What is the “correct” or best way to work with multiple geometries?

Let's say you have several, parametric geometry nodes of furniture.
A chair, a table, couch…

How can I use these in another geometry, e.g. a house?

The only way I found so far is “Object Merge”, but I assume
that is not the correct way to do it, right?

I've been viewing and browsing through tons of videos and tutorials,
but most of those always focus on one geometry.

Any advice or recommendations for best practices welcome.

AS

Normally you keep them in separate objects, unless you need them to interact geometrically in some way.

Or pack and and treat each thing as a single point.

Thanks!

Let me ask it this way, if I have a skyscraper in one
geometry and one chair geometry in another one, which
should be copied/instanced/packed 3000 times,
I would assume that you don't put all these on the top
of your object tree, but into another geometry below “/obj”.
So either directly into the skyscraper or into a “many tables”
geometry node.

Inside there, which operator do I use to reference
any of the “originals” or assets to multiply it?
Or isn't this the way one would work in Houdini?

Sorry, maybe my question is just too basic
Just asking this because there seems to be no other
node capable of organizing things other than a
geometry- or a subnetwork node.
No “folder”, e.g…