Actually their true core activity is that of 'trusted third parties'. Accordingly, the business model they all use is to provide an 'authority', to help building trust in a community of web users (either suppliers and customers), in order to better lock a vertical market to their only profit (or the shareholder's one...).
Best illustration of this paradox?... their 'market approach / strategy': almost all of them strive to have an exhaustive and centralized supply, in order to satisfy as best as possible a centralized demand (difficult to anticipate). In such an organization, mass supply stimulates mass demand via a 'top-down' approach, and the market is mostly regulated according to the interests of the regulator.
'Collaboration' is a bit different thing... This word is intrinsically incompatible with any concept of centralization. 'Collaboration' rather means 'transverse', or 'lateral'... and sometimes 'bottom-up'.
Whatever is the market, 'collaboration' is naturally based on the very initial and basic principle of all trade: the one where demand (barely standardized) drives supply (mostly tailored), and not the contrary.
Therefore, adapted digital architectures (or solutions) for 'collaboration' issues are necessary distributed and peer-to-peer. Such relevant tools should allow the actors (citizens, prosumers, suppliers...) to auto-organize in real-time with each others, autonomously, whatever would be the context.
In such a system, never centralized, trust should also be built (spread) step-by-step, on a close-to-close basis (word of mouth). Each actor would (or not) recommend his partners to his own community... unlike those '1 - 5 scales' of recommendation, often backed by 'crap' comments.
Today, these truly collaborative models are quite rare ... for a simple reason: their economic models are less obvious to investors; or, at least, more difficult to apprehend, according to their usual understanding (disruption). This explains why suitable projects have so much difficulties to start.