[RDFN – Redfin] Business Model Analysis and Growth Constraints

[I am currently writing up a somewhat brief companion piece on Zillow. It will be posted next week.]  

Like healthcare and education, real estate is a massive market plagued by inefficiency, protected by incumbent interests, and resistant to reform. If you’ve ever bought or sold a home, then you have no doubt been struck by the oddity that in a fragmented landscape staffed by 2mn licensed agents and 86k real estate brokerages, fee structures seem obstinately uniform and inflated. It takes little incremental effort and expertise to sell a $1mn house than it does a $500k house, and yet the agent will receive twice the fees in the former case, as commissions are based on a percentage of the home sale. A buyer’s agent is financially motivated to maximize the price that the buyer pays. Redfin’s business model doesn’t change any part of that fundamental structure. It still recognizes most of its revenue as a take rate on the proceeds of a home sale. But, by integrating labor and technology to deliver a comparable service at lower prices the company at least makes a small dent in the $80bn of annual commissions garnered by agents and brokers.

In a typical residential real estate transaction intermediated by agents, the seller is charged a ~5%-6% commission that is split between the buying and selling agent. The agent, typically an independent contractor, will split the commission with his sponsoring brokerage, typically ~60/40 in the agent’s favor. If it is part of a franchise operation, like ReMax, the brokerage will in turn kick up some part of its commission share plus monthly franchisee fees to the franchisor. Redfin, on the other hand, charges 1% in most areas (and 1.5% in others) as the selling agent, and refunds a portion of the 2.5% it receives as the buying agent to the buyer. [A home seller rep’ed by Redfin still pays the buying agent the standard 2.5%-3%, so the seller’s all-in savings comes to 1%-2%].(1)

Like other real estate brokers, Redfin has access to all the homes listed for sale in its markets’ Multiple Listing Services (or MLS’s, private databases on which licensed brokers share information on listed properties and offer compensation for brokers who can deliver a buyer…hence the splitting of commissions), but traditional real estate brokers with offline roots don’t operate websites with nearly the sophistication or reach of Redfin’s. Visit the sites of leading brokers and broker franchisors like HomeServices of America (owned by Berkshire), Realogy (brokerage franchisor with brands that include Century 21 and Coldwell Banker), RE/MAX, Douglas Elliman, Keller Williams, etc. and prepare to travel back in time to an age of “About Us” landing pages, stultified maps with jerky zoom-in/outs, and mapless listings. Those firms have effectively outsourced online lead gen to Zillow, Facebook, etc., and still heavily rely on client referrals and personal networks.

Zillow is not a broker in its own right and so has historically relied on intermediaries like ListHub for MLS data. It has managed to secure “hundreds” of direct MLS feeds – in 2015, following an acrimonious spat(2), ListHub and Zillow terminated their feed agreement, forcing Zillow was ink deals directly with MLSs in piecemeal fashion – but the completeness of those the listings and the speed with which Zillow obtains them, varies widely by MLS. In San Diego, for instance, out of the 250 data fields in an MLS listing, Zillow only gets 50. Also, it appears that most of Zillow’s listings are still sourced from agent syndicates, brokers, and third party feeds(3), rendering it a step removed from the swift and comprehensive access that Redfin, as a licensed broker, is privileged to. Practically, what this means is that Redfin can remove claimed inventory and post new MLS homes on its own site within minutes, enabling it to notify customers about newly listed homes 3 to 18 hours faster than other real estate listing portals. And its detailed MLS data yields more accurate automated home valuation estimates to boot: according to an “independent” SSRS study commissioned by Redfin, 64% of the listings for which Redfin provided a public estimate sold within 3% of that estimate vs. 29% for Zillow and 16% for homes.com.

Given that nearly all prospective buyers start their searches online and nearly 90% end up purchasing their homes through an agent or broker, integrating a well-trafficked web portal with an efficient brokerage operation is a salient point of differentiation. Unlike either traditional brokers who don’t have a viable online presence to acquire their customers or aggregators who can’t fully monetize their visitors, Redfin is positioned at both the front and back ends of the funnel to efficiently capture and convert prospects. Visiting redfin.com, you’ll notice that the names and brands of third-party listing agents(4) who list on the site are obscured and your engagement is directed to Redfin buyside agents at various points. So if you’ve ever house shopped on Zillow, you might recall typing your contact information into an agent card and having that agent email or call you back to coordinate a visit. Redfin, rather than force third party agents and prospects into a game of phone tag and risk conversion slippage, simply offers a menu of visiting dates and times. You can schedule a home tour just as you would a restaurant reservation on OpenTable, and upon doing so, are given the option to connect with a Redfin buyside agent (and are, in fact, forced to declare whether you are already working with an agent before booking a visit).

