Thread by @GiordMarco96

Most useful ideas from #NLP to improve your #SEO workflow or find new ideas.

A short recap on my threads about Natural Language Processing too.

A thread about some concepts that you don't want to miss out on if you love data and making analyses. 🧵

Let's start with the definitions and all the cool stuff:

https://twitter.com/GiordMarco96/status/1499842120068177920

No, you don't need to learn it all or be a master of the subject. Enough is fine to get a high-level understanding of the new stuff.

Just because something worked today it doesn't mean you can get the same results by iterating over and over.

Google is getting smarter.

Text generation: for content automation or brief creation. It usually requires a little bit of editing according to the length, niche and type of information.

It can work for English but I had horrible results in other languages.

It's not always the case in some markets and/or languages. For instance, Italian SERPs can be horrible and have a lot of generated or low-level content.

I noticed the same for other European languages.

Needless to say, they are not always penalized by updates. Eventually, they will be probably hit.

I was thinking of giving an example but there may be people that would pollute the SERPs even more 👀

This is ""acceptable"" for your projects, not for big brands or clients.

Generated content is great if you spend some time editing and mostly for some languages. It can be a powerful weapon if used properly.

It's not the most suitable strategy if you want to build quality content imho. This is not true if you're heavily editing the text, of course.

For some types of content, it can be extremely effective.

You can also try to generate content outlines with AI.

N-gram analysis: analyze word occurrences and spot the most common combinations in an article.

Do you see some frequent combos? Focus on them then.

Google may use phrase-based indexing, i.e. checking if you have some phrases in your copy.

While it's not confirmed, Google released several patents on the topic and I think it makes sense.

Anyway, if you already optimize for entities it should be more than fine. Just be sure not no reduce SEO to a set of boring steps.

Named Entity Recognition (NER): extracting entities from a text.

You should use it to get entities in a given text and understand which ones are more frequent or are in a more prominent position than others.

This is essentially the key to getting entities and understanding what you are supposed to include.

There are some tools for that as well, you can just search for NLP SEO tools. Just beware of the LSI scam.

Part-of-speech tagging (POS tagging): analyze sentences and assign part of speech to each word.

This is useful to analyze sentence structures in competitors or even your own.

It can be effective for getting featured snippets, especially if you don't speak the language.

Text Clustering: my favorite use case. You can cluster keywords to see patterns, groups, and much more.

There are tools as well but for simple tasks, you may want your script.

searchenginejournal.com/semantic-keywo…

There are tools like @keywordinsights that can be used to cluster keywords and do that based on SERP similarity too.

Use it for bigger projects where you need accurate clustering based on SERPs.

Topic modeling: spot topics in documents. In this case, it would be web pages.

This is useful to evaluate the ""topicality"" of your competitors and see what you can extract.

Here is an example:

holisticseo.digital/python-seo/top…

It can be quite expensive task for big websites, I warned you.

Nonetheless, you can use it to evaluate smaller websites and understand what topics you can spot.

I've talked about one #Python library for the task as well.

https://twitter.com/GiordMarco96/status/1492289867908366336

And if you need a short primer on why you should use Python I got you covered.

This is mainly for people with a background in data or those who want to explore this new field.

https://twitter.com/GiordMarco96/status/1488885080596619266

to to to

There are some strange acronyms in this world, I explain the main ones here.

https://twitter.com/GiordMarco96/status/1501996323918123022

Google is getting way pickier and not just in English.

I don't have 25 years of experience but I can tell that the difference between 2017 and 2022 is astonishing.

Now words really matter and there are cases where you can even rank w/o backlinks.

I always suggest having a holistic strategy, if possible. If that's not the case my priority will always be on value proposition and quality content.

Getting backlinks is important after you have some pages that can receive these links.

NLP is not only the future, it's already the present.

History teaches that those who don't adapt cannot survive in the job market.

Being always ready for new topics is important in the ever-changing competitive landscape.

Newest SEO tools already include NLP features and that's great.

My opinion is that you don't necessarily need them for all the projects.

There are cases where your content is so better than competitors' that you just need to hit publish.

I'm talking about different scenarios. Competitive SERPs requires the perfect mix between backlinks and great content.

This takes how much coverage on that topic you have into account.

In the last few years, SEO became more diversified and complex, which is great.

You know, always reading about the same topics gets stale after some time. With the availability of data, there are more interesting options now.

Big companies have access to tons of data and need someone to help them.

Sure, this is the ideal task for Data whatever position but only an SEO can have this much domain knowledge in the field.

Hybrid roles are becoming more and more frequent.

NLP is just one of the many options you have at your disposal.

In-house SEOs or some professionals usually have to do other tasks to complete what they do.

Instead of isolation, I propose integration, i.e. being holistic.

SEO is just one of the many industries hit by innovation.

Just think of Economics, Finance and Statistics.

Compare the uni programs they had 10 years ago to what they have now.

The (in)famous voice search will eventually become reality with better NLP techniques.

I am not so hopeful about it but it has to become effective sooner or later.