Today's identity reality has users pay a large up-front cost (in their effort) when establishing a relationship with an SP, the value of that relationship only to be (hopefully) realized in the future. Kinda like front-loading in mutual funds (but without bailouts). Users have to decide too early, and based on insufficient data, whether a relationship will be net positive at some point.
Federated (in the inclusive sense) models can enable a cost curve that more closely matches the expected/hoped for value curve.
The up-front costs are minimized relative to today's model.
The relationship may still go sour (or prove of little value), but the user won't have invested as much sweat equity in it.
Your financial advisor will of course make their commission either way.