Honestly, Zillow’s announcement that they would lay off 25% of it’s workforce came as a shock to me–in the middle of the biggest housing boom in living memory, what could possibly have gone wrong? But from this shocking event, we are learn 3 lessons from Zillow for scaling with automation.
But the real real behind the story leads to a fascinating warning to me and other companies that look to scale through automation.
iBuying and RoboInvesting- Automation
The backstory here is that Zillow and other real estate companies have been foraying into the practice of robo-buying homes in markets using algorithms, not appraisers, to choose a reasonable price for a home, pitch them an offer, and write them a check for cool, hard cash.
The objective was that Zillow, Redfin, Opendoor and others in the space reasoned that their superior data insights (and elite grad school engineers, no doubt) felt that they could outsmart the market with an algorithm to ensure they could buy a house and flip it, earning a tidy profit on the side but this side of scaling and automation will come with many consequences.
Too Good to Be True? Yeah
According to the New York Times, the attempt flopped — fairly hardcore:
On Tuesday, Zillow, which said it has 8,000 employees, said the division had been the source of huge losses and had made the company’s overall bottom line unpredictable. Zillow Offers lost more than $420 million in the three months ending in September, roughly the same amount that the company had earned in total during the prior 12 months.
# Zillow, facing big losses, quits flipping houses and will lay off a quarter of its staff. Stephen Gandel for The New York Times
Wait — But, Why Though?
This is where the question comes in. What actually went wrong here? How could Zillow possibly have been so wrong that they lost as much money in three months as they earned in the 12 months prior during the biggest housing boom in modern times?
The hosts over at the podcast Robinhood Snacks Daily actually defined a few of the reasons Zillow got zapped so mightily, and here is where the lesson can be for automation as an attempt to scale:
Three Market Forces Algorithms Can’t See:
- Zillow’s algorithm wasn’t able to identify differences between homes that are obvious to anyone who walks through it — Things like a smelly basement, rowdy neighbors, an unpleasant view or that strange extra nook in the kitchen that nobody can quite explain.
- When a seller knows their home would likely go for above market value, they will probably turn down the Zillow offer, opting to go straight to market to get a prettier deal through a human agent.
- Yet, when a seller knows their home would likely go for below average market value (because of said smelly basement and grandma’s kitchen nook thing) they will probably jump at a middle of the road offer like Zillow’s because it’s way less hassle and probably more money than they would have received directly on the market.
Those three factors lead to a cascading series of problems for the algorithm and it’s benefactors, namely:
- Not enough top-notch homes
- Too many homes that they overpaid for
Combine this with a pandemic shortage of the building materials needed to fix up said smelly homes let alone the contractors to do the work, and you suddenly have hundreds of homes in inventory that you can’t move
The Lincoln Journal Star puts it this way:
The result: Zillow faced selling homes at a loss. Analysts who reviewed 650 of Zillow’s more than 3,142 homes found that two-thirds of them were listed for less than Zillow paid to buy them, with an average markdown of 4.5%, according to an Oct. 31 analysis from Ed Yruma, managing director at KeyBanc Capital Markets.
In its third-quarter earnings announced Tuesday, Zillow wrote down about $304 million worth of homes it purchased and expected to sell at a loss. The company said it expects additional losses of $240 million to $265 million in the fourth quarter.
Zillow to close its home-flipping division, lay off 25% of staff by Heidi Groover for the The Seattle Times
What’s the Takeaway?
If you’re a business (like mine) that leans heavily on automation of processes to drive for scale, you have to recognize that volume and quality are not the same thing and that you are always at risk of these and a few other issues when you have robots (or humans) repetitively doing tasks and missing the nuances of each interaction:
- VIPs gonna VIP. The very best people/clients in your market will not respond well to automation for one reason or another. AKA: Nobody likes being treated like a number, and top clients or talent know their worth. 🛩
- DANGER! Less qualified people, candidates or clients will respond hungrily to your requests. They will seem motivated and check off all your boxes on paper. If you’re smart, you’ll weed them out quickly, but if you’re not so smart, you won’t notice until you’re weeks in and, like Zillow, literally bought the farm already. 😞
- Garbage In, Garbage Out. Scale doesn’t fix crap quality, it only scales it. Worse, it multiplies it. If you have 10 people working on the wrong process, you have 10x or even 100x the problems. If you have bots and automations running, they will make this even worse. You have to control your inputs to get solid outputs.
- Supply Chains are Fickle Things. Depending on your business, as fun as it is to be on-demand, if 2021 has taught us anything, its that externalities are real. Developing real engagements, connections and networks is an old-school but verifiable method of softening the impact of these shocks.
As we continue to build out ConnectedWell and our systems, tools and processes, we need to keep these things in mind