If instead you choose to “Start an Offer” on a listing, your contact details, along with your offer price and payment means (mortgage or cash), are sent directly to a Redfin agent. If a prospective buyer schedules a tour at one home but neither signs up for tours at similar homes in the area nor subscribes to online alerts for new homes in her price range, Redfin will direct these signs of waning interest to one of its buying agents, who can give that prospect extra attention.

Besides owning both the gateway and the transaction, Redfin benefits from a second point of integration, at the brokerage level. Unlike most brokers’ agents, who work as independent contractors compensated predominantly on a commission exclusively tied to the value of the home sale, Redfin’s agents are full-fledged employees who enjoy base salaries, benefits, training, and expense reimbursements. Redfin agents are plugged into the company’s lead gen engine and so don’t have to dedicate valuable time prospecting for leads. They are supported by a centralized support staff that responds to online inquiries, beautifies listed properties, and prepares closing documents, leaving agents more time to develop clients and provide a superior level of service, which improves the odds that a lead will convert to a buyer. The average Redfin agent completes 3x the transactions of a typical brokerage agent and is paid more than twice the industry median ($90k/year vs. $40k)(5). That goes a long way toward explaining why Redfin’s 73% agent retention rate – with most of the churn occurring within the first 6 months – exceeds the industry median of 58%. A traditional broker operating without the benefit of Redfin’s infrastructure is left crudely toggling commission splits to retain agents (as was demonstrated in late 2016 when NRT – the non-franchised part of Realogy, operating under the Coldwell Banker, Corcoran, Sotheby’s International, ZipRealty, and Citi Habitats brands – was forced, after years of losing market share, to more generously share the commission pie).       

That said, an agent at an incumbent broker typically has far more years of industry experience than an agent at Redfin. Realogy, for instance, recruits agents who have worked in the field for 10+ years. This makes sense considering that traditional brokers, who don’t have robust and well-trafficked web presences, really have little choice but to rely on agent experience for leads: 2/3’s of Realogy’s leads are sourced from the personal networks of its agents; less than 10% come through the internet. However, there doesn’t seem to be much evidence that industry experience offers much of an advantage. Compared to a customer who transacts through a traditional brokerage, a customer who sells her home through Redfin is, according to management, 69% more likely to use Redfin to buy a home (up from 42% in 2016). It may be that the centralized support Redfin agents receive and the standards they are required to follow – unlike independent contractor agents, all Redfin agent-employees funnel their workflow through the company’s technology infrastructure and abide by common internal processes on how to schedule tours and price homes – more than compensate for the lack of field experience.

So Redfin’s two points of integration – between its web interface and its brokerage and between its brokerage and its agents – produces superior customer experiences compared to the more common modularized configuration of a traditional broker. Redfin sells more homes, faster, and at higher prices; its NPS score, 50, is well above the industry average of 38. The proportion of customers hiring Redfin to sell the home that the company helped them buy years earlier is significantly better than at competing brokerages, with repeat transactions growing by ~50% per year over the last 2 years. Over time, as advertising and positive word of mouth has percolated through neighborhoods, Redfin has claimed ever more share in its markets, fueling gross margin expansion. The steady progress across all cohorts and the gaping disparity in performance between the earliest and latest ones – markets opened in 2006-2008 have 8x the market penetration and nearly double the gross margins of those opened in 2014-2016 – suggest a long ramp of gross profit growth as Redfin’s more recent cohorts season.

