CROSS STREET MOBILE APP
A mobile app to empower real estate agents with smart searching, automated scheduling, and streamlined messaging. Realtors can freely meet with clients in the field instead of being stuck at the office tied to a desktop based workflow.
I was the sole designer for this project and was responsible for the full design process from research through design, prototyping, and user testing.
The flowchart above maps out the steps a buying agent must complete to schedule a showing for a buyer client. Each buyer client will see on average 30 homes, (about 5-8 a week) before making a purchase and each of those homes are represented by a listing agent. So the buying agent will be contacting each of these listing agents, but they typically work with 2-5 clients in a given month. With 60-100 homes and agents on a monthly basis, this problem gets complicated really quickly as there isn’t a standardized way to keep track of all of the different showing requests, the status of each request, and ongoing conversation between the buying agent and listing agent.
Nearly every agent needs to establish their own ‘lifehack’ or workflow to address the chaos set before them, but much of it is still really manual due to the limitations of the MLS (Multiple Listing Service) system where all homes are listed.
Build and validate a MVP mobile app that empowers real estate agents to streamline their daily workflow of searching for homes, scheduling showings, communicating with clients and other agents, and managing client data so that agents can close more deals.
Cross Street (mobile app) was a second iteration product after building and testing Avenue. Many of the decisions made across the project were influenced by the feedback and learning gained from Avenue. At a high level, we needed to go from a webapp to native mobile apps in order leverage more of the interaction and performance expectations from our users.
In order to build an MVP for both iOS & Android, we had to prioritize features that were more unique in value to acquire users and then build the more standard real estate app features like search by filter and saved searches later on.
USER RESEARCH & ASSUMPTIONS
Where a decade ago agents were in control of sending properties to their clients, realtors no longer feel that they're perceived as necessary and knowledgable to their clients since sites like Zillow, Redfin, and Realtor.com publicly offer property data. They had to pivot from being the gatekeepers of knowledge to becoming the middleman to get their clients every showing they request.
I conducted user interviews with real estate agents, brokers, buyers, and sellers to understand their pain points and processes when dealing with buying and selling homes. From speaking with these 20 individuals, I distilled our learning into user personas, specifying each type of user's goals and needs were. But ultimately focusing on building the app around the buying agent persona because they were the most burdened and desperate for a solution. (A real estate agent actually handles both roles and responsibilities of buying and selling agents, it just depends on if their client is a buyer or seller. Working with buyers is more common since it’s much more difficult to ‘win’ a listing and represent a seller.)
Some assumptions we wanted to validate with agents were:
Scheduling showings for clients is tedious and should be automated as it would save them hours per week to find and service more clients instead.
Communication with their clients and other agents is chaotic since they interact with a lot of different parties without saving their contact information.
Agents would like to have a digital assistant to help them with administrative tasks such as prospecting potential homes, coordinating their schedule and corresponding client notes, and following up with existing clients.
Popular real estate platforms such as Redfin, Zillow, and Open Listings are geared towards consumers and certain MLS boards have their own app for agents specifically such as HomeSpotter, HomeSnap, and MLS Touch. They have all the house data that agents need as some of it is private to only realtors, however outside of holding data and having a basic CRM where clients can react to properties, they lack the heavy lifting required of scheduling multiple showings for a client.
ShowingTime and AgentInbox are the most direct competitors as they are apps that focus on the phase of scheduling & communication with each one focusing on one more than the other. The vision of Cross Street was to work with the existing mental model that agents held that their client communications were centralized around the homes they wanted to see and they would also be messaging other agents for those same properties they are requesting to show.
1-10 years experience in real estate
works with 3-10 buying clients a month
Very Type-A personality, needs to lean towards being a control freak to make sure there are no mistakes with the details in the contracts.
Work schedule is very sporadic, days are often pretty long, but they're working in chunks of 30 mins to 2hrs depending on the day.
Beginning of the work week is slower with planning and responding to clients, end of week is more being outside of the office with showings and client meetings.
Agents go through a general progression from working with entry level buyers to later representing sellers and eventually higher end luxury priced listings.
Comfortable with technology to help alleviate all the different kinds of tasks required of them. Tech solutions and knowledge is shared mainly through their brokerage and other colleagues in the office.
