The gradual coevolutionary process can drive the divergence and long-term coexistence of ‘good' and ‘bad' mutualists. How different can they be, how many are expected to evolve, and how fast should they accumulate on the evolutionary timescale? To answer these questions we focus on a two-way, obligate mutualism, which we study theoretically in the adaptive dynamics eco-evolutionary framework. Ecological processes working at the scale of organisms' lifetime (the ‘microscopic' scale) turn out to strongly influence macroevolutionary dynamics and patterns. Specifically, the ecology of partner competition determines the speed of evolutionary diversification, the breadth of evolved polymorphism, and stationary phenotype diversity. These macroevolutionary attributes are also correlated with each other. For example, stationary phenotype diversity increases strongly with the speed of evolutionary diversification, suggesting that highly diverse mutualistic communities need not be old communities; rather, high diversity arises quickly given the appropriate ecological conditions that model analysis identifies. These results provide predictions for mechanistic links between organism ecology and macroevolutionary dynamics and patterns. Empirically documenting variation in influential organismal traits will thus prove critical for a process-based understanding of macroevolution in mutualistic communities.