At what point do more access points become self-defeating?

I have a legacy Openmesh network that’s become very busy due to increased activity in my area - let’s say resource development that’s brought a lot of workers to a campground (together with the typical yearly trend of increased traffic.)

The devices are working well enough, but the suggested load is around 50 users per AP; over a day I’m seeing around 18 gigs down and 7-8 up and around 150 devices connected to the busiest AP - obviously probably not 150 devices simultaneously.

I had a nearby repeater (mesh) that we’ve disabled for the moment, because I think it was causing more slowdowns than it was helping.

I’ve run fibre to that part of the park, so that busy AP is now a gateway; we’re adding two more APs nearby to cover some new sites, and the repeater will be hardwired back to the spot where the fibre is to add another gateway.

So that spreads the load to more APs, but I’m wondering if adding another AP right next to the busy one would help things, or at what point do I actually have too many? Should I be experimenting with reducing power on some of the APs to limit their reach? (There are lots of RVs in this place, so I’ve always figured we need signal penetrating power.)

We are limited to 120 megabits on the internet service, but could add a second 120 meg connection, in which case, I would split the network in two, and most devices would, I presume, more or less automatically connect to the stronger (ie closer) network, helping spread the load.

What are your thoughts on this? The Openmesh/CloudTrax network has its own channel optimisation system that seems to work pretty well; I’ve tried setting the channels manually, and it doesn’t work nearly as well.