Thread by @GiordMarco96

#SEO has gone through a lot of changes and a lot of people have to catch up.

In this thread, I want to analyze some SEO trends for 2022. Some of these topics are actually old but recently exploded in popularity 🧵

Knowledge Graphs. I often mention them, one of the key ideas people often forget about. I don't understand why this isn't more popular!

It's one of the most important concepts for Semantic SEO.

inlinks.net/p/knowledge-gr…

Without a proper understanding of how a semantic search engine works, you will face a lot of troubles in the years to come.

The future of search is going towards entities and their relationships. It's a better way to provide search results.

oncrawl.com/oncrawl-infogr…

Entities. The focus on things and not strings (keywords) is more true than ever now. Still, a lot of people prefer to disregard this important aspect.

Including proper entities is the key to improving your topical authority.

Entities are NOT keywords. Don't confuse between the two. I will make another thread on the topic soon.

You should use both because they serve different purposes. Think of entities as the core topics of your entire website for now.

rankranger.com/blog/keywords-…

Topical maps. While Semantic SEO is not suitable for small websites and can require plenty of resources, it's worth pointing out that you should get comfortable with drawing maps.

This allows you to get the bigger picture of a topic/set of topics.

rankranger.com/blog/topical-m…

If you are creating a website from scratch, consider starting from the topical map.

In this way, it will be easier to understand which entities you will cover and how the user is supposed to navigate through the website.

#Python for SEO. In the last few years, some SEOs decided to use coding to overcome most of Excel's limitations.

And I have to say it, great idea! In this case, there is a clear influence from Data Science.

Data Science and all the AI hype influenced other subjects, and SEO is no exception.

Programming languages like Python and R bypass the limits of Excel, in exchange for a steeper initial learning curve.

Learning both was one of the best things I could ever do.

While Python is super popular and cool in the SEO community, it doesn't mean you should learn it.

It's a helpful tool if you want to pursue a certain path and if you are ready to invest some time into it.

And I have to mention automation too. Automating boring tasks is what I think is needed to make the most out of computers and robots.

Spend more time on quality tasks and less on boring and stupid activities.

Data analysis itself is another interesting activity you should focus on. I don't refer to the traditional way of dumping data into a spreadsheet and filtering them.

We have so many options nowadays that we can get more insights in a reduced period.

That's why I always say that people with a more statistical or technical background will reap the fruits of their hard work.

Data analysis should combine your domain knowledge (SEO), business acumen and good technical skills.

MUM. The new multi-modal model promises to be 1000 times better than BERT.

It's not live yet but seems like a big deal. You cannot optimize for it, as I told you in my other threads when talking about BERT.

The only action you can take is to improve your strategy.

For those of you that want proof. We will receive the news in the next future.

seroundtable.com/google-mum-ann…

The only way a more accurate model like MUM can benefit you is by creating quality content.

If you are helpful, factual, and can satisfy the needs of your users, then you have nothing to fear.

I started liking Google more after BERT and I hope this trend will continue.

MUM interconnects data sources of different types, text and images for now.

The idea is to increase the likelihood of finding authority pages and understanding what is accurate.

MUM's goal is to increase factual accuracy. No one wants inaccurate information.

In the future, we may have access to a perfect blend of text, image, audio and video data. I guess this will be the true future of search.

MUM is trained across 75 languages and is capable of handling several tasks.

For these reasons, MUM is said to be multimodal and multilingual. It can operate across different formats and supports more than one language.

Now you know what these 2 adjectives mean.

How will this affect global search? Let's say your language doesn't have any good answer to your query but English does.

MUM transfers this knowledge and the search results will show you what you were only available in English before.

If you want to future-proof your website you need to improve the quality of your work. At the same time, be sure to think in terms of entities and structured data.

Think of your website as a big Knowledge Graph.

Sustainable websites. Save the environment and think more about being green.

This is a new topic for me as well, you can learn more here:

jcchouinard.com/sustainable-we…

Some countries care a lot about the environment and others, well... I cannot say the same.

This is not about ranking or money, it's about saving our planet. I will allocate some time to learn more about sustainability since it's a priority.

Passage indexing. If you don't know what I mean by that, study it now! You should already know that Google can read passages of your text to find answers to rare queries.

And guess what, rare queries are the majority.

blog.google/products/searc…

Phygital for Local SEO. This is not my field, really. I will link you to a good article that also covers this topic.

wordlift.io/blog/en/seo-tr…

LaMDA is a Transformer-based model to improve conversation capabilities. It was trained on dialogue and should produce sensible and specific responses.

Easier said than done. This could represent the long waited revenge of Voice Search.

blog.google/technology/ai/…

The introduction of LaMDA (Language Model for Dialogue Applications) could change search behavior and how users interact with Google.

Conversational technologies will probably be more and more important in the next few years.

Let's connect Knowledge Graphs, factual accuracy and MUM. Google is interested in providing better results with accurate information to enhance search.

I introduce you to KELM, one of my favorite models so far.

KELM (Knowledge-Enhanced Language Model Pre-training) attempts to reduce bias and toxic content in search.

There may be connections with MUM. I guess this could be very much likely.

searchenginejournal.com/google-kelm/40…

The idea is to convert what's in Knowledge Graphs into natural language text that will be used to improve existing NLP models.

This is a very smart idea because graphs are considered trusted sources and can be subject to human editing.

No, we don't know anything about the status of KELM.

When discussing algorithms/models please don't make the argument "how can I optimize for it". You can't.

Think about your end-user and quality.

Imho asking some questions can betray you. Why should I base my entire mindset on an algorithm that can change tomorrow?

That's not even supposed to be the goal of Marketing.

I listed some concepts or ideas that are not even recent.

If you want to future-proof yourself and stay ahead of the competition, then you need to spend some quality time studying the "newer" stuff.

And you should start from the models I mentioned and the Knowledge Graph.

SEO is becoming harder and we are seeing a shift towards the importance of topical authority and branding.

Some may say that building a brand is more important than anything else. I agree! Build a brand, not a website.

As search gets more "picky" you should plan carefully and be honest about the quality you're offering. Is your value proposition that strong?

https://twitter.com/GiordMarco96/status/1503823357673832449