BERTopic library in python is a great tool for SEO. It helps to identify the topical coverage of the website.
What is BERTopic?
BERTopic is an open-source Python library, it’s a powerful tool for SEO; utilizing transformer-based language models to perform advanced topic modeling. To put it simply, BERTopic closely mirrors how Google’s Natural Language Processing (NLP) API identifies and categorizes entities within content.
How Advanced SEO Specialists can use BERTopic?
The main purpose behind using BERTopic for SEO is to get insights what topics does a certain website covers in depth.
With that said, I use it for many SEO cases, here are some:
- Topical authority areas (ranking on SERP1)
- Low-hanging fruits (SERP2)
- Topical weakness (Beyond SERP 3 )
- Benchmark dropped traffic especially for content websites or those who rely much on Blog SEO content; as they are more likely to suffer from dramatic drops.
Hint: SERP means search results page. So SERP1 means page one of google search (ranking positions 1 to 10).
The following screenshot demonstrates the first use case of BERTopic in doing SEO for a healthcare provider website in Arabic.
How to start using it for SEO?
First, you need to export all queries from GSC, or at least queries with position 1 up to 20, which means keywords ranking on Page 1 & 2 on google. To get full queries, use Google BQ API instead of the traditional GSC dashboard which has limits.
Second, sort them out in a CSV spreadsheet. Then run the sheet in a python script in google colab that calls BERTopic and make it function across the keywords.
In case, you’re doing that for a competitor website or a website that you don’t have access to its GSC, you can export the keywords from ahrefs (or even semrush) and run the sheet into the same python script. However, note that latter method won’t be as much accurate as GSC exports.
It can be used as well for extensive keyword research driven by semantic search. Furthermore, you can use to get more insights about low CTR issues; a common issue in SEO nowadays.
And perhaps the website is lagging behind due to crawl budget issue and decaying content, two issues that are often overlooked yet significantly impact overall search performance. You can utilize this method along with ScreamingFrog’s new feature for ‘semantic search’.
Interested to learn more about this use-case? Feel free to contact me, and i’ll happily send you the full python script with instructions.