Thread by @GiordMarco96

How can you apply #Python for #SEO to get outstanding results?

I will show you some examples of what you can do without spending money.

This thread is all about Python and quick wins and insights 🧵

Disclaimer: I am not listing source codes here, I will make separate threads later.

The goal of this thread is to give you the footprint and the direction.

https://twitter.com/GiordMarco96/status/1590067832225611776

And remember, Analytics is different.

Scripting alone is not doing analysis and who tells you the opposite... is lying.

I recommend the following framework below:

https://twitter.com/GiordMarco96/status/1625948113541427201

Scraping - It's no secret that Python can be used for this purpose.

There are many types of projects, in this case, the most important revolving around sitemaps.

You can get a lot of information from sitemaps alone because they store time data.

And this means...

...that you can often analyze how much a website is publishing and their structure easily.

Sitemaps can be found by looking at the robots.txt file, though there are exceptions.

Analyze them to get an idea of how much content you could need for a project.

For large publishers or news websites you can scrape RSS feeds.

This is super underrated but you can do crazy stuff if you combine 1 month of data for 6 publishers.

This can be used to get fresh news from foreign sources too...

The only limitation is the RSS format sometimes.

Text analysis - Python is king when it comes to extracting insights from text data.

SEO is largely based on this type of data, lucky us.

There are many things you can do, the only limit is your imagination (and computational power).

Scrape ranking pages, extract their text and get the entities via Google NLP API.

What you learn:

- Most used entities, include them in your copy
- What a machine can recognize
- Terms you didn't even mention

You can use spaCy too but Google API's official.

Scrape the content from a website and identify the topics.

LDA is a technique that allows you to classify these pages and find what they are about.

BERTopic is another candidate for this task, so you have more than one option.

Understanding the main topics of a website (topic modeling) without visiting it allows you to work on other tasks.

This one is a computationally expensive task, not suggested to beginners.

Still, it's very powerful with the right mindset.

Text generation - The most (in)famous topic of the moment.

You can do a lot of crazy stuff in this case too and I am studying it.

While generating dozens of articles is not a gamechanger, it is if you do it 1 million times.

Disclaimer: I am against this strategy.

You can use the openai or transformers libraries for this task.

In some cases, it's better to use models suitable for that niche and test.

More on this in the future.

I want to thank my friend @theDrewDag, drop him a follow!

Text generation is super effective for meta descriptions, alt texts and ads.

While this is not my favorite use case, it's certainly effective.

Long-form content is almost impossible to publish without any editing.

Clustering - Yes, you can do that too.

You can cluster keywords, pages, pretty much whatever you want.

The process can be so demanding that you may want to use paid tools instead.

Still, you can use DIY scripts for a lot of projects.

The official™️ god of Clustering is @LeeFootSEO, whose work you can find below:

patreon.com/leefootseo

Internal linking - Yes, you can use coding to check internal links on your website and plot them.

This is quite hard for a beginner and that's why I don't recommend it unless you have some practice.

It gets more and more complex for larger websites though.

SERP Analysis - You can check what's happening and the history for rankings.

Or you can scrape content once again and repeat some of the steps mentioned before.

You can get SERP data via APIs or by doing it manually, but the former approach is more consistent.

Quality reports - Check broken links, canonicals etc.

The most boring thing you can do, useful if you cannot use tools and you need some fast and free alternative.

Do it once, repeat it forever.

Data Analysis - While Python is not my first choice for analysis, it's still super powerful.

Sometimes Excel is just not enough and you may want to use Python.

SEO tools won't give you a lot of info that you need or you have to pay crazy prices.

Checking Google Analytics, Search Console and even other data sources enrich your reports and your understanding.

We know that many SEOs use data dumps or the usual templates from tools.

Having your own style adds value and is good for clients.

https://twitter.com/GiordMarco96/status/1588620643737534464

You don't want to delegate this stuff to other people unless you're in an agency.

You should handle analysis because you have SEO knowledge.

This is not about coding, more about data understanding.

Web apps - Want to have that custom-ish look?

Build a web app with some Python libraries, it's easy and it can be used by your client too.

Still better than a PPT presentation.

You can build your own dashboards and scripts too...

Automation - repeat what was said before, don't waste time on tasks.

You can automate a lot of things as long as you get value.

Python is the king of automation.

I will list some examples of what you can automate...

- 404 reports, quality reports
- RSS feed/sitemap scraping
- SERP data extraction
- GSC data extraction
- Scraping prices from a set of websites
- Any type of scraping

As a consequence, if you merge some of these processes you can get some nasty stuff.

I don't advocate black hat methods but Python can definitely do that as well.

Coding opens up a lot of doors.

There are many other variations and use cases so you can just test them out.

Find what is more suitable for your style and your needs.

Data Analysis and text stuff are my priorities.

Mainstream SEO tools will never give you a competitive advantage.

Everyone is using them and your client will have probably seen those tools for quite some time.

Having something custom forces you to use your brain and propose something new.

The aforementioned reason is already enough to justify learning something new.

You want to be unique and differentiable.

If you present data dumps you cannot expect special treatment lol

Building your own stuff is still a strange topic in the SEO world.

A lot of people want to be spoon-fed and standardize everything (in a negative way).

I get extremely pissed off when people ask me stuff w/o giving me context.

So yeah, find what works for you and test.

Don't pay $1000 for a template or a course lol

Be sure that you have the correct mindset before you do something.

Building your tools helps you to understand what you really really need.

I am writing an ebook about Analytics for SEO, available starting from March 8 (circa).

This is NOT scripting, it's all about insights, practice and management.

https://twitter.com/GiordMarco96/status/1625425763280498689

Follow me for threads, tips, and case studies (coming soon) about SEO, content, and Python/data.

If you liked this thread, consider liking and retweeting it!🧵

Do you want to know more?

And maybe you have a lot of data and you don't know how to use it.

You can book a call with me (if you have a B2C content website/publisher):

bookk.me/marcogiordano