A lot depends on player tendencies. Since I don't play on the same site as you, take my comments with that caveat in mind.

The way I approach 3betting spots is the following.

First, we need to develop a baseline range. This can be anything, but Snowie ranges are a good start. You can also look at Peter Clarke's "Modern 3-bet ranges". Keep in mind that most of the ranges are mixed. I usually simplify the frequencies to 100%, 50% and 25%. (This point will become more important below)

Then we want to look at Villain tendencies. The main ones are their VPIP/PFR and fold to 3bet stats. If they fold too much to 3-bet, 3betting polarized can be useful. We also want to look at postflop tendencies. If they tend to call a lot of 3-bet but fold a lot postflop, 3-betting linearly can be very useful.

An easy way to adjust to player would be to have the same ranges in each spot, but simply increase/decrease the frequencies of different hands depending on player type. Say, you want to 3-bet more polarized, then you increase the frequencies of the wheel Aces or something like 87s. If you want to 3-bet more linear, increase the frequencies for broadway-type hands. This way makes theoretical sense because (handwaving a bit) you're basically fiddling with frequencies of choices which have the same EV.

You can think of many other factors, but the above can be a good starting point.

With rake considerations flatting IP is pretty -EV at 25nl, personally I would just play a linear 3B or fold range and maybe play some cold calls with some implied odds hands against weak players who will pay you off when you hit and with passive players behind who won't punish you by squeezing a lot.

Also linear 3B ranges are generally a lot easier to play post flop which I think is an oft understated factor because the % efficiency that you can play your strategy at is just as important as the EV that it actually generates.