Goes into the office 2-3 days a week, depending on what is going on. If he more escrows and closings, he'll be in the office, otherwise he's out meeting clients and showing homes.
With a sporadic schedule, some deals he has to drop everything and get it done regardless of where he is, even on his personal time. Work life balance is difficult to manage because clients are naturally more free after their 9-5 work hours.
There's a lot of co-dependency on others: clients, agents, escrow officers to respond timely. With much back and forth interactions for each transaction, there's no guarantee to get another party to respond so there's a lot of dead time and unnecessary follow up.
Between building client relationships, marketing, showings, meetings, contracts, email correspondence, comparables analysis, and home searches there's always more to be done and managed than one person can handle efficiently.
Growing his clientele, building long term relationships with buyers as they will eventually become sellers.
Aspires to develop a team and have people under him to grow his business. Perhaps adding consultants that handle administration and marketing tasks so that he can focus on getting more clients.
Having a reliable assistant to follow up and check in with client needs so that the lead stays warm.
Cross Street transforms realtors' workflows through automated showing scheduling and a unified messaging platform. Agents save hours each day as Cross Street removes the burden of back and forth coordinating between multiple parties while keeping all of your conversations in context to each property your client wants to see. A tour is scheduled within seconds and agents can sit back while the Cross Street smart assistant confirms showings seamlessly.
SOLUTIONS - “HOW MIGHT WE…”
Solutions to these problems we considered were:
Reduce the manual task of calling, emailing, and texting other agents to confirm and reschedule showing appointments
Automate scheduling for showing requests as well as shuffling time slots to make a collection of showings work with their client's availability.
Address the chaotic nature of communication with other listing agents who the buying agents do not save their contact information
Organize messaging where each conversation has context to the property and client who is interested in buying or selling the property. Utilize existing MLS data to populate profiles of active agents so that there's always a name to each property.
Improve the responsiveness from other parties so that the buying agent can move forward with closing the deal
Enable one-click or smart confirmations, cancellations or rescheduling for other parties.
Maintain a system of client activity and communication that snowballs at an exponential rate
Enhance a CRM experience that synchronizes all property showings, conversations, notes, and transactional activity around homes of interest and then suggest properties based on a learning algorithm from the client's activity data.
With a majority of buyers and sellers avidly using Zillow, Redfin, and Trulia to find homes, we prioritized our build and design with automated scheduling first because it drew the most excitement in our user interviews and it had the least competitors already in that space. Messaging was the next feature on our priority list since scheduling appointments took place through the medium of text messaging and so these features needed to work harmoniously together in order to have a functional user experience.
Avenue started with a full featured search experience built around a realtor's workflow, however the time and resources spent for a very common feature led us to build a product that was more unique given the limited runway we had as a startup. The founders of Cross Street envisioned the product to be a platform for the entire real estate transaction to take place, but the MVP was to strategically focus on the step of going to show homes, communication between parties, up to the step of submitting an offer.
Above are some examples of sketches during our ideation design studios to create a tour, manage client contacts, layout the overall user journey, a user flow for rescheduling, and an interaction pattern to edit a tour.
One of the first initial flows from transitioning from Avenue to Cross Street with basic wireframes.
BUILD & PROTOTYPE
After doing a branding exercise as a team we established these principles to guide design decisions for style, tone, and interaction. These principles were selected based on our user research and persona of what realtors value as well as how we position ourselves to other tech products in the industry. In order to unify the visual experience on all platforms and streamline the handoff process with engineers, I started with building a component library that I planned on detailing out into a full design system when time allowed.
An overview of the main UI screens for the Android app. (From left to right) Automated Tour Itinerary, Organized Messaging, Personal Smart Assistant Responses, Agent Focused MLS Listings, and Status Updates of Upcoming Tours.
Responsive web UI for buyer clients (non-agents) who are not on the Cross Street platform.
TEST & LEARN
USER TESTING: INVISION PROTOTYPE
I conducted initial user tests with 10 realtors using an Invision prototype to gather feedback validate the solution of a automated tour scheduler and messaging platform. The test was to measure the user flow of creating tours and messaging other agents. There were a lot of questions regarding how the app would perform the automation in the backend since each realtor had their own workflows of how they setup showing appointments and getting them all confirmed. We used that time of questions and answer as valuable learning for us to improve the UX by documenting these edge cases to be included of how the backend logic should weigh into our tour creation algorithm. Specifically, agents had a range of how long they would stay at certain properties more than others, however we pushed this feature request to a later release as a baseline of 30 minutes was broad enough to work with.