Source: Redfin IPO prospectus

Source: ReMax investor presentation, May 2018

[Redfin, with its ~1,400 employee-agents, ~50k transactions, and less than 1% national market share, is a seedling compared to its more established peers]  

[Note, Remax and Realogy operate franchise operations that derive most of their revenue from management fees and annual dues of the underlying agents, so their business model and profit margins are not comparable to those of Redfin]

Many agents profess that a 3% commission rate properly captures the value of high quality customer service that they offer, but Redfin’s success suggests that the industry’s intractably elevated fees really just subsidize pervasive inefficiencies. Traditional brokers still insist on brand moats, and perhaps once upon a time brokerage brands signaled to home buyers a trusted hub of information, but that role has now been subsumed under internet-enabled aggregators who bring widespread price and selection transparency. And maybe legacy brands once signaled high quality service, but Redfin, which doesn’t have widespread brand recognition, even in its mature markets – according to a Redfin survey, less than 1 in 3 home buyers and sellers named redfin.com as a top 3 real estate website; less than 1 in 10 named Redfin a top 3 brokerage – has nonetheless demonstrated that it can serve customers at least as well, at lower prices. Of course, traditional brokers – or, at least the large, publicly traded franchisors, Remax and Realogy – are not unaware of the changing competitive landscape and have recently redoubled their investments in tech through internal development and acquisitions. But I don’t think they are culturally equipped or financially prepared to commit. Plus, as franchisors, they rely on brokers who hire independent contractors, granting them full control of neither brokers nor agents, making it difficult to efficiently implement productivity-enhancing innovations. Realogy’s grand rhetoric does not square with pusillanimous catch up measures like data-driven approaches to commission splits and agent productivity tools, table stakes efforts that still allow the company to abide by the strict profitability targets promised to shareholders who are not looking for radical, margin depleting makeovers.

By contrast, technology is embedded in Redfin’s process: the company analyzes billions of customer interactions and has over 100 experiments running at any given time to inform high return uses of marketing dollars. Redfin’s infrastructure is leveraged by agents to convert leads to commission-paying transactions at greater than industry levels, which means it can charge lower commissions than its competitors without compromising profitability, build market share, and recycle incremental profits into agents, technology, and marketing to further boost leads and conversions. That said, Redfin still relies on Google and Facebook for traffic and will need to get better and better at conversion just to maintain ROIs as ad rates continue to escalate. Also, as Redfin has gotten larger, it has seen y/y traffic growth dramatically decelerate, from 43% in 2q17 to just 18% in 2q18, as incremental traffic must now be increasingly yanked from well established real estate portals like Zillow. In an attempt to offset this alarming trend, Redfin is growing proprietary listings, weaning itself off the same MLS listings that all the other real estate portals use(6). It is also leaning on mass media campaigns, which seem to be showing promise with the cost per customer touch from mass media 25% lower than it was a year ago. As Redfin’s listing share climbs in a given market, brand awareness will be further fueled by the growing number of Redfin signs that litter neighborhood lawns. So, while the company still generates the same amount of EBIT losses as it did 4 years ago, scale economies are evident:

Still, I don’t want to overstate this point. After all, this is a relationship business that requires handholding – a while back, Redfin had customers host their own open houses; it didn’t take – and agents can only serve so many customers a day. This limitation is exacerbated in tight housing markets: as a commission-based business, Redfin’s success is ultimately predicated on how many of its customers purchase homes, and when inventory is hard to come by, Redfin is forced to throw bodies at the problem of guiding dispirited clients to successful outcomes (i.e., getting them to bid more or bid on other homes). And with current housing inventory at half of normalized levels, Redfin has recently cut the number of prospective homebuyers per agent by 10%, diluting agent productivity and anchoring gross margin expansion in the process. While this strategy of more intensively engaging the customer has invigorated re-engagements (that is, prospects who return for a second meeting with an agent, a leading indicator of closings), it has not yet translated into stronger close rates.

