A UX Researcher, Writer and Consultant.
Shaping design direction for an AI Assistant for Real Estate Agents
Discover how I optimized the Auctm AI assistant for real estate agents, achieving a 93% reduction in unsaved notes and a 25% increase in task creation through targeted UX research and testing.
Methods
Stakeholder Interviews,
Surveys,
User Interviews
Tools
Miro,
Qualtrics,
Mixpanel, Dovetail
Timeline
4 weeks​​
​
Background
Auctm was in the early stage and a beta version was released with limited no. of users.
​
Auctm is an intelligent assistant that helps Agents focus on leads that need attention, creates AI-driven plans by recommending the best next step and is constantly learning from the best agents in the office to drive newer agents to higher levels of productivity.
​
​
Auctm shall help Realtors to:
-
Prioritize the most important leads and tasks.
-
Take quick actions on leads that get updated on the CRM.
-
Easily take notes on leads while on the go and log these into CRM.
Problem Statement
Although the clean interface of the Auctm Mobile app provided a quick way to perform the key tasks on the app while they were out, going on appointments and showings, there was a 25% to 40% drop in the user flow for the task completion of these key actions:
Log a note
Update actions on the lead e.g. moving the lead to the next stage
Create a task for the agent​
Key Objectives
-
Understand the needs and pain points of the customers
-
Identify the key blockers preventing them to perform key tasks on the app
Methodology
Stakeholder Interviews
Understanding the current state of product and expectations for target customer base.
​
I took interviews of the following key stakeholders:
-
Product Manager
-
Design Lead
-
Customer Success Lead
-
Engineering Manager
Survey and Screening
I conducted a survey and got responses from 50 customers
The goal was understand the demographic and working/employment status.
Took a sample size of 20 customers.
Criteria:
-
Customers actively working on >50 leads
-
Used Auctm app in last 3 months
Participants - why select this segment?​
The reason behind taking this criteria into account was because the product was catered toward high functioning agents who produced more deals hence people who work with more than 50 leads at a given time are assumed to be producing more deals. Also, they needed to have used Auctm so that I can probe in with Critical Incident technique where I will ask for participants to recall incidents therefore they needed to have used the app.
User Interviews and Usability Tests
The agenda for the interview had the following:
-
General Questions - getting to know them and their background
-
Discovery/Contextual Inquiry with CIT technique
-
Usability Test
Data Analysis
Methods used to analyse the data gathered from the user sessions:
-
Transcribing notes
-
Affinity Mapping - thematic analysis
-
Rainbow Spreadsheet
Key Customer Insights
-
Pattern of work and preference of desk time: Users who have been in this profession for several years follow a routine and have a stringent way of carrying out their daily operations. This often includes taking notes on their paper files and updating them on their CRM at the end of the day and they prefer having that desk time rather than working on a small screen. It shall be hard to get a new technology introduced in their working environment.​
-
Risk of replicating a CRM: Another risk we might eventually run into is replicating CRM-like features. A lot of the feature requests or needs that users talked about indicated replicating CRM features because that is what they have been comfortable with throughout the years.​
-
Lack of trust: 67% of the users expressed that after logging a note or a task through Auctm, they check CRM app right to see if the note was logged. Due to API issues, this is not instant(takes a minute or two sometimes).
Usability Issues
1. Agents struggle to Save a Call Log after entering it because of the position of the Save button.
Suggestions:
-
Error-prevention dialog box that prevents user from Closing the note before saving.
-
Reposition the Save button in place of Close button.
-
Enable disappearance of keyboard when user clicks outside the textbox.
​
​
​
​
​
2. Agents want to be able to edit the note once they have logged it in Auctm.
3. Providing a Task Name doesn’t make sense because the Task Type - Call, Text, Email is the task itself.
User Stories
Results & Next Steps
-
This research helped the entire organization understand how customers navigate our product, in what context do they use it and the most painful experiences they encounter.
-
The findings aided to new design directions to be explored such as chrome plugin of the AI assistant.
-
This project led to the initiation of a new project to accelerate the logging of data entered to the CRM.
-
Before this project, the product roadmap had a bunch of features to be implemented because of high demand which were now put on hold for analysis further in terms of their significance to the AI vs CRM.
-
Juan & Janet became the key roles targeted to design for in product roadmap planning and prioritization sessions.
Impact
Biggest impact of this project is how it steered the product direction to a customer focused approach.
93% reduction in unsaved notes
25% increase in Task creation