Consider schema markup as the language you can use to build a knowledge graph

Martha says: “Start to think about schema markup as your language for building a knowledge graph. Don’t just go after rich results. Really focus on being understood by these new AI engines – not just search engines, but AI large language models.”

Does AI understand your content in a different way, and does schema markup need to be different to how it was before?

“You have to think about translating your content for it to be understood. Up until now, when people did schema markup, they focused on trying to get that rich result and hitting the minimum that was required and recommended by Google for the elements of the page that you had to translate.

Now, you need to think about what that page is about and how you ensure that it’s truly understood. How does it connect to other things in the world and on the web? How does it relate to other things across the brand? We’re now aiming for a comprehensive translation instead of the minimum.

This is important because large language models are trying to make inferences. They’re trying to understand the intent of the searcher (or the person they’re interacting with). To make those inferences, they need the context of more than just the basic pieces. They need the context of where this puzzle piece fits in the much bigger picture of all the other things that they’ve indexed or understood.”

How do you fit into this bigger picture?

“Let’s start with links. Links continue to be something that you need to have but, with schema markup, you can actually have links in context. You have a link on a page – so you know where it is, you know how often the link is used, etc. Majestic does a great job of illustrating this in their tools. However, you don’t know what the link is, in the context of the topic of the page.

If I’m talking about a product, is the link taking me to who created the product? Is the link taking me to a related product? Is the link taking me to understand more about the accessories of the product? With schema markup, you can give context to that link. It also brings in the context of all the other content on the page.

If you want the page to be understood in a certain way, use the schema.org vocabulary. It has been around since 2011. Google and Microsoft collaborated to define all the things that they wanted to know about. The Schema.org vocabulary helps you bring context and allows you to disambiguate with external definitions.

If this is a page that’s talking about all the hotels in Paris, you want to be clear that we’re talking about Paris, France not Paris, Texas or Paris, Ontario. However, you can go much deeper than that, especially when you’re bringing out products and trying to figure out and position where they fit.

In your marketing language, you might be talking about footprints, but you don’t mean footprints in the sand, you’re really talking about carbon footprints. Schema markup is a control point for marketers to make sure that the content they are building on the site is truly understood and positioned in the right place, not just for humans but also for machines.”

Do you need to do anything differently to appear in the different AI-based search locations, like SGE, BARD, Bing, and ChatGPT?

“Currently, there’s very little documentation with regards to how SGE works. In fact, I just saw Gary Illyes speak and he said that he’s not allowed to talk about SGE. One thing that’s really great for digital marketers is that Microsoft has been trying to articulate what we need to do. They’ve asked us to make it easier for them to index our pages and help them understand when things are changing.

This plays to the point that we need to help them make sure they understand. If they’re investing money to make sure that they’re crawling the pages so that you have a chance of showing up, you also then need to make sure they truly understand it – and make it easier to understand.

If you look at any research around large language models and knowledge graphs, a knowledge graph actually grounds large language models. It removes hallucinations. It provides a standard, structured way for them to make inferences with less investment and energy. Again, we need to think about how to help them do what they’re trying to do and understand it – through things like structured data or schema markup, and doing it in a really connected way. You’re building a knowledge graph that will help them achieve their goal faster and cheaper. That is some of the key messaging we’re hearing from other thought leaders.”

Are we going to get to a stage where AI understands the meaning behind the content on your web pages and schema might not be required?

“It is a possibility. We work with mid-to-large enterprise, and there is a control point that content creators will want to have. I was speaking to someone today who was saying that it’s kind of scary that we won’t even own the experience. This is not just true with what we’re seeing from Google and Bing, but also, who are the other innovators in that experience that are going to give us the conversion?

I’m a firm believer that, whether we’re talking about search or about marketing overall, the brand is going to want to have some control over what they’re consuming and where they’re going. If the experience doesn’t exist, then they’re going to want a marketing data layer.

The knowledge graph (whether that’s using the standard language of schema markup or something else) is going to be a way that enterprises can have that control. That’s how they will be able to continue to play and tell these AIs how they want to be understood.”

How could this position SEO as a hero in preparing marketing data for broader AI initiatives?

“Search engines are just one consumer, and we’re so focused on that all the time. However, when you look at a knowledge graph and how one can be used, it can be used for lots of other things. What’s exciting about the vocabulary of schema is that it’s very comprehensive.

