You are correct in your assumption of what I call a "match". Technically, it should also meet the 700 SNP threshold, but I don't think the triangulation program requires that. I've never thought of triangulation as a means of determining more matches. It's just a way to narrow the list of things that you can check into further.

A few years back, I was experimenting with an enhancement to the old GEDmatch triangulation software based on the assumption that if A matches B1 and B2 on the same segment, then B1 and B2 also must match on that segment. Turns out that isn't true. Works OK for really large segments, but when you start getting to the smaller ones, above 7cM but not really huge, it falls apart real fast. The only way to be sure is to also compare B1 and B2.

John Olson