Political outcomes is the biggest. Businesses (and individuals) can be strongly (financially) affected by who's elected. I'd been very curious in the past about political outcomes but had never used prediction markets, so I built a little tracker that inputed the probabilities based on scraping ~14 betting sites. I had to do it on a market-by-market basis, and there were lots of edgecases. Having them all in one place (or two; kalshi and polymarket) makes life much simpler.

Prediction markets have also directly affected my travel planning, helping me avoid a country that wasn't at war at the time but had (in my judgement) a too-high risk of war.

You're completely right about thinly traded markets; those are way, way less accurate. But they also offer the greatest opportunity for someone to bet the other way, which has the effect of 'correcting' the market.

Elections usually also have good liquidity.