Accept and prepare for the fact that schema is becoming less relevant
Anne says: “We’ve reached the point of declining relevance for schema.”
What does this mean for SEOs and what structured data will still be essential in 2024?
“From my vantage point, the state of natural language processing means that structured data is no longer serving a critical role in training search engines what your content is about. Our content no longer needs to be optimized for machine consumption through the addition of specific schema markup. We’ve reached the end of the training period where that was a value-add.
As evidence for this, the Google Merchant Centre feed is slated to become less robust because Google has said that they can find what they need to generate product rich results directly from your page. You no longer have to structure this for them in a way that you have been previously.
I used to be the absolute number one advocate for structured data and schema implementation. Now, in many cases, it’s difficult to maintain, it’s poorly optimized, and it’s brittle. The return on investment is just not there anymore for most applications of schema. Unless you’re in the position to do a case-study-worthy, best practice, advanced implementation of schema, I wouldn’t embark on the project.”
Do certain sites, like recipe websites, still benefit from using schema?
“The argument I would expect against my advice is that there are still some rich results worth fighting for, whether it’s a recipe rich result, a product rich result, review schema, etc.
We only have to look in the recent rearview mirror to the revocation of the how-to and FAQ rich results to say that this is potentially a moving target, and the time is better spent optimizing our content for users.”
Do you still need structured data regarding what your company is, the history behind your company, and the thought leaders within the company?
“There’s a lot of buzz right now about semantic SEO and semantic markup. I am really playing the cynic here but, very often, the way that semantic markup is being prepared is that you’re putting your content into some sort of text analysis software, and then allowing it to tell you what the main entity of the page is. If external NLP (natural language processing) can already tell you what your article is about and understand what the content of the page is, then adding that as structured data doesn’t add any value.
The other way that this type of semantic schema is being generated is by individual editors who author content for a site. In most cases, these teams are not deeply trained in taxonomy for machine learning. They may be applying taxonomy to articles in an inconsistent way that doesn’t necessarily denote the core topic but instead indicates that a particular subject matter is mentioned.
Human-generated taxonomy for semantic structured data is flawed and machine-generated taxonomy is something that can already be interpreted. If you can’t do best-in-class, then I don’t know how much it’s helping.
There is some value to doing explicit entity SEO using structured data, I just haven’t seen it done in a way where it’s worth the effort.”
Are there any specific page or content structures that make it easier for the machines to digest?
“I am a big fan of returning to fundamentals: effective use of well-nested header tags, readable paragraphs that are informed by keyword research, etc. It seems almost as though these fundamentals are being neglected, but that is exactly what works.
Think about structuring a page so that it could be an award-winning high school essay. It should be well-outlined, and the topics and subtopics should be nested in a way that allows the reader to scan it and understand the argument. That’s going to help the machines to interpret the subject matter and it’s going to make it easier for a real user to consume as well.”
What is your opinion on AI-generated content?
“Large enterprises that are trying to tackle content deficits at scale (having unique product descriptions, having unique and differentiated product metadata, etc.) want to try and solve that problem by using some external content generator.
However, we know that it has to be evaluated in some way by a human because inaccuracies could be introduced. I don’t have an objection to using AI for content generation. The hurdle is going to be organising teams of content fact-checkers at scale. Of course, this may stop being necessary as AI improves.
It’s a fascinating thing. I love that there are tactics in our industry that used to be considered table stakes but may be completely obsolete come next year.”
Is content like short-form video and audio podcasts preferred now? Should you be trying to publish multiple forms of content on your page?
“That makes me think back to the Facebook-led original push towards video that ultimately resulted in a lot of teams pivoting to create in a format that they didn’t have the right expertise for. In some cases, teams drew away from other areas where they could have been growing because of the influence of these tech behemoths on our access to an audience.
I consume a lot of short-form video every day, so I know that is certainly an addictive way to get information. However, I don’t think every brand has the right team in place to really compete in that space. Moving into multimedia and new-format content that you are not currently equipped to handle is a risk. You are spreading your resources thinner and thinner instead of applying discipline and focusing against the buffeting headwinds to put long-term emphasis on areas where you know you can differentiate and grow.”
Can short-form video be appropriate for every type of content or is it only suitable for a certain stage of the marketing funnel?
“I’ve been paying a lot of attention to this lately because I am in one of those few phases in life where we do a large burst of new consumption. I just bought a home, so I have been using short-form video to learn how to do some critical home repair projects.
I feel that this kind of content is actually very bottom-of-funnel. If I’m learning how to repair a crack in my driveway, I’m going to need a product to do that.”
Where and how can effective short-form video content reach its intended audience?
“Personally, I am a Reels and YouTube user. I don’t usually like having targeted ads but, because of the change of address form and the new charges on my credit card, the algorithm knows that I am a new homeowner and has been seeking me out. I didn’t have to go out and say, ‘What do I need to know to tackle the likely challenges of being a first-time homeowner?’ I’ve been riding that wave of suggested content and finding that it is both engaging and educational.
Of course, I am always a big fan of word-of-mouth and speaking to local experts never lets me down. I will ask the people who have inspected my home or done small projects, ‘Where do you get your advice? Who do you trust?’ Then, I can use that to explicitly seek out content creators so that I can learn how to safely do projects on my own and what products I’m going to need to accomplish them. It’s a combination of both push marketing and pull marketing to find the kinds of credible sources of short-form video content that are useful to me.
Going back to the previous question, I don’t think there are any areas of the funnel where short-form video content isn’t relevant. There is a lot of transactional intent in that space.”
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?
“Since my overall tip is to stop doing something, I’ll answer it like this: Where do I think elite SEOs should be focusing in 2024?
You should be doing deeply researched competitor benchmarking. You should be strategic visioning and setting up a long-term growth plan that you can stick to, despite the buffeting winds of change. Also, you should be using experimentation to gather data and support your big, bold moves.”
Anne Berlin is Senior Technical SEO Analyst at Lumar (formerly Deepcrawl), and you can find her over at Lumar.io.