Use Data Or Be Used By Data!

The May 6 issue of Seotistics is here for you!

With all the AI buzz around, we don't have to forget about the useful techniques for analyzing websites.

Yes, some evergreen methods you can use now to add value to businesses.

P.S. I've added a referral system (scroll down)! Go invite your friends and get crazy rewards!

​Read this in your browser​

1. Clustering / Fuzzy Matching

These 2 methods are TOTALLY different. But I had to introduce them together.

Semantic clustering involves grouping pages/keywords together based on semantics.

Concepts like king and queen share no words in common but are semantically connected.

The biggest contributor to Semantic Clustering in SEO is Lee Foot, like shown in this SEJ article.

Lee tested with keywords but not with pages (as far as I know). The only difference is that you'd be using slugs rather than keywords.

Fuzzy Matching serves a different purpose and my mentor Andrea d'Agostino provided a good example in this article.

If you already know where some pages should go, Fuzzy Matching is the best approach.

Its downfall is relying on your slugs, so if they are extremely messy or non-expressive, there is no way to tag them.

The reason why you should have a good URL structure from the start is to improve labeling.

(Please, don't change your URL structure unless it's an extreme case!!!)

2. Topic Modeling + N-Grams

If you want to detect the topic of a website without checking all the pages manually... this is for you!

The easier approach is using N-Grams, sequences of n words to detect the most common patterns.

I've talked about in my ebook (and this OnCrawl article) as they are a nice way to quickly evaluate a website.

The more complex method is topic modeling, like the one you can do with BERTopic, a powerful Python library.

You can do a keyword analysis with BERTopic but also select pages...

It's also possible to go beyond SEO and combine BERTopic and LLMs for analyzing help desk inquiries, like Tyler Gargula did:

3. Gephi

Gephi is a tool to work on graphs. These data structures are made up of nodes and edges, which you can see below.

This is invaluable to represent websites instead of using tables. With proper knowledge, you can easily assess the linking structure of a website.

The picture above depicts clusters by color, according to their links.

The plot per se doesn't tell you anything except that you have 4 groups.

You need to dig deeper and investigate within them but as you see, it's quicker this way.

For simple use cases, you can use Screaming Frog, which offers similar features.

The advantages of such solutions are clearer once you are very specific with your goals.

Gephi offers more flexibility and customization than Screaming Frog.

This evergreen article by Patrick Stox is enough to get you started.

4. Knowledge Graphs/RAG/AI

I've already talked about it in this past issue and I will avoid repeating myself once more.

The only thing I want to mention again is that AI can be used for labeling your pages and classifying them.

5. Basic Rules

Yes, the most basic rules and filters are fine for the majority of the cases.

Instead of worrying for complex methods, start from the easiest and cheapest solution.

One of the most common examples is by dividing pages or queries into groups:

• Put this page under the category "Top" if you have the top 1% clicks
• Put this other page under the group "Bad" if it falls below a certain threshold

SERP/Keyword Clustering also falls under this category.

You are creating a set of rules, no Machine Learning or crazy stuff involved.

This is also covered in my BigQuery Handbook (again, my subscribers have it).

6. Content Decay/Evergreen Score

My favorite topic and one of the most useful activities for large content websites.

My mentor Andrea d'Agostino worked with me to create a process to detect pages to be updated.

It requires some more tuning depending on your goal but it's overall the best solution to detect decay at scale.

Our approach is simple:

• Plot a line for each page, based on clicks over time
• Take the slope​
• Based on its value, label it as growth, decay or stagnation

It's not immune from a lot of problems, such as:

• Seasonality
• New pages

But if you are half-good at cleaning data and subsetting, you will detect these patterns and filter them out.

The Evergreen score is something I created too, the lower, the more a page is evergreen!

My subscribers can see how I use it in SQL thanks to my BigQuery handbook.

You can learn more in my article about Content Decay.

