March 05, 2019 - 1:42pm EST by
2019 2020
Price: 20.50 EPS 0 0
Shares Out. (in M): 91 P/E 0 0
Market Cap (in $M): 1,860 P/FCF 0 0
Net Debt (in $M): -315 EBIT 0 0
TEV ($): 1,550 TEV/EBIT 0 0
Borrow Cost: Available 0-15% cost

Sign up for free guest access to view investment idea with a 45 days delay.

  • Nice catalyst
  • I’d rather be long


I am short Redfin. I believe the company’s business model is fundamentally flawed and they are structurally incapable of generating a profit.


Redfin is a residential real estate agency that employs about 1,500 agents in most major MSA’s across the country. There are several key points of differentiation with most real estate agencies.


First, Redfin has a disruptive pricing model where they refund to customers about half of the commissions to customers. Redfin and others have argued that the traditional 5-6% commission rate charged by real estate agents is unnecessarily high in today’s world.


Second, Redfin has a suite of tools available to its agents and customers that are considered industry-leading in terms of usability. Their website is admittedly fantastic. It is simple and elegant and even technology-challenged people can easily slice and dice search results and find lots of useful information such as comp sales, school districts, etc. There are other tools such as scheduling showings which are normally done the old-fashioned way but are now moving online.


Third, Redfin’s agents have a relatively low-touch relationship with their clients, it is not a full-service model. If you see a house you are interested you would just schedule it yourself and you might not even have the agent come with you on the showing. Because of this, the majority of Redfin’s transactions are on the buy side, since selling a property has more manual processes and work involved.


Fourth, Redfin has a unique compensation model for agents which is the opposite of other agencies. While almost every other agent is paid almost entirely by commissions, Redfin agents are paid almost entirely by salary. This allows Redfin to attract agents with the promise of less stress and a guaranteed good living. The company has published plenty of its own compensation data which shows an average agent makes over $100,000 including benefits, while top agents make over $175,000. (There is sometimes a small incentive compensation component, and agents can earn higher salaries with higher sales volumes)


Putting these elements together, Redfin believes there are a lot of people who buy and sell homes who don’t require a full-service approach and would be happy to save thousands of dollars with a shallower relationship with their agent. Redfin is right about this, as evidenced by the 135,000 transactions they’ve completed in the last five years. I personally love their website, and it’s used by millions of people who aren’t Redfin clients. Their business model is actually really clever. So, if Redfin’s general belief is right and elements of the company are great, why is this a short and where do things break down?


We know what Redfin’s proposal to customers is, but what is its proposal to its business owners? The idea is that there is a virtuous circle here, where low commissions drive lots of low-cost traffic to the website. The quality of the website and other technology tools then convert people into customers, while simultaneously offloading tons of tedious manual labor from Redfin agents to their clients. Even though Redfin’s commissions are 50% of industry average, that offloading process allows agents to be greater than 100% more productive in terms of transactions, thus more than offsetting the foregone commission revenue. By paying agents fixed compensation, Redfin will capture the economic gains above a certain point of agent productivity (defined as transactions per year). On paper, this is a great idea. Unfortunately, it hasn’t worked in the real world, and there are no encouraging signs that it ever will.


The main variable in this equation is agent productivity. Most of the other things are either fixed or relatively predictable. Here are the main elements to the model:


  • COGS. This is agent compensation and direct expenses to agent activities. COGS per agent is about $265,000 and has trended up at a mid-single digit rate in the last few years. This amount should stay consistent going forward.

  • Number of agents. It has grown at a 25-40% rate and I expect it will continue to grow at least 20% for the next few years.

  • Revenue per transaction. This has remained consistent at ~$9,400 over the last four years. I believe it will stay at this level for the foreseeable future.

  • Opex. This consists of R&D, S&M, and G&A. These expenses are primarily fixed but should have a variable component to them which allows Redfin to generate economies of scale. (How much, we’re not sure) Encouragingly for Redfin, Opex per agent has decreased over time, falling from $135,000 in 2014 to $118,000 in 2018. I believe they will continue to realize economies of scale over time, although for reasons I’ll explain in a bit, I could be wrong.


