Where do I place the mic (for EQ)?

This is the document I show over and over again when people ask me this question.
Optimum Measurement Locations for Loudspeaker System Equalization
@mark-emerson

Great doc - very helpful.

Nathan showed me this this about a year ago and I was always intrigued about how it might work. It's a short article, and I would recommend reading it.
The principles are:

1. Measurement area: source coverage (1⁄4 to 3⁄4 in audience depth, away from walls)
   Don't make measurements at the front of the audience or the back. Keep it to the middle half. Also, at least 3 metres from the back or sidewall.

2. Layout: multi-locations spread in width and depth (no single line equidistant to side
   wall)
   No grids. Don't make measurements in a line or parallel with the back or side wall.

3. Minimum number: 8 locations
   More locations are not bad, but don't significantly affect the result, so there is no extra benefit.

The ideal pattern recommended is a 'Zebra' pattern. I can't find any reference to this pattern, but I have come up with a simplified pattern based on a grid. I call it a chessboard.

I haven't played chess since my youth, and this is not connected to any grand chess strategy. Simply that a chess board consists of 8 rows and columns. Real Chess players might find it an amusing game.

The idea is that no chess piece can occupy the the same row a column. Putting it on a grid shows the principle. I have chosen the King to mark the position of the desk (if possible). It's the center and about half way in the audience. There is also an on-axis speaker position, also about half way.

From to the front to the back wall there are 8 equally spaced distances. Likewise, from the side wall to the centre.

The final piece(!) in the puzzle is what about the distances to the source speaker? It turns out that they are about equally distanced. Of course this depends on the ratio of your room dimensions. It is assumed that the space is twice as wide as it is long.

I thought I had attached some images. New to this.

Thanks great @marcus-painter . Nice way to think about it. The goal is to randomize the comb filter enough to generate an average that best represents the global response of the system so I think whatever method you follow that can generate random locations will work.

Most of the time I will have placed the mains to attempt to match the aspect ratio of the room/audience area. I almost always place so the throw of the speakers cover width wise and add more delays if possible and the line of minimum variance lands on the halfway mark of the depth of room.
Either way, once my place and aim has been measured, I then move to mid depth of the covered area of a particular source and measure 3 different heights of the mic stand for an average.
Since this ONAX mid-depth position is the one most effected by an EQ correction, I dont think it matters anywhere else because you can do anything about it.
If your design is good, the XOVER points will match the ONAX EQ.

@michael-reed If it works for you, I love the efficiency of it. Meanwhile, I'm over here not taking 8 measurements for an audience area or array, but for every zone of the array. Maybe I'm obsessive, but I like to play with audio analyzers. ;)