Equal allocation based on multiple variables

This kind of problem feels to me like one of those problems that reduces to a knapsack/subset sum problem. So, not knowing where to even start on this, I would probably start there:

Subset sum problems are a somewhat common question.
One thread with a couple of additional links: https://www.excelforum.com/excel-for...en-amount.html
Another one: https://www.excelforum.com/excel-for...l-to-cell.html
or put "subset sum problem excel" into your favorite internet search engine.

Once you have a handle on how to do one subset sum, then it can be expanded and applied to your problem. If I understand your problem it is to find 3 subset sums where the variation between the 3 subset sums is a minimum (so that the subset sums are all about 1/3 of the total sum). It sounds like there could be other criteria thrown into the mix.

One warning -- the built in Solver that comes with Excel has certain limitations (100 or 200 decision variables and so on). If something built on Excel's Solver will work for you except for these limitations, you may need to explore other Solver engines. Frontline (the programmers of the built in Solver) offer (for a fee) a more robust and expansive Solver engine. I have seen an OpenSolver that gets good review from a couple of users here. I also expect that this sort of task is common in some corners of the real estate/financial/investing markets and you may find stand alone apps that perform this kind of "portfolio allocation" problem (either inside or outside of Excel).

Grouping in two dimensions is a complex problem, and more dimensions is harder yet, but you're only dividing into three groups.

The first thing you need is an objective function that expresses the similarity of two groups as a single number between 0 and 1. I think no one here could help you with that. Someone where you work would intuitively recognize a good partition, but it needs to be reduced to a formula because computers suck at intuition.

Then one way to proceed is called annealing: you initially assign properties to groups at random, and then start swapping between groups, seeking to minimize the variance of the three measurements (A:B, A:C, B:C). You always accept swaps that reduce the variance, and initially, you accept some that make it worse. Over time, you only accepts swaps that make it better.

If you had 15 or fewer properties, you could do an exhaustive test of all groupings.