New Feature: Semantic Content Validation

Posted by admin In: Blog & News No comments

The Most Advanced PBN Content Assistant In The World

Our new semantic validation tool ensures that your team is creating highly relevant content that stays true to both the topic of your original PBN domains and the keywords you are targeting from your links.

Using TF*IDF analysis from popular tools, the system allows you to seed the content with terms/words that are already rewarded in Google’s algorithm, so you can dial in precisely what Google identifies as relevant.

These extra pre-publishing checks ensure that all your PBN links are rewarded with the maximum amount of value and of ranking potential.

This post explores the topic in three sections, providing an overview of the problem, our solution, and how the tool fits into your content workflow within ExecPBN.

If you understand the relevancy issues we’re facing as PBN owners and general TF*IDF analysis (or it just bores you lol), skip to section 3 to see how the tool fits seamlessly into your PBN content production workflow.

1# The Relevancy Issues

Google has worked extremely hard in the last decade to increase its understanding of the context of the content it indexes. Primarily to improve the quality of the search results it provides for the world.

It’s made massive steps in understanding search intent and matching that up with the right contextual answers via which pages rank for an individual query.

They have also become a lot better at understanding topic authority, and which domains it can trust to provide reliable answers within specific fields and industries.

By extension, it also much better at understanding to what degree a link is relevant to the target URL from the source page and domain.

How does that affect serial link manipulators like us?

Well, it means we’ve had to get smarter about where we’re placing our links.

After Google rolled out the Penguin update, we’ve also been unable to signal relevancy, or what we want to rank for, via using exact match anchor text.

So, link relevancy has usurped pure power and narrowed the pool of desirable links.

This focus on relevancy has put massive inflationary pressures on the prices of PBN domains in popular affiliate niches. Health, finance, gambling and even marijuana domains regularly attract multiples of 5-10x that of generic domains that have similarly powerful link profiles.

That means that to stay within a reasonable budget and to protect our ROI, we often to have to entertain buying domains on the fringes of our niche or more general topic domains.

The problem is, that when we come to placing content on these domains, we’re often trying to manufacture relevancy.

We need to produce content that stays relevant to the authority a domain might have on a topic within Google’s algorithm but also stay closely related to the content we’re trying to rank with the links we include.

Flying Blind

We usually take weak measures to ensure that we’re trying to make relevant content. From the feedback I got from most PBN owners, this usually extends to being creative with the page or article title.

Creating these titles involves meshing:

  1. The topic of the blog.
  2. The keyword we’re targeting in the page(s) we link to from the article.

However, the issue comes from the article content itself.

While most of them make very respectable efforts, it unreasonable for our content writers to be experts in every topic we assign them. It’s also unreasonable to expect them to use the natural language Google is programmatically anticipating on those topics.

So, in reality, despite our best efforts, many of us are not doing a great job of manufacturing relevancy for our links.

At best, we’re flying blind and hoping that Google appreciates the full value of our links. At worst, we’re going to have our links heavily discounted or even have our whole domain flagged for being unnatural.

2# Better Relevancy Analysis Via TF*IDF

What if we can look at applying some of the popular and successful methods that we’re all currently using in our on-page SEO arsenal for our affiliate and client sites, to help solve our relevancy issue?

It’s safe to say we’re a decade removed from just keyword stuffing and putting our keywords in our header tags to get Google to find us relevant enough for a keyword.

We understand that we must use the Latent Semantic Index to find language Google finds relevant to the topics we want to rank for.

Many tools have crystallized the science of finding relevant semantic terms by their application of TF*IDF.

What is TF*IDF (a 2-minute butchered explainer)

Wt,d = TFt,d log (N/DFt)

‘Term frequency-inverse document frequency’ is often shortened to the super catchy and easy to remember acronym of TF*IDF! However, don’t be intimidated by the technical name.

It’s a statistical measure we can use to understand how important a word (the term), or a 2- or 3-word combination, is to a document within a collection of other documents.

In the context of a search engine like Google, let’s say our collection of documents is the top 100 results for ‘Blue Apples’.

You could analyze all those documents to understand how important the words/term ‘blue apple’ is compared to all the other terms/words used in those 100 pages.

Let’s do the math quickly using the imaginary page that ranked first in a hypothetical top 100 Google results page:

Term Frequency: TF(t) = (Number of times term t appears in a document) / (Total number of terms in the document).

Let’s say the term ‘blue apple’ appears 20 times in the 1000 words on our page ranking 1st.

(20 / 1000) = 0.02

Inverse Document Frequency: IDF(t) = log_e(Total number of documents / Number of documents with term t in it)

Let’s say the term ‘blue apple’ appears in 50 of the pages in the top 100 pages for ‘Blue Apples’. (Blue apples don’t exist, so there were only 50 pages to index in our imaginary search result lol)

(100 / 50) = 2

Thus, the final TF*IDF weight of ‘Blue Apple’ in the Top 100 results of ‘Blue Apples’ is the product of these calculations:

0.02 * 2 = 0.04

==============

Let’s contemplate this, but with two more terms that appear in our top page, that might be related to blue apples.

