> Don't they have MBAs and financial models to assess this sort of thing?

All models are wrong, some are useful.

Zillows were wrong, and apparently not useful.

Yeah, throw your model at: "contractors took six months longer than expected to finish the kitchen"

Extrapolate that out a hundred thousand times and you get a sense of what they were trying to manage. How did they predict remodel costs? Pre-pandemic material costs and back-of-the-napkin labor estimates?

LOL. It’s kinda like estimating software!

It's worse. These people are relying on unskilled labor hired by contractors to fulfill requirements they dreamed up in a game of telephone.

Mobilizing professional contractors is very difficult unless you are the highest bidder. And even then, shit happens...a lot.

Isn't Zillow more like a post-IPO startup than a bigcorp? If so, it makes complete sense that they might "bite off more than they could chew," (e.g. Pets.com) a lot of startups end that way.

They're not amazon/google big but they're not small. They have tons of employees and billions in cash.

If MBAs and financial models can predict slowdowns, we won't have them at all. If anything ML models will exacerbate the situation IMO and they probably relied on them too much and bought more than they should. Of course what I said is also a projection (without even an ML model), so that is not accurate either :)

The models don't have to predict a slowdown. They just have to account for the total amount of risk Zillow is assuming.

Which should really just boil down to carrying costs vs future expected profits

Their risk model should really account for the possibility that housing prices go down. I mean, there were famously in 2008 banking firms whose models didn't even allow for that.