Use Data Or Be Used By Data!

​

The January 29 issue of Seotistics is here for you!

Last time we saw a process for Content Optimization that can be split into several subprocesses.

Today we are going to analyze the details of what I outlined last time and how these subprocesses work.

Agencies can implement most of them because there is no reason to do manual work in 2024.

P.S. There will be big announcements in the next 2 weeks!

​Read this in your browser​

Recap

(N.B. Before we start, read the past issue first to understand the context.)

​

We left here! I am going to describe the 3 flows on the right in no particular order.

N.B. Data cleaning is a huge and specific topic that I covered in my ebook.

1. Tagging Pages

Now, this is what most of the time I get asked and what I deem a crucial activity for any website.

It's impossible to analyze the entirety of a website page by page if it's decently sized.

For this reason, we resort to groups to get aggregated metrics.

In most real-life scenarios, you won't have much order, excluding the tags and categories in the CMS.

If you want a different type of classification, things get even messier.

The solution here is breaking down the problem and thinking methodically.

There are several approaches:

• Clustering
• Fuzzy Matching
• Rule-based

Prevention: use Airtable to do it from scratch. New websites with labeled data are a joy to work on.

P.S. If you need tags/categories but can't use APIs, then you need to scrape the whole website!

Clustering: A Common Misconception

Clustering refers to a set of unsupervised Machine Learning (ML) algorithms.

In plain English, you don't tell the machine anything about your data, it has to figure out itself!

Now, the commonly called SERP Clustering isn't actually Clustering.

Other types of clustering can be used for classifying, labeling and tagging pages, which is invaluable for Ecommerce.

I have shared a lot on this topic in my Discord server since this is quite a common question.

Fuzzy Matching

In most cases, you want something quick and effective at grouping.

Fuzzy Matching is based on distance, it tells you if 2 strings are similar or not.

This is one of the best methods if you have good URLs and can come up with some descriptive groups.

I share with you this example that tells you how to classify URLs with this approach.

2. Page Performance

I briefly mentioned last time that you should classify pages by performance as a way to speed up decision-making.

Now, how you do it can vary a lot but I've found the following methods to be quite consistent.

Once you can categorize pages, you can tie a group to actions, like shown below:

​

This means that you should already know what to do with such groups!

There isn't any advanced SEO here, it's the usual same stuff under a different flavor.

I am just showing you the power of Operations.

The script for running Content Decay is contained in my ebook and can be easily tweaked.

For evergreen content, I invite you to check the BigQuery handbook you received via email.

3. Research Flow

Keyword Research is a spicy topic because everyone has a different process.

The one I outline below is a generic one that can fit the needs of many:

There are too many sources you can use for Keyword Research and it's pretty much the usual SEO stuff.

Reddit/Quora/Forums offer great insights and I always recommend checking what they are ranking for.

Or... you can scrape subreddits yourself with Python!

Once you have all the data, you combine them and feed them into a SERP Clusterer.

With that output, you can take the name of each group and feed it into AI... to generate ideas, briefs and then output it to GDocs or your CMS.

This is what automation looks like.

If you want to be even more methodical, you can add Airtable as a middleman to store your data and use Make/Zapier to connect to OpenAI.

The Tools To Use

Check my previous Seotistics issue to get information about the stack I currently use for SEO Analytics.

Even if you are non-technical, you can easily make your way through with ChatGPT and a healthy dose of Make/Zapier to connect to APIs.

Really, now writing Python code is super simple and more advanced use cases are handled by engineers anyway!

Tying Everything To Action

The big problem of data (or Python) when applied to SEO is that most stuff is pointless or is a shiny object.

This is the main reason why I focus on processes and analysis rather than testing the craziest models.

After a LOT of failures and years of experience, I can say:

• Reject fluff
• Simplify as much as you can (just not too much)
• Question the status quo

The reason I write is because the current SEO and even Web Analytics landscape isn't able to meet our needs.

As long as you keep it grounded, your work will produce much more value than "traditional" SEO.

The Pitfalls Of Traditional Processes

The real issue about current processes in SEO is that they are not scalable.

Most of what we read online works for small websites or boutique use cases, it just doesn't scale up.

The most common pain points I see are:

• Agencies/SEOs can't report on meaningful metrics quickly
• High % of inaccuracies across reports (thanks, spreadsheets)
• More time is spent on pointless tasks than on SEO

Actual digitalization goes through progress but many seem caught in the AI trap.

As you have learned today, engineering is where the gold is at.

The Solution

The best solution to all these problems is investing in data, in this case, architecture.

The cheapest and easiest solution is Google Cloud Platform if we talk about SEO and Web data.

This will be the focus of the next issue of Seotistics!

👥 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.

We also plan weekly calls on Sunday at 9pm CET where you can ask me WHATEVER you want.

🔎 Analytics For SEO Ebook (v6)

If you want to expand on the use cases I showed and get the best resource on Analytics for SEO, 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.

v7 will include additional info on DataViz, SQL, Google Cloud and updates to some Use Cases.

I know, it's been delayed but you will get crazy stuff!

📚 Recommended Reads - Peak Content 🗻

Some cool resources to help you with processes and with SEO in general.

Here is what I recommend:

• ​Ultimate Content Repurposer by Caitlin Hathaway
• ​Content tagging with fuzzy logic in Python by Andrea d'Agostino
• ​Semantic Keyword Clustering Tool by Lee Foot
• ​Connect OpenAI (ChatGPT, Whisper, DALL-E) integrations​
• ​5 Data/Strategic Mistakes That Destroy Your SEO Projects​
• ​The 8-Step Content Auditing Framework For SEO & Marketing​

To new subscribers: check the welcome email to get the most important resources.

📞 One-Hour Call

Want some advice on how to use your data and maybe even build an entire automation system?

Want help with your content strategy or even a content audit?

Look no further, I can help you:

​

❗️ 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​