Pink Apples

1 / 1000 = 0.001

(1 time in 1000 words)

100 / 12 = 8.33

(12 out of the 100 results)

TF*IDF weight: 0.001* 8.33 = 0.00833

Apple Dye

10 / 1000 = 0.01

(10 times in 1000 words)

100 / 40 = 2.5

(40 out of the 100 results)

TF*IDF weight 0.01 * 2.5 = 0.025

In this example, Apple dye is more of an important term in the documents about Blue apples than Pink apples are.

So, with all things being equal, you would expect a page that contains mentions of ‘Apple Dye’ to rank better than a page that contains the word ‘Pink Apples’ for the search query ‘Blue Apples’.

This example is exceptionally crude because the formula actual down weights super common words like the/is/at etc. However, to understand the basics of creating relevancy with TF*IDF, you now you have a good primer.

 

TF*IDF has been creating a buzz in the SEO world for a few years now.

It’s utilized to help tune the content we want to rank by using language (or terms) that the algorithm is already using to identify relevant content to a search query.

Several tools can help you analyze the semantic index for an individual query. These include:

Text-Tools: https://www.text-tools.net

Cora: http://seotoollab.com/

Semrush – SEO Writing Assistant: https://www.semrush.com/

SEOBility: https://www.seobility.net/en/tf-idf-keyword-tool/

Website Auditor: https://www.link-assistant.com/website-auditor/

All these tools work by taking the top 100 search results for a query and using TF*IDF to work out the most important terms within those pages.

What makes these terms/words so potentially powerful, is that by reengineering the search results, we know that they are already rewarded in Google’s algorithm.

This process creates a workable list of terms and their frequency, so you can tune up your content. By undertaking this tune up, we are hoping Google see our content as more relevant to the query, therefore boosting our rankings.

Why aren’t we using these tools on our PBN?

What if we drop one tier below and start utilizing these tools on our PBN content to solve our relevancy issues?

Could it help make our content and links both super relevant to our target keyword of the post we’re linking to, but also retain natural relevancy to our PBN domains?

Why don’t we create two sets of seed semantic terms for our content writers to utilize in their content production?

Firstly, let’s create a list of semantic terms that would be natural to our PBN domain’s original topic.

We can do this by just thinking of some keywords that the domain is or could naturally rank for and then run them through our TF*IDF tool. (We could even choose 4 or 5 keywords that might be relevant to the blog, to create a single list.)

This process creates a seed list of semantic language Google would expect to encounter within new content published on our PBN domain. If we utilize these semantic terms, our content appears natural and is unlikely to get flagged or discounted otherwise.

Secondly, let’s create another seed list of terms in our TF*IDF tool from the keyword we’re trying to rank for.

Utilizing this language also means Google encounters a lot of semantic terms that are relevant to the content we’re linking to and the query we are trying to rank for.

When we combine both, we’ve been able to scientifically and diligently create content that is considered both natural and relevant to the algorithm. Thus, maximizes the content’s potential value, and therefore ranking ability, of our PBN link.

Theoretically, this all sounds excellent, but how do we start to utilize this technique in our PBN content creation process, without creating more overhead and editorial oversight?

3# How the validation Works within ExecPBN

Luckily, we’ve built the automation of semantically validating your content into ExecPBN, so you can start using this process with minimal effort.

ExecPBN’s Content System

If you’re unfamiliar with ExecPBN’s content order system, our task management and publishing solution for PBN owners, read this before reading on.

One of ExecPBN’s key features for high-level PBN owners is that it allows you to quickly and centrally set up, manage, and publish your content.

Once you’ve imported your blog details into our network manager and then selected which domains you’re linking to are your clients or money sites, you can begin to use our content ordering system.

This system is where you can create blocks of articles that need creating for your client site. You select how many articles you need, and the links and anchors you want to place within each one.

You can manually select where each piece of content is to be published on your network or have ExecPBN auto-assign each piece using our customizable rules. The most straightforward rule is that the system finds a blog where you haven’t linked to the client site before.

There are more advanced rules you can also set, like creating sub-networks, a minimum/maximum Domain Authority rating, or even have them not clash with other clients.

Once you’ve confirmed a few details, and how much automation you’d like to use, new articles are placed within a content queue. You can assign these to individual writers, who have their own account and personal dashboard. Or have them going into a general pool where any of your writers or VAs can pick up and have the article created.

When a writer submits the content, we ensure that they have met your word count and included your chosen anchor within your text. The system inserts the hyperlinks for you when we publish it directly to your blog.

The whole process is entirely customizable, and you can be as manual or as hands-off as you want to be with your content process, including having editorial control over the summited content before it gets published.

As well as validating your content requirements, if you required them, we can now semantically validate your content before submission. If it doesn’t include the minimum threshold of semantic terms, it asks your writer or content submitter to correct it.

