SendSmart Web App
SendSmart is a web application utilized by Auto Dealerships to increase customer engagement through automated messaging. According to SendSmart, their solution can be expected to double web lead response rates.
Role
UX Research & Design
User Surveys & Interviews
Affinity Diagram
Empathy Map
User Personas
Signal Detection
Problem Statement
Ideation
Feature Prioritization
Competitor Analysis
Task Flows
Wireframes & Prototype
Usability Testing & Analysis
Iterations
Team
Solo Project
Tools
Figma
Miro
Google Suite
Zoom
How does SendSmart work?
The Problem
Dealership representatives have reported SendSmart doesn't allow them to organize and filter their messages and notifications. This makes it difficult to detect and navigate their assigned customers, which could result in lost opportunities.
Proposed Solution
Increase dealership representatives’ sensitivity toward detecting their assigned customers by providing ways to discern them from the pool of customers being assisted by other team members. This includes adding a personal inbox, improving the layout for key identifiers to be displayed, and adding personalized notifications.
User Research
User Surveys & Interviews
Six dealership representatives were surveyed about their use of SendSmart. Follow-up interviews were conducted after narrowing down their pain points and blockers.
Topics Explored:
What type(s) of dealer representatives use SendSmart?
How helpful is the message notification system?
How easily do dealer representatives navigate messages?
It became apparent that some responses differed depending upon job title / position.
Affinity Diagram
To analyze the data further, an affinity diagram was constructed to find patterns among responses.
Additional Key Findings:
SendSmart is used for different reasons based on job title / position
Dealer representatives use SendSmart most days on both desktop and mobile devices
Auditory & visual notifications are utilized by dealer reps on both desktop and mobile devices
Notification preferences vary by job title / position
An additional diagram showing the dealership process was made to provide context.
Empathy Maps
Using qualitative data from both user surveys and interviews, multiple empathy maps were developed to better understand the experience of the different types of dealer representatives.
BDC Representative
BDC Manager
Appraiser / Salesperson
User Personas
Each persona represents a different user segment of SendSmart. Their development was guided by the empathy maps above.
BDC Representative
BDC Manager
Appraiser / Salesperson
Defining the Problem
Signal Detection Theory
Based on the user research, there are dealership representatives who experience pain points and blockers with regard to filtering and detecting their assigned customers. This can be expressed in terms of Signal Detection Theory (SDT).
If needed, a brief overview of SDT is provided below.
Signal Detection Theory | Brief Overview
Signal Detection Theory can be used to understand situations wherein an operator detects whether a signal is present or absent. In this case study, it applies to whether a dealership representative detects their customer’s message. Their internal response is categorized as either “Yes” or “No”.
Signal and noise can be modeled using distribution curves. In most situations, there will be at least some overlap between the noise distribution and the signal distribution. The overlapped region is where breakdowns in performance occur.
“Yes” / “No” Response
“Yes” Response: If the signal is present, it’s a HIT; if the signal is absent, it’s a FALSE ALARM.
“No” Response: If the signal is present, it’s a MISS; if the signal is absent, it’s a CORRECT REJECTION.
Response Strategy (Bias)
The operator can bias their internal response to lean liberal (i.e. responding “Yes” more frequently) or conservative (i.e. responding “Yes” less frequently). This can be used as a strategy to maximize their HITS or CORRECT REJECTIONS. However, a trade-off will typically occur, resulting in a higher occurrence of FALSE ALARMS or MISSES.
Sensitivity
Sensitivity is represented in the model as d’ (d-prime). This is the measure of separation between the signal and noise distribution curves. The greater the separation, the higher the sensitivity.
When increasing sensitivity, it becomes easier for the operator to discern the signal from the noise. This improves their performance by increasing the probability of HITS and CORRECT REJECTIONS, while decreasing the chance of FALSE ALARMS and MISSES.
