Research summary: We show that the existing alliance portfolio of a firm can impede the adoption of a new business practice. We analyse the private equity industry which features alliances in the form of deal syndication and has recently seen the rise of a novel investment practice: add-on deals. Using algorithm supported induction, we first document robust empirical patterns using machine learning techniques, and then test the theory we construct to explain these patterns using standard econometric methods in a hold-out sample. We find that when the capabilities required for the new business practice require new partners, existing alliance portfolio members who support current practices can impede access to these new partners (and hence the adoption of the new business practice) through capacity constraints and inter-partner rivalry.