Redfin’s business model has actually been tried before, with mediocre results. In 1999, five years before Redfin was founded, a real estate broker called ZipRealty was founded on the belief that agents armed with productivity-enhancing technology could charge lower commissions and make up for it with faster turns. ZipRealty charged sellers 25% less than the market average while rebating buyers 20% of commissions upon closing. Zip operated a well-trafficked website(7) of MLS listings that were supplemented with relevant third party and proprietary information like neighborhood crime states and price trends, and funneled consumers to ZipRealty’s employee-agents (ZipAgents) who managed all their workflow on the company’s proprietary systems. By studying site behavior and email communications, Zip could rate a prospect’s likelihood to transact, allowing its agents to focus on the highest value leads. From the website, prospects could schedule dates and times to see homes and even submit offers online. As the prospect moved through the transaction process, in the background ZipRealty would deal with all the forms and signatures, leaving ZipAgents free to serve clients. Sound familiar? This worked all right for some time. Transactions nearly doubled from 12k in 2005 to over 23k at its peak in 2009, while the number the agents more than doubled from ~1,400 to over 3,000 (peaking at 3,400 a year later, in 2010). But these metrics unraveled shortly thereafter. Beginning in 2010, the company was forced to adopt an independent contractor model, variablizing its cost structure while also unbundling the technology/online marketing part of its business from the brokerage. In 2011/2012, ZipRealty closed offices around the country, exiting nearly half its markets, and started licensing its technology to Realogy franchisees, who used it to run their brokerage operations. Zip was acquired in June 2014 for $166mn by Realogy, who has rolled out what is now a CRM and lead gen tool to over 250k of its agents.

But even in the boom times, things at ZipRealty were never as good as they appeared, with agent growth far outpacing revenue. The amount of revenue ZipRealty generated per agent and per transaction steadily declined while EBIT losses mounted.

In essence, Redfin runs nearly the same business model that Zip did in 2006 but seems to be implementing it far better. An “engaged” (really good) ZipAgent bought or sold 14 homes and made $65k per year. For a typical Redfin agent, the respective numbers are more like 30 and $90k. Redfin completes twice as many transactions as Zip did at its peak with 1/3 the agents. If an exceptional ZipAgent transacted 14 homes a year, maybe the average ZipAgent did more like 8, which is about typical for the industry…and if the average real estate agent makes $40k/year on 8 transactions, then you can see how offering 20%-25% discounts might have created an untenable compensation arrangement for ZipAgents, especially considering that Zip homes tended to skew towards the lower end. If you’re going to compete on price, productivity is everything. Ok, so maybe Redfin is running its shop better than a former competitor. That’s still no reason to buy the stock. More pertinent than the question of whether Redfin’s pairing of technology and real estate agency can “work” (it obviously can, Redfin has been at this for 15 years), is whether Redfin is doing something different enough to allow it to sustainably claim gobs of value in a mature and insulated industry. It is on this point that I must demur. Glenn Kelman, Redfin’s CEO, refers to the company as the “Amazon of real estate” but what I’ve described up to now is a brokerage using technology to do the same ol’ thing, only better. Redfin is riding scale economies to undercut staid competitors, but there is nothing fundamentally different about the configuration of its core business model. It has bundled mortgage origination and escrow agency in recent years, but these moves aren’t particularly novel in the industry.

Redfin’s recent capital intensive experimental foray into direct home purchases (Redfin Now) – where it will use its own balance sheet to purchase a seller’s home for less than its market value with the hope of re-selling it at a profit(8) – is certainly an abrupt change from the traditional broker model, but is it a good change? Or is it simply a reaction to startups like Opendoor, OfferPad, and Knock, who have raised over $1bn in venture capital over the last 4 years to fund some version of a “buy it now” model, and Zillow, which launched its own Instant Offers program last year? In replacing a buying agent with principal capital, this strategy is as subversive to the existing paradigm as it is risky for those committing serious capital to it in the throes of a seller’s market. Any competitive advantage that Redfin can now claim rests squarely in its ability to extract more productivity out of the traditional agent scheme and up until recently, management seemed diffident, borderline skeptical, about a system that compromised agent utilization, circumscribing the addressable market for Redfin Now at less than 10% of all home sales and describing the program as a mere “website feature”. But the company is proceeding more vigorously, now that it has apparently convinced itself of its putative competitive advantage: unlike Zillow, it has transaction expertise and unlike Opendoor, it has the audience. After seeding the program with $10mn in capital, it upped the capital limit to $20mn, then $25mn, and has for now settled on $35mn.

