accumulates knowledge, and you need to be patient and understand that this child learns slowly. We can help the algorithm by using semantic HTML5, by using schema markup on the main content page to show it that this is the main content, because we obviously don’t want to waste resources.

If you want to be considered by any of the algorithms further downstream, you have to make sure that Googlebot or Bingbot can annotate your content accurately, and you need to be sure that it can do so confidently.

So if you’ve got a confidence score of 99 because you’ve correctly used your H1, put it in semantic HTML5 within a section, and correctly marked your resides, your footer, your header, and your menus, then you’ve got a much better chance of getting that higher confidence score. If you haven’t done that, your comfort score will be lower, and you’re likely to be further back in the queue.

Schema Markup

This is the machine’s native language. So you need to provide schema markup in HTML5 that explains explicitly the role and content of each element on the page.

Then you’ll be winning all the time because your confidence score will be higher than others, and that’s where we’ll start feeding the knowledge graph and other knowledge algorithms. Because by being explicit, incredibly clean, consistent, and clear, we’re going to be providing Google and Bing with information in their native languages that will help them understand who we are, what we do, and which audience we serve.

Knowledge Vault Algorithm

This is a machine-readable encyclopedia. The knowledge vault is what most people call the knowledge graph, but that is a mistake. The knowledge vault algorithm, the knowledge extraction algorithm, and your entity at home have to be incredibly clear and consistent, incredibly well written, and have this HTML5 and the schema markup in place.

And for the knowledge vault, if you want to get in there, the entity home is the hub of that information. You need to tell the machine the facts about the entity home.

It’s the place where the entity lives; if you like, it’s your canonical URL for the entity. You repeat that information on every reliable, authoritative, trustworthy source that Google is looking at, and you link from your entity home to that source and from that source back to your entity home as much as possible.

That way, Google goes round and round in circles, seeing the same information confirmed both on your entity’s home and every source around the web that talks about that entity. And that repetition for Google is always going to nail its understanding of who you are.

Knowledge Panel Algorithm

We should look at the knowledge panel as a SERP feature. Just like any other algorithm, it’s more complicated, but it’s still just a SERP feature, and if you look at it this way, it’s simply lots of mini featured snippets.

If you look at the many featured snippets of information that Google believes is true, it doesn’t necessarily have to be in the knowledge graph to be in the knowledge panel. It can add information to the knowledge panel simply by being incredibly confident in the way that it would for a feature snippet.

Google struggles a great deal with ambiguity, and that is going to be a huge struggle for anybody unless you have a completely unique name. If it’s not unique, it’s a huge struggle to get Google to understand the named entity, the specific entity, and present the information that relates to that specific entity.

On kalicube.com, you can find a free checklist to help manage your knowledge panel and Google’s understanding of your entity so it can present what you need to your audience as accurately, positively, and helpfully as possible.