The revenue per agent is about $315,000 – this is variable but mostly on volume instead of prices. The COGS per agent is about $265,000 – this is almost completely fixed. The Opex per agent is $118,000 – there are economies of scale here but we’re not sure how much. All-in, Redfin loses about $70,000 per agent per year, or $80 million annually. Why? Because Redfin’s agents aren’t productive enough to offset the reduced commission. That’s it, that’s the short thesis.



Trx / Agent


GM %






















At the current 31 transactions per year run-rate, Redfin has a gross margin of about 23%. As long as they are growing the number of agents and the transactions per agent is flat, the gross margin will deteriorate, which has been the case in recent years. They lose another 25-33% per year to opex. Simply, Redfin agents need to start selling a lot more homes. By my math, transactions per agent per year needs to rise to 36 for the company to break even, and that’s giving generous credit for large economies of scale with opex. At 42 transactions they generate a 10% EBIT margin ($.90 EPS) and at 50 transactions they generate a 20% EBIT margin ($2.25 EPS). But how are they going to get there? In five years, transactions per agent has barely grown at all…how can someone underwrite agent productivity increasing 50-60% in just a few more years? This is the part I don’t understand about the long thesis, there is no evidence to support this one key factor. Anything less than 36 and Redfin loses money in perpetuity.


Thinking about the qualitative issues that drive the transactions numbers, I believe one of the main reasons why agent productivity is both inadequate and stagnant is because Redfin has a self-selection bias with the agents they attract. Most agents are almost 100% compensated on commissions and have no salary at all. This is why there are a lot of part-time agents out there, because they cost the agency nothing even if they only bring in a couple deals per year. But the point is that good agents want to be on the eat-what-you-kill model. I’m not saying that all Redfin agents are bad, and I’m sure some are very good. But I think it stands to reason that compensation structures provide a self-sorting mechanism for the industry. My own conversations with agents (I will be the first to admit these anecdotes are nearly worthless) are nothing but hyper-negative feedback on the experience of being forced to be on the other side of Redfin agents.


Another problem related to this is that Redfin probably encounters elevated levels of defections among its highest ranks. The highest turnover is most likely to be from the lowest tail (ie they got fired) and highest tail (ie they figured out they could probably triple their compensation by moving to a traditional agency). Replacing top agents isn’t easy.


I think the two most interesting questions are what the distribution of agent productivity looks like, and what is turnover like among the top ~20% of agents. Unfortunately, Redfin wouldn’t provide this data to me, however, since real estate agents are a promotional bunch, there is some publicly available information on Redfin’s website. If you search Redfin’s website by a city, it will give you a list of all the agents covering that market. (I am not sure if the search results provide all of the agents in each market, I can only go off the information they provide. There are a number of agents, primarily newer ones, that don't have sales data provided) Each agent has his/her own page with their photo, bio, number of closed transactions (and list of each transaction), when they joined Redfin, and some other random stuff like customer ratings. I went agent-by-agent for six large markets and compiled some stats for buy-side transactions:




Avg Trx

% > 30

Avg > 30











Los Angeles





Washington DC





















The “% > 30” column is the percent of agents that have average transactions per year above 30, while the “Avg > 30” column is the average number of transactions of that group of agents who average more than 30 transactions per year. It’s worth noting that for half of these markets, there is one agent with an average of about 120 transactions per year which are huge outliers. I can’t tell if there is something weird going on with these, maybe these people are somehow getting credit for deals that should belong to other agents, or maybe they really are that productive. For example, Washington DC has one agent who joined Redfin in 2018 and closed 116 deals – excluding this agent the average transaction would fall from 30.9 to 23.8. Superstar agents or not, the data helps validate what seems like a reasonable assumption: there are a minority of top performing agents who are highly profitable for Redfin, but even the best performing markets, on average, are significantly unprofitable due to insufficient agent productivity. It’s a significant risk that Redfin is highly dependent on a small number of agents who are probably incentivized to leave.


