If you're not seeing page latch waits as your number one wait type, by an enormous margin, then you should not be using the In-Memory tables & indexes. Period. They solve a very narrow range of problems. They're not magic. They come with a lot of limitations and overhead.
The exception to this is the possible use of in-memory table variables. Now those can radically improve performance IF you're using lots of table variables. If not, again, no real win here.
The In-Memory indexes & tables are very much an edge-case tool. Don't get hung on the marketing hype. Now, if you're in the edge case, and you can deal with the limitations, they are incredible performance enhancement tools.
Now, on the other hand, Columnstore indexes are a great solution for a common set of problems. If you are running a data warehouse or similar type of reporting system where there are likely to be lots of aggregate queries, clustered columnstore indexes are going to be your best buddy. If, on your OLTP system, you're getting more than a few aggregation and analysis queries, you may see huge performance gains by adding nonclustered columnstore to your clustered index tables.
Columnstore indexes are not magic. Or, rather, like all magic comes with a price, so do columnstore indexes. They have a different set of maintenance requirements. They add overhead to inserts/updates/deletes, so you need to be cautious about when and where you add them, like regular indexes. They especially suffer in a very high volume of data modification queries. The way they work requires a delta store to be maintained until a threshold is reached upon which a rebuild occurs to move stuff from the delta into the compressed, pivoted data store. If you're doing that a lot, it's a massive performance hit on the system. Columnstore indexes are absolutely horrible at point lookups and limited range scans: WHERE ID = 42;. They can degrade performance there.
We've talked about this several times online now. It feels like you must be getting pressure to use these things without understanding why. The key to implementing any of these technologies is understanding their usage (I sure hope that's been explained, but happy to try again) and their limitations. Then, cautiously, with careful testing and evaluation (measure with Extended Events, check the execution plans), implement them in appropriate places.