So, when does an eventual consistency model just cause more headaches than it helps for scaling out distributed systems?

Okay, so I'm on a new project, and we've got a bunch of distributed services. We're arguing about whether to go with eventual consistency or strong consistency for some of our data. I get why eventual consistency is good for scaling and availability, especially for stuff like user profiles or activity feeds where it's okay if data takes a second to show up. But honestly, I'm worried about how much more complicated it makes the application code. Like, you have to deal with read-your-writes stuff or figure out what to do when data conflicts. Are there specific times when all that extra work and development complexity just isn't worth it? When does trying to use eventual consistency just cause more bugs and problems instead of actually helping us scale?