Honestly, unless you’re having a rather specific difficulty, IMHO upgrading your card would be a complete waste of money. That doesn’t mean you can’t or shouldn’t do it, because the same could be said for a lot of what we -that is, the gaming community- spend our money on. What you spend your disposable income on is up to you. Just don’t think that upgrade is going to change your world.
You’ll see a lot of online advice about how you “need” to have 8 Gb VRAM now that you can get it or what’s even the point of living. If you’re absolutely obsessed with every frame-per-second you get out of a game, you’re welcome to spend as much as you want on a new card. If you’re willing to turn down the settings a bit and just enjoy playing, you might be surprised with what you can get away with.
Case in point, I just upgraded my whole PC about two weeks ago now. I had been using my previous machine since 2013 and the rebuild cost me about $700. Obviously there were significant savings because I didn’t waste any money on things that were completely reusable: case, PSU, optical drive, OS, storage, monitor, peripherals, etc. Also, just as obviously, I wasn’t going to spend that entire budget on the VGA card - because you sure can if you want to!
So I got an RTX 2060 for about $300. Is it better than the GTX 760 it replaced? Absolutely. Did it change my gaming? No. I could run Shadow of the Tomb Raider / Dishonored / Deus ex / etc. just fine before. They looked fine, they played fine. They may look slightly better and play slightly better than they did before, but like most computer upgrades, you could only tell if you went right from one to the other. After two weeks, “better” is the new “normal” and it’s pretty much just business as usual. I suspect that if I could play everything I wanted on a 760, you can probably do the same and more with a 970.
As a side note, going from an FX-8320 to a Ryzen 5 3600X was a major improvement, but honestly I can only tell when I’m running a simulation or doing some other computationally-intensive Python task. Cutting a 10 ms operation down to 5 ms is beyond my ability to notice, but cutting a 30 minute calculation down to 15 minutes is a big deal.