Here is a quick easy 2 step way to know what your competitor is up to on LinkedIn

Step 1: Grab the data for each month by using jina.ai

http://r.jina.ai/https://www.linkedin.com/ad-library/search?accountOwner=NUS+Business+School&countries=SG&dateOption=custom-date-range&startdate=2024-04-01&enddate=2024-06-30

- For notebookLM: this would be adding as a source each link
- Claude could be a paste of each output

Note, sometimes I've found r.jina.ai to be inconsistent - If this fails
- Ideally, retrieve from the API (granted permissions required) - it is a process to get the token
- or Respectfully scrape the data
Reach out if this is of value to you, I'm happy to setup a script that could do this.

Step 2: Start questioning what you want to know!

Qualitative Insights - NotebookLM


One question might be:
How has messaging evolved over time? (e.g., value propositions, CTAs, content types)

nus

So here we see that the National University of Singapore Business School has changed their messaging in 2025 wherein the headlines now call out more specific aspects of the programme:
"Be part of the NUS business school success story"...etc

Also, NotebookLm's assiduous annotation shines here. You could easily click in to find the exact ad URL it is referencing to get to the visuals.

expertise

Above is the typical ad headline used last year

and below is the change in messaging found for the National University of Singapore:

cso

Quantitative Insights - Claude

Claude functions as a good analyst sidekick.

Here I've asked it to cluster to understand better the type of programmes they are advertising.

Again ALWAYS CHECK IT'S CODE AND ASSUMPTIONS

I have found it cheats by just hard-coding random values for the chart.

By reminding it to be factual, and looking through how it does the clustering (mainly keyword matching) - I could then trust the data

Here is an example of what Claude can do:

2

3

Next steps some consideration

One main issue with the above approach is that the visual element which has a huge impact was not extracted for analysis.

However, it is fairly easy and accurate to implement with Vision AI.

A second issue is that these methods rely on jina.ai which may not be so generous with its free tier in the future. (Which could be easily overcome with some polite scraping scripts)

Now Image:

Currently - the process is still quite manual and could be made into an automated workflow quite easily.

Imagine being updated with a message or email when your competitors switch strategies!

Reach out if any of the above has value for you!!