Analyse csv file?

[mint rapids]

I got the idea that gpt might be good in analying a csv file proposing improvements and suggestions.

At first it seems to work, but it looks like it only checks the first couple rows and gives feedback for it, then it just leaves everything empty until the same row conditions are there and it repeats its feedback.

To make it more clear what I try to do, I have a csv with asset names and a list of tags foir each asset. I'd like gpt to check if the asset names and tags can be improved to make it easier to search.

I made sure to chunk the csv so they are not too big. Each row has a max of 100 characters (many have less) and I make 1000 row chunks (according to gpt it can process 1200 rows with 100 character each)

Am I doing it wrong, or am I not prompting correct?

Here is my prompt if it helps:

The goal is to enhance the searchability of asset data by analyzing the following fields: asset name, pack name, and tag names. Each row in the CSV should be analyzed individually, without skipping or grouping rows. 
Add a new column "tag suggestions" where you only add if you propose any changes to the tags.
- Are there any tags that should be removed?
- Are there any tags that should be added?
- Are there any new tags that might help search results?
- Give specific feedback for each row

csv format example:

Pack,title,tag_titles
artisan,Forge Fireplace Lit,"dwarves, fireplaces, kitchen, stone"
artisan,Forge Fireplace,"dwarves, fireplaces, kitchen, stone"
artisan,Anvil,"metal, workshop tools"
artisan,Anvil,"metal, workshop tools"
artisan,Bellows,"magic, workshop tools"

It only gave results where the tag "workshop tools" appears.

I tried again and removed the first couple lines, then later appearances of "workshop tools" didn't receive that feedback.
Now mannequin and dress where the top entries, so it gave feedback to all entries containing manneqin or dress, but no workshop tools anymore or anything else