---

[Analytics For SEO Course: Preorder]

Limited offer until June 1, 30% off my Analytics for my Analytics For SEO course!

You will:

✅ Use GSC and GA4 Data to their fullest potential

✅ Learn Python/R/SQL for your needs

✅ Get a complete blueprint for auditing websites

✅ Learn how to 10x your productivity

✅ Learn BigQuery to work on large websites

I teach you what's needed to go from 0 to a professional Data Analyst.

Even if you leave SEO, the foundations are the same for other jobs!

​Learn Analytics For SEO Now!​

---

7. CausalImpact/AB Testing

CausalImpact is an R package created by Google. Given some observations over time, you can assess the causal effect of an intervention.

Let's say you change the titles of some pages and you want to know if the traffic increased due to that change.

Well, CausalImpact can help you!

A good example is provided by Giulia Panozzo's talk about this library.

If you want to run multiple tests and can't bother with coding or being super technical, just use a tool like SEOTesting.

The true limitation of SEO testing is that we usually deal with small samples and it's unlikely to come up with meaningful conclusions.

As such, this is recommend to in-house or people dealing with big-enough websites.

8. Entity Extraction

Extracting entities from a text is another way of summarizing a website and great for On-Page SEO.

Despite the fact that your average content website is now dead 😭, this is still important.

Before I'd have told you to test spaCy but now you have access to GLiNER and even AI.

This article on Medium is a great start for using GLiNER:

Avoid the Google NLP API as it's quite incomplete and not even that good!

9. Market Basket Analysis (MBA)

One of the most underrated techniques (for no particular reason).

Market Basket Analysis involves finding which items are usually associated together.

E.g. people often buy these 2 products or visit these pages during the same session

So you can use MBA to create bundles/promotions or even for better internal linking.

This is something any Ecommerce should consider because you'd like to know what to upsell or cross sell.

A great example is provided by this article on MBA for GA4 (transactions).

10. Analyzing Time

I don't have enough space to talk about this point but predictions and seasonality analysis fall here.

Every time you have to analyze data over time, there are many many thing you can do.

Some methods of performing Anomaly Detection also fall inside this group.

Even the DataViz below has some degree of sophistication:

There are many other methods that I will show you in the next Seotistics issues!

The New Referral System

Invite friends with your unique link and get rewarded! Since this system is new, let me know if something doesn't work!

P.S. This is still a WIP as I need to make it nice, go stack your rewards in the meantime 👀

👥 Join Our Community

Our Discord community offers a small place where we can talk business and SEO.

If you hate all the noise of social media, then this place is for you.

Join our Discord Community!

🔎 Analytics For SEO Ebook (v7)

If you want to learn about Analytics for SEO and get the best resource out there, this is for you.

It will teach you or your employees to:

👉 Prepare audits that make sense and are actionable 🔥

👉 Avoid common pitfalls that cost you money 💸

👉 Create analyses that add value and money💯

👉 Move the needle faster with efficient SEO systems ⏳

This comes with periodical updates to keep the content fresh.

Unlock Powerful Insights Now!

Recommended Reads - Peak Content 🗻

Other than what already shared, please read the following resources to get better at your craft:

• ​Simple Introduction to Graph Theory​
• ​Combining GSC & GA4 Data In BigQuery (SQL Code Included)​
• ​The 8-Step Content Auditing Framework For SEO & Marketing​
• ​The Definitive SEO Analytics Stack [+ Workflows]​

📞 One-Hour Call

If you have doubts about SEO or Analytics, you can book a call with me.

Have doubts about your content website or with your data?

Look no further, I can help you:

Book A Call Now!

❗️ Feedback and Recommendations

If you have ideas/recommendations for the next issues of Seotistics, you can simply reply to this email.

Marco Giordano
​SEO Specialist & Data Analyst

Follow me on 🔽🔽🔽:

Bernerstrasse Süd 169, Zurich, Switzerland
​Unsubscribe · Preferences​