I think Kagi more aggressively blocks garbage, which is probably the tradeoff you are seeing.
The last time kagi failed me, I flipped on claude research mode. It found a microfiche archive, OCR’ed it, then started retro-stalking obscure 90’s celebrities (read old interviews, found coworkers, etc).
It took 15 minutes, but it produced an accurate 20-30 page report, then converted it to machine readable form for downstream use. I tried doing the same research myself a few years ago and failed to find anything at all.
As the kids say: Google’s cooked.
As for Google, I typically only use its "Web" search so I skip all the junk Google adds to the `All` tab. With Firefox or Chrome, you can make a custom search engine with `&udm=14` appended, like `https://www.google.com/search?q=%s&udm=14` if you want it to be your default search engine.