Tools like Schema App allow you to build a knowledge graph very quickly, which allows the SEO to be the hero and raise their hand within organisations that are trying to figure out how to pull data together from all these different places. You can say that you’ve already done it. You’ve done it for organic search.

We also have indications of the mental model of how this becomes a data pipeline that users consume, that you can build other applications on. When Fabrice Canel spoke at PubCon recently, he talked about how they’re not using multiple data sources. They’re pulling in the indexing and data all in one piece, and then building the services on top. If Bing’s doing that to deliver a different level of value, other people can do that too.

My question is, what are marketers doing to make sure that they have their data in a really consumable way, which allows people to consume and use it? We’re seeing innovators within our enterprise clients who are already thinking this way and using their knowledge graph as a way to inform and accelerate some of the things they are doing, even internally.”

Should SEOs consider turning into analytics professionals?

“I don’t know about analytics professionals, but we have to start asking ourselves how we think beyond the click as our one value piece. We need to be thinking about our business relevancy on a much larger scale. How do we help the business stay relevant to capture conversions, get appointments, and make sales – beyond how analytics are today?

We need to be thinking about the other analytics we might need to be going for. Conversions are key. As we see more and more of that data exploration and research happening off our site (or being consumed and packaged in other places), the ability to measure conversions is going to be really important. If your business is struggling to measure true conversions, that needs to be where SEOs are asking questions.”

Could the internal role of SEOs be turned into an SEO consultant as opposed to an SEO manager?

“The need for the cross-functional nature of the job is accelerating, especially when it comes to content. The other big play here is unique content – that is truly tied to your authority and expertise. I don’t use those words in an EEAT context, but truly what you are an expert in and why users need to go to your site and the content you’re creating as opposed to anybody else.

It has to be unique, it has to be well positioned, and it has to be worth putting you in the initial position for those answers. You have to be the number one result, because the other links aren’t going to exist.”

Do SEOs also need to become more specialised in their expertise or can they still handle multiple roles?

“It depends on your talent. Being able to connect the dots and explain how these things all come together is going to be more important. Everyone needs to become a translator. If you’re a technical SEO, you can’t just say that they need to do IndexNow. You need to explain that they have to do IndexNow because, if you are costing the search engines too much money, you’re not gonna even get the chance to play. You need to tie it back to the business.

Whatever role you’re in – whether it be content, technical SEO, or analytics – you need to be able to think four steps ahead about how the business achieves their primary goal. That’s going to become more and more important. People aren’t going to just accept that you need to build links. They want to know why you build links, how that plays in the space, and how it’s evolving.”

How do you articulate the value of schema to non-technical or even non-marketers?

“Today, a lot of what we’re talking about is how things are changing – and then we’re trying to break down large language models. These days, even my mother knows what a large language model is because of ChatGPT. If you articulate that you want to play in this landscape and be understood there, someone in content can understand that, someone in the business can understand that, a sales leader can understand that, etc.

Schema Markup is a language. I’ve had t-shirts that say, ‘This is the language of search engines’ but I now say, ‘This is the language of machines and AI’. This is the language that can articulate what you are about. Explain that you need to translate things so that you are being truly understood, which is the practice you’re doing around schema markup.

You want to explain that it’s not just working page-by-page, but how you are telling the whole interwoven story – and then talk about a knowledge graph and how all these things are connected.

As humans, we don’t just read one piece of content and move on to the next. We read a piece of content and it makes us think of related concepts and ideas. Even when we talk, we’re trying to find commonalities and things we can connect on – whether that be children, things that we love about our work in SEO, companies we work for, etc. That’s how the neural brain works.

How the machines’ neural brains work is not so different from ours. It’s about making meaningful connections through different pieces – and how you can tie that changing journey to this control point or vocabulary that we can use to explain and speak to machines. We’re all trying to find these connections in order to experience and make decisions. That’s how I would explain it and get people bought in.”

If an SEO is struggling for time, what should they stop doing right now so they can spend more time doing what you suggest in 2024?

“Stop writing really generic content. If you’re just doing the basics, and things that aren’t very differentiated, put that aside and go deep.

Figure out how you can articulate what you have today using schema markup so that you can tell the story. Make sure that unique content is standing out.”

Martha van Berkel is CEO at Schema App, and you can find her over at SchemaApp.com.