No-code LDA Topic Modeling: Website Content Clustering Output Organiser (Google Sheets Template)

LDA topic modeling can show you the content clusters that already exist on a website, but the raw outputs are hard to read. This Google Sheets template by Lazarina Stoy is an easy way to organise them: it pairs with jsLDA, a no-code, publicly available web app from Cornell that runs LDA topic modeling in the browser, and turns its downloads into a structure you can work with.

The workflow, as laid out in the sheet:
1. Scrape content-rich pages with Screaming Frog or an equivalent crawler.
2. Clean up the data: replace commas with spaces, remove special characters, save only the content as a CSV, and keep a copy of the order so you can match each row back to its URL afterwards.
3. Paste the URLs and content into the template.
4. Open the jsLDA web app, upload your content CSV and a stoplist for your language, choose the number of topics and iterations (experiment with both; the size of the dataset matters here, and you want to avoid overfitting), and explore the visualisations.
5. Download the outputs and paste them into the template: the topic words into the Keys sheet (the main keywords defining each topic) and the topic-to-topic connections into their own sheet, so you can see how often topics co-occur.

Use this for:
‧ Understanding and processing LDA topic modeling outputs without code
‧ Identifying the content clusters that already exist on a site
‧ Understanding your site’s topics better, as input to internal linking and content planning

This is ideal for SEO consultants trying topic modeling for the first time. The process is shown in Lazarina’s guide to internal link audits and topic clustering and in her YouTube walkthrough; for a code-based version, see the LDA notebook in the Resource Library.