So far, all we can really say is about Redfin Now is that it is not a disaster. All 17 homes that the company purchased in 1q18 were sold in 2q18 at higher prices than it paid, yielding breakeven profits after the cost of renovations, labor, and capital. By scaling and more efficiently managing the renovation process, management thinks it can eventually realize 1%-2% of the value of the flipped house, yielding profit dollars that are roughly equivalent to what it current recognizes as an agent. In other words, relative to its current agency business, Redfin Now is both less of a sure thing and requires a capital commitment. Redfin hasn’t yet run into Zillow, but this is just a matter of time. And a home owner trying to decide between selling her home to Redfin or Zillow (or Opendoor or OfferPad, etc.) will simply go with whoever offers the best price(9). Earning excess returns in what is sure to be a market characterized by vigorous price competition will come down to having the lowest cost structure. And I have no basis for assuming that Redfin will possess such an advantage.

Overall, Redfin has some semblance of a moat: incumbent brokers trying to replicate Redfin would struggle under overwhelming cultural, organizational, and financial burdens, while startups would face the generic challenges of creating brand awareness and scale. But, I’m certainly not awed. The MLS listings that comprise Redfin’s site are available to any licensed broker and from what I can tell, there aren’t meaningful self-reinforcing dynamics to protect and further perpetuate scale economies.

Redfin’s growth opportunity is straightforward. Even in Seattle, its most mature market, the company has less than 5% share and nationally, it is less than 1%. When asked to name 3 real estate brokerages, only 8% of those surveyed in Redfin’s markets named Redfin. But 8% is a big improvement from just 4% in 2016, and as the company’s brand awareness and transaction share continues to grow, it will presumably scale its fixed costs (though, every year for at least the last 4 years, Redfin has been generating losses and burning cash, even while its revenue has more than tripled). In its most mature market cohorts, Redfin’s real estate gross margins are 35% (vs. 28% for the company as a whole today). Layer in some opex leverage and maybe you can impute mature operating margins of 15% (operating leverage is constrained by how many houses a real estate agent can flip in a given year). Even after declining over 50% ytd, Redfin’s stock is still priced for above-industry growth, though it is cheap if you simply extrapolate the 30% growth in brokerage real estate revenue that the company has generated ytd. Trivially, 15% revenue growth over 6 years with 15% operating margins on year 6 revenue at 20x implies low/mid-teens annualized returns. Agent productivity is key to this outcome and there are basically two ways to drive this higher: 1/ attract more leads per agent and/or 2/ boost the conversion of leads to sales. With respect to the latter, Redfin agents are already transacting several times more homes than their peers and there are physical limits to the number of transactions an agent can close. I fear that the accelerating y/y declines in this metric over the last 4 quarters, while partly self-imposed as discussed earlier, may also in part be due to the fact that agent productivity is near a structural plateau:

And with respect to the former, leads depend on site traffic, and as mentioned earlier growing traffic at Redfin’s current size means pulling traffic away from established portals, namely Zillow, whom even Redfin admits has made improvements in listings quality. Assuming constant lead conversion, you might look at traffic growth as a leading indicator of transaction growth:

Footnotes

(1) More of Redfin’s brokerage transactions come from the buy-side than the sell-side, reflecting its 2004 origins as a map-based real estate search service, though mix has shifted over the years towards seller transactions, which have gone from 25% of transactions in 2015 to 35% in 2017 [even so, all else equal, Redfin pulls more revenue out of each buy-side transaction, so brokerage revenue is even more tiled towards the buy].

(2) ListHub is owned by Move.com, which competes directly with Zillow-owned Trulia

(3) Zillow’s Listing FAQs notes: “Listings are sent to Zillow from sources such as partners, brokers, agents, third-party “feed” providers, and in some cases, directly from an MLS”. (emphasis mine)

(4) Redfin partners with outside agents to reach customers in areas that are, for now, outside Redfin’s domain. The referral fees that these third party agents pay to Redfin represent just 6% of the company’s real estate revenue and are only growing half as fast as revenue from its own brokerage.

(5) 3/4 of a Redfin agent’s compensation is variable, a function of the number of closed deals and customer satisfaction.

(6) Notably, following a disagreement of one kind or another, Redfin agents no longer advertise on Zillow, a development that Redfin’s management claims has not effected its business.

(7) In 2010, according to Hitwise, ZipRealty’s website received more traffic than any other residential real estate broker.

(8) The seller is willing to accept a discount to market value for the convenience of a fast, hassle free close (maybe she needs to extract equity from her existing home to purchase a new home).

(9) Let’s safely assume that Zillow and Redfin can both complete a home buy in the same amount of time.