Setting Up Semantic Validation in ExecPBN

For the process to work to its full potential, we need to set the semantic terms in two different places. Note that just because you have this setup, you can still elect not to use them in future content orders.

1# Enter Blog Semantic Terms

Firstly, let’s set up the terms for each blog we want to publish our validated content on.

1# Browse to Manager > Network Blogs. Find the blog you wish to add semantic terms to and select [Settings].

2# Scroll down until you find ‘Semantic Topic Terms’. You can either manually enter each one on a new line or paste them in from your favorite TF*IDF tool.

3# Press the [Update Blog] button to save the semantic terms.

Note: The validation process does not automatically become active during every content order involving this blog in the future. The blog semantic term validation won’t be active unless a Semantic Keyword is selected during the order process.

Text-Tools.net Integration

You’ll notice in both semantic submission screens that we have the option to import your semantic terms via a CSV download from Text-Tools.net’s TF*IDF tool.

We use the tool for creating the semantic terms for our own network. However, to grab the terms you need, any TF*IDF tool that gives you a full list by term weight is perfect.

Instructions:

Open: https://www.text-tools.net/semantic-analysis/termweight.php

Semantic Analysis Job > TFIDF

Enter the ‘keyword’ and the ‘main target country’. The process usually takes under a minute to complete and appears under ‘Jobs > Completed’.

Select [Analysis] and then click on the [data] tab on the right-hand navigation.

Right Click & choose save as on the purple [Maximum & average WDF*WDF values]

Head back to ExecPBN & select ‘Choose file’ under ‘Text-Tools.Net Import’ and choose the download CVS file from your hard drive.

Click [Update Blog], and the terms are imported into your setting automatically

2# Creating Semantic Keywords

Secondly, we need to import the semantic terms relevant to the keyword we want to target.

Navigate from the menu [Content] > [Semantic Keywords]

From here, you can see an overview of your existing semantic keyword that you can update or edit.

To create a new one, select [New Keyword].

Enter your keyword name, and again you have the option to enter or paste your terms manually. Alternately, use the CSV file uploader from Text-Tools.net.

3# How to select semantic validation during an order

Once you’ve created your first semantic keywords, you’ll have the option to select them during step 2 of the order process. You can add a single semantic keyword to each article.

Note: This doesn’t affect your link or anchor requirements; you should still set them up separately.

How the validation works

When it comes to picking up an article order, we would recommend that your writers browse through the semantic terms for inspiration at first. They should still attempt to create a natural piece of content, based on your requirements.

They should then import their work into the ExecPBN editor and see how it validates against both your blog and your semantic keyword terms. Then they make any adjustments needed before submitting it.

They don’t have to include all of these unless you want them to. You can set a threshold (semantic threshold in the post settings) of what percentage you’d like included before accepting the content back into the system (for your editorial approval, or for it to be published directly into your blog).

Note that there are two separate lists of semantic terms, one for the blog and one for the target keyword. The threshold target must be met for each list individually.

Writers can see which terms are missing from the content in the semantic terms card above the editor. There are two tabs for each set of terms. Missing terms are in pale blue, while the included terms are marked green.

There is a process bar that lets them know if they’ve met your minimum inclusion limit. Note there is no upper limit, so they may add more than is required.

If they don’t meet your requirements, but submit the content, the validation fails, much like it was missing your link requirements or minimum word count.

Tool Settings

We have two options in the software settings to help refine the tool.

You can find these under [Options] > [Posting]

Semantic Threshold: This sets, by percentage, the threshold of semantic terms you require to be included in your content before the system accepts the submission. 70-80% is a good range, depending on your preference.

Suppression list: As TF*IDF tools have to analyze the content of a whole webpage, you’ll find that your tool often picks up navigation terms like ‘About’, ‘Contact’ and ‘Privacy Policy’. You can add terms you’d like the system to avoid adding to your article order requirements.

Some Weaknesses

The tool isn’t perfect in its full implementation of TF*IDF. For example, it can’t tell your writer the exact frequency a term should be utilized within your article compared to the documents we used to create our seed list.

However, we found that during testing, suggesting language meant that our writers tended to use critical terms several times within an article, once the suggestion had been planted.

Final Words

The semantic validation tool allows you to manufacture perfect relevancy in your content. This process helps provide peace of mind, that you’re doing everything in your power to ensure your PBN links realizing their full value, while further minimizing risk.

It’s just one over 80 tools ExecPBN can provide you to make running a high performing PBN or SEO network seamless.

As always, we’d love to hear your thoughts on this concept and your feedback once you’ve begun using the tool.

TFIDF Resources

http://www.tfidf.com/

https://medium.com/@nick_eubanks/tf-idf-and-how-it-works-with-seo-79b76d9db5c0

TF*IDF for SEO: Test Results and Tool Comparison

Subscribe to our mailing list

* indicates required
0 Likes

Comments: 0

There are not comments on this post yet. Be the first one!

Leave a comment