As for what evidence exists about turnover at the top, I was able to find a press release from Redfin from 2013 that touted 49 of their agents who had been named as one of REAL Trends’ best real estate agents. Again, I just went agent-by-agent from the list provided by Redfin. Of the 49 agents, here’s what I found:



Current Status


Still active Redfin agents; average transactions 44.2


Moved to other agencies (Keller Williams and Compass most popular)


Moved to other roles at Redfin (usually area managers)


Left the real estate industry


A third of this group of top performing agents ended up leaving for traditional agencies and in total half of this group is no longer an active Redfin agent. I can’t say for sure whether this is representative of the overall agent pool, but I think it’s pointing us in the right direction. Again, common sense says that the best agents will realize they could make a lot more money working on commissions. Someone selling 50 homes per year, with an average transaction size of $500,000, will make something like $350,000 pre-tax, double what they would make at Redfin. I'm sure some people are just more risk averse though, and are content staying at Redfin for the peace of mind with having a good salary.


Redfin isn’t sitting back doing nothing, they are instead trying to become more of a turnkey real estate services firm. They can provide mortgages, title services, and perhaps most importantly they have RedfinNow which will buy a customer’s house for its own account within about a week. Zillow (and private company Opendoor) is doing the same thing to much more fanfare. It’s effectively a business line where Redfin tries to profit from panicky and/or price-insensitive homeowners buy putting in a lowball bid and but dangles the ability to close in a week in the hope the seller bites. It’s basically like a payday lender got into the real estate business. Redfin and Zillow try claim their data allows them to buy and sell homes better, which may in fact be true to some extent though it’s unclear how much. Profiting off peoples’ homes is a very dicey business to be in (eg mortgage crisis, servicing practices) and I believe this business practice will inevitably come under serious regulatory scrutiny for being borderline predatory and having conflicts of interest.


Zillow is forecasting that they will generate $100-115 million in revenue in their home buying business in 1Q19 but that they will lose $33-38 million in the process. Their goal over the next 3-5 years is $20 billion in annual transaction volume while generating a 2-3% margin…at scale, of course. I believe they are just doing this because it’s a way to show growth. I’m skeptical of Zillow’s and Redfin’s abilities to create a lot of value flipping homes, and in the short-run it will probably create some diseconomies of scale with opex.


As a side note, I have the eerie feeling that one day, many, many shareholder losses will be traced back to the words “at scale” in the same way people have done for “platform” and “new loan underwriting model” and “eyeballs.” Don’t you think it’s funny how no company ever clearly defines what exactly “at scale” means? Why can’t anyone achieve scale today, why is it so impossible? These companies are already gigantic. Wayfair is going to sell something like $10 billion in furniture this year but somehow they’re not close to scale. In a world where massive fixed costs for things like computing power have been permanently altered from fixed to variable expenses, how is it that none of these online business models are achieving scale? What fixed costs specifically need to be eclipsed on a per-unit basis? As far as I can tell, companies mean “at scale” when growth slows to 0-5% - growth seems to be the definition of scale. I guess that would mean companies like Groupon, Pandora Media, and have finally reached scale? Be careful what you wish for.



I can think of some ways to value Redfin but they strike me as overly wonky and theoretical. I don’t know what this company is worth, but its current $1.5 billion EV strikes me as fantasy. It’s the same EV as RE/MAX which generates over $100 million in EBITDA. If someone (or Redfin themselves) can convince me that the average transactions per agent will increase materially I’ll change my mind. In the meantime, maybe real estate commissions are 5-6% for good reasons instead of being a grand conspiracy to fleece customers. Maybe lowering the costs of buying and selling homes by 50% isn’t economically feasible after all.


I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise do not hold a material investment in the issuer's securities.


When I cover the short will start working, probably

    sort by    
      Back to top