Overall Landscape: Redfin is in an interesting position in an evolving market. By pivoting to an on-staff agent team with a superior lead system, the company is able to create a couple of advantages for the consumer (in addition to advantages in the business model, which are less relevant here). The basic advantages that Redfin creates for the consumer are:
Buyer rebate for certain houses and geographic regions
Interactions with real estate agent incentivized for service not commission
Easier way to browse UI for houses
Centralized interactions and data points (i.e. homes checked, toured, paperwork, through agent all on the same portal)
At the same time, it can be argued that residential real estate is a zero-sum game. Access to more information and an increased population looking for houses might be able to create a marginally expanding overall market, but generally there is a stability to the number of buyers, sellers, and amount of time spent on looking for a house in a given part of an economic cycle. That is to say that if Redfin is to grow its total user base and transaction volume, it has to look to take share from current competitors existing in the market.
So given Redfin’s competitive advantages, where can this expansion come from?
Traditional Brokers: Traditional brokers and their top agents still control about 10% of all residential real estate transactions (Coldwell, Re/Max, etc.) while Redfin has significantly less than 1%. This is in spite of the fact that brokers offer real no advantage for the agent or the buyer: buyers are now better aware of pricing standards and the process through online information, while great agents no longer need a physical office space and are only offered a singular pipeline of good leads. In contrast, partnering with a Zillow gives these individual agents the best chance of deal flow since it’s aggregated, but the agents now have leverage over the brokerages to negotiate favorite rates over paying a Zillow.
Zillow/Trulia/Realtor: People use Zillow, in its founders own words, for transparent on-demand access to real estate data. Zillow itself isn’t a direct competitor to Redfin; their primary business model is based upon ads and individual agents paying to be contacted for certain listings. However, one of Redfin’s selling points is full vertical integration and also an easy way to browse listings based on comprehensive data points. If Redfin can acquire a small percentage of the people using Zillow for the early part of their real estate buying process and then sell them on the advantages of Redfin’s agent system while on the platform instead of finding an external agent, it could pay handsomely
Possible Ways to Differentiate:
Undercut Traditional Broker’s Word of Mouth Advantage: A strong hypothesis is that the reason that the traditional broker is able to succeed is based on years of reputation and more specifically, word of mouth for certain agents or the local branches of their brokerages. Someone looking to buy a house might reach out to one of their friends or a friend of a friend in the area they are looking to buy to get a recommendation for an agent or local brokerage branch and is bound to get to find a large number of people have bought from a specific brokerage over time. Redfin’s agent system offers all the benefits of a broker with additional advantages, and automating a built-in trust and reputation system beyond 5-star agent ratings could increase the conversion rate on the site rapidly. In short, a direct recommendation from a trusted contact still holds a lot more value than a basic ratings system for a serious purchase such as property, and bridging that gap could be largely beneficial
Attract the “serious buyers” portion of Zillow’s user base: Zillow, Trulia and Redfin’s current home browsing experience is geared heavily towards those really early in the home-buying process, with the ability to view loads of houses seamlessly through easy visualization and recommended houses features. By the time the home-buyer reaches the later parts of the process, having already toured all their options multiple times, they have a thorough understanding of their options. However, in the middle part of this funnel, where the buyer has narrowed down their options to about 15-20 or has yet to hire an agent, there really isn’t an easy way to organize and compare information on houses in a standardized way. This also applies to when the home-buyer might be trying to narrow down the specific area they are trying to buy in. It’s only possible to hire an agent once a buyer has settled on a broad geographic area, and for first-time buyers, this itself represents a huge decision which isn’t captured in current tools.
Possible Solutions:
“Social Buying”: For a given area, it might seem unlikely that a person’s direct friends have bought with Redfin or a specific agent on Redfin, even as Redfin continues to expand rapidly. However, odds are that in any area, a 3rd degree connection has had an experience with Redfin, especially in densely populated suburban areas (i.e. NY, SF, Jersey, etc.) By having a user login with Facebook and grabbing their friend list, users can see 3rd degree connections that have bought a house using one of Redfin’s agents in a certain area by overlaying these connections over the existing map feature on mobile. This saves the buyer the time of having to contact various people who have bought houses to find recommendations for great agents in the area, as well as gives a potential introduction to someone who might be able to provide great product validation for Redfin or one of its agents. This should in theory help drastically increase conversion from Redfin browsers to those who actually end up buying a house from Redfin.
