Zillow wasn't pursuing the strategy of a traditional iBuyer, which as you note is to lowball. Zillow thought they were providing a service of market-making and didn't consider the risks (I still don't see why it wasn't obvious to them...). If they offered some discount from the Zestimate, that would have been a lot smarter, but it would have undermined their goal of making a market, as well as undermining the very concept of the Zestimate (if Zillow says my house is worth X, why are they offering me 80% of X?). The alternative would be to appraise each house in person instead of basing the purchase price on the Zestimate, but that wouldn't scale.