When decreasing sensitivity, it becomes difficult for the operator to discern the signal from the noise. This weakens their performance by decreasing the probability of HITS and CORRECT REJECTIONS, while increasing the chance of FALSE ALARMS and MISSES.
SDT Application to SendSmart
SDT applies to personas unequally.
BDC Representative
The BDC representative uses SendSmart to actively engage with customers in real-time. To do this, they must sift through the noise of notifications/messages of their teammates in order to detect their assigned customer’s signal. Because this noise is excessive, they often experience false alarms. This leads to representatives taking a more conservative response to incoming notifications and messages, resulting in an increased probability of missed opportunities.
BDC Manager
The BDC manager uses SendSmart to passively monitor their team’s conversations to ensure quality. Unlike the BDC representative, the manager is interested in all the conversations as they happen and doesn’t perceive the high volume of notifications and messages as noise. However, it is difficult to evaluate individual performance given the inability to filter messages by user.
Appraiser / Salesperson
The appraiser / salesperson refers back to conversations in SendSmart to get up-to-speed on the customer’s situation before they arrive at the dealership. Much like the BDC manager, they generally don’t perceive the high volume of notifications and messages as noise. However, if an appraiser / salesperson decides to message a customer using SendSmart, they’ll find themselves dealing with the same detection issues as the BDC representative.
Problem Statement
By framing the issues dealership representatives face in terms of SDT, it helped to develop a concise problem statement to be considered for the design phase:
It has been observed SendSmart doesn't afford dealership representatives the ability to sort and filter their notifications and messages making it difficult to detect and navigate their customers, which leads to missed opportunities.
Ideation
How Might We ... ?
How might we boost dealership representatives’ sensitivity toward detecting their customers’ signals so as to increase HITS and decrease MISSES? How might we inhibit the noise they experience so as to lower the probability of FALSE ALARMS?
How might we redesign SendSmart to result in a signal / noise distribution similar to the one below?
Current UI Design
The current UI design only displays the names of customers who have responded (blurred) No other indicators are present for determining which dealership representative is working with which customer.
Having both top and side navigation was identified as being problematic. During user interviews, it was discovered this feature leads to confusion in terms of navigational hierarchy.
“I Like, I Wish, What If?”
To generate ideas on how to approach this problem, an exercise using the “I like, I wish, what if?” method was performed.
Competitor / Familiar Application Analysis
The personas’ use of familiar applications was analyzed for the purpose of examining their existing mental models. The idea behind this exercise was to take these cognitive mappings into consideration for approaching a design solution to SendSmart.
DealerSocket and Gubagoo were selected given their in-app messaging features.
Key Takeaways:
SendSmart distinguishes itself by featuring automation as part of its messaging application
SendSmart is the only messaging app that doesn’t provide a personal inbox to users
Gabagoo implements a “My Chats” tab to separate an individual user’s messages from their teammates’
Both DealerSocket and Gubagoo notifies representatives for only their assigned customers
Both DealerSocket and Gubagoo utilize a collapsible sidebar
Feature Prioritization Matrix
Potential solutions generated in the previous exercises were placed into a prioritization matrix to determine which features should be selected. Priority was placed on features predicted to have the greatest impact (based on user research) and least complexity (in terms of design).
Using this method, the following design solutions were selected:
Display lead assignments within the inbox
Personalized notifications
Ability to sort/filter messages
Personal inbox
Collapsible sidebar
Design & Prototype
Task Flows
Two features implemented in the redesign: the ability to (1) filter messages and (2) set up personalized notifications. Flows were created for each task.
Filter Messages
Set Up Personalized Notifications
Sketches & Markups
Sketches and markups over the original design were made to better understand the spacing requirements for each added element.
Sketches
Markups
Digital Wireframes
Digital wireframes were designed and prototyped.
Improved Inbox
Various elements were added or redesigned to help boost representatives’ sensitivity toward detecting their customers’ signals.
