GoHalves

GoHalves matches non-competing B2B tech brands whose audiences overlap, then drafts the joint campaign for both sides.

Most matching products rank on similarity. Co-marketing cannot. The two companies need the same buyers and different products, and those two requirements pull against each other. A pure similarity score hands you your closest competitor every time.

So every company gets scored twice. One score is who they sell to, and that drives the match. A second, separate score is what they sell, and that one is inverted and used to rule people out. Past a set threshold on product similarity the match is killed, however well the audiences line up.

Which meant the interesting work was all in the exclusion rules rather than the ranking. Not what I expected going in.

Matching opens once the pool is dense enough to produce good pairs rather than merely available ones. That is what I am working on now.

For anyone who has built matching into a marketplace, where did you set your threshold, and did you tune it by hand or from outcomes?