Overall, we had very positive feedback from realtors because they saw Cross Street as a solution that they could use. They were excited to see how the app would turn out and were willing to pay a subscription cost, which validated our unique value proposition. There were small details that we identified and revised regarding copy, iconography, and screen flows from this session of user testing.
USER TESTING: BETA TEST GROUP
After building out our MVP and releasing our beta on iOS and Android, the feedback we received was mixed. The concept was still exciting and valuable for realtors to see it live and working in their hands, but when they ran into multiple errors they became frustrated since now they have created a new problem for themselves to fix. This is expected as a beta experience, however realtors became less invested to use the app with their real clients due to the experienced instability. We quickly learned that portions of the app needed further development to service more edge cases than we had initially outlined for our MVP. Scheduling is a challenging problem to solve since the context significantly complicates the puzzle as additional factors are required to be part of our algorithm.
The three primary areas we had to address that affected the user experience were (ordered from most urgent to least):
The digital assistant which handled appointment confirmations had inconsistent success. We utilized the Natural Language Processing engine based on the Alexa platform that we had trained through seeding realtor appointment conversations. When it could not confirm an appointment time, we created a graceful fallback called 'Takeover' mode where the user needed to step in and confirm, reschedule, or cancel the appointment manually through a button group UI element and private notification. Even after several rounds of testing and iteration to train the assistant, the success rate was about 75% to either confirm or fallback into 'Takeover' mode. With the assistant making a mistake about 25% of the time, realtors felt that fixing the assistant's error was a disappointing experience as now it has caused confusion for other parties.
Our tour creation algorithm needed to be more dynamic. From our MVP build, users were satisfied and even excited with successful tours created, however the baseline functionality was not robust enough to handle their specific rules to each property they're scheduling an appointment for. They requested to preview the tour route, order the showings and enter their own start time before submitting.
Users wanted to have a map to search to look up properties before creating a tour. They didn't request to have a full search experience with saved searches and filters, but just a lookup on a map. Their mental model of finding a property on the MLS is to go from Look up Property -> Create Showing Appointment/Tour instead of Create Showing Appointment/Tour -> Look up Property.
We learned that points 2 & 3 in the list above were features that we did not include in the initial Invision prototype became real issues to our larger group of beta testers. Naturally, there is a difference between conducting a user test with a prototype versus testing with a live build, especially one that is used for productivity with real clients and appointments. So we listened to our users and revised the user experience by expanding the user flow to have a property look up as the initial start screen that’s separated from Tours. We improved our digital assistant by including more edge case scenarios and constraints while having users utilize the 'Takeover' mode more frequently, which they claimed as fine with them.
LEARNINGS & TAKEAWAYS
In hindsight, our greatest challenges in validating the MVP was anticipating technical complexities in the scheduling automation algorithm that worked in conjunction with the digital assistant experience. It was only after we shipped out the beta experience that we saw the gap of what users expected that we needed to hit.
To address this issue, if I were to do this differently, I would really push for a standalone prototype that we could test automation and assistant flows with users. The feedback from the static Invision prototype served its purpose to validating the user journey, but the features that were more complex and had more edge cases needed a period of QA and UX testing parallel to level of technical complexity. We were patching, fixing, and iterating this feature as fast as we could through our beta release, but should have allocated more time to do field testing and iterate even before releasing it as part of the beta.
One Step Backwards for two steps forwards
Ultimately, we decided to pivot our focus and validate a fully manual tour experience (as seen in the user journey and mockups above) so that each realtor can personalize how they create tours, their appointment requests, showing times and routes. Our team decided that by scaling back the experience we are able to provide a more stable experience to build value and acquire users and then introduce automation later on after the QA issues were more ironed out.
Feedback from the fully manual experience through an Invision prototype was positive as users did not have a tech solution for scheduling showings, however there was reluctancy for some realtors to pay for the subscription service as they stated that they already get a few apps from their MLS board for free. The business model struggled to be viable even with an experience that was valuable and delightful due to the existing costs of being a realtor. With the limited funds remaining in the project, the company pivoted towards Cross Street Lending.