If you're providing some expertise or startup costs to purchase expensive investments (buildings, machinery...) or creating some sort of efficiency of scale, you've probably got a viable business.
If your business plan is to skim money off of workers who could do their jobs and get paid without you, your workers are going to eventually figure that out and cut you out of the equation.
Lead generation is a viable business, as demonstrated by Yelp (and previously Yellow Pages), it's just not hot and fast-growing.
There's going to be a huge difference between businesses where customer acquisition is needed regularly (eg, taxis; it makes sense to take a different taxi with a different driver, based on whichever is closest to you) and where customer acquisition is needed infrequently (eg, cleaners; it makes sense to use the same cleaner over and over, on a fixed schedule).
The problem was precisely that Homejoy was a leadgen product but attempted to extract a cut of the action for the rest of time, rather than one-time as most leadgen goes.
saalweachter is right - the big question is what value Homejoy provides. In this case it's clear that Homejoy offers (to the cleaner and the customer) a substantial one time value, and then provides little to no value for subsequent engagements. Because the product doesn't offer any value into the future, both participants were incentivized (and IMO well justified) in cutting them out of the deal.
This doesn't change what you said; they may still have had a viable business if they had been structured for a one-time transaction.