Standardized Comparison Tool: For serious buyers, having a side by side comparison tool to shorten the browsing experience, as well as having an option for an agent to give their thoughts between multiple choices would be a promising tool. Fundamentally, this tool allows buyers to narrow down their realistic choices in an efficient way and takes the pros and cons list that lives in every buyer’s head and puts it in digital form. The second layer of this tool would involve assigning a standardized, quantitative “Redfin score”, that would take into account the basic financial and preference profile of the user and assign a score out of 10 or 100. Advanced iterations of this score using machine learning, would be a really easy way for users to compare their options based on their personal preferences in a quantitative way. In short, current recommendations are based on browsing history and location, but having the user’s personal preferences and financial history could help differentiate Redfin’s platform from competitors.
Furthermore, for Redfin, have the user lead the way in eliminating certain options and having a way to organize information online frees up time for Redfin agents, allowing them pursue more leads. This tool would also serve as an intermediary step for the casual browser on Redfin and actually hiring an agent to facilitate the process, which should also increase website conversion rate.
Flaws In Solutions:
From a technical perspective, this seems a little brutal. You have to cross compare each users FB ID with every other user who has bought a house in a local area’s FB ID in real time (the Facebook API says this a purely implementation server side, which means a slow-down that could affect overall page load times). The saving grace is you only need to gather this information from people who are specifically registered and have looked at/found houses on Redfin (so the lists of friends you have to cross compare are shorter). Also, as I mentioned before, in certain areas, there probably aren’t enough third degree connections using Redfin to validate having the tool available. That ends up with the user thinking the tool and by extension, Redfin, as an inferior product. The alternative is rolling out the tool market by market, which is a logistically and engineering nightmare.
There are a couple of big flaws for the comparison tool. First, creating a truly accurate standardized score matching a user’s profile with a house is an extremely engineering intensive process, which is touched upon a little more in the technical implementation section below. A second big problem is that the tool might only appeal to a small niche of users who use it at as a basic price comparison tool, which might just amplify the resources devoted to problem one.
Technical Implementation
The actual implementation of the Facebook feature will be simpler than the comparison tool because it requires adding features as opposed to modifying existing infrastructure. From a UI perspective, hopefully it only requires adding a couple of small buttons on the info section overlaying the map. The next time users login, it’s necessary to request their permissions to access their friends list as part of the Facebook API. The tricky part is optimizing when you compare the friends list of a person to those who have already bought a house. One possibility is using the Facebook API to send a GET Request which takes the two IDs and gives the common connection between the users. If the app is written in Obj-C or React Native another possibility is creating a massive hashtable. If I press the looking for button, every one of my friend’s Facebook ids is saved as a key (if it does not already exist) and then mapped to a value with my unique FB id. Thus, if the next user has a friend with a key that already exists in the hash table, I can access the FB IDS of our mutual friends. I have no idea how the app is currently built so these are just out there suggestions. From what I know, it seems smart to limit this feature to places where Redfin has a strong presence, but not in the most prominent presence because comparing the friends list for 75,000 people seems like it’s bound to create a delay in load times, no matter what the solution is.
The solution has a wide range in terms of its potential technical difficulty. For a barebones feature where the user has already created a wish list and merely wants to view the features of each house next to each other, the feature is more of a design- focused challenge. All the data would come from making the same API calls used for an individual house page, but visualizing the data in a way that makes for a comprehensive breakdown is a bigger challenge that would presumably be front-end heavy. Adding something such as a Redfin score, which would classify how each option quantitatively maps to each user’s individual need would be more complex. At first, Redfin could use a small sample of 5-10 questions to gauge the user’s profile and then run it through a slew of conditionals to come up with a basic score (i.e. if users income >50,000 and the house < 400,000 add .5 to the score), which isn’t too complex. However, coming up with a truly accurate model at scale would involve building some type of supervised machine learning model using agents for the initial classification, which would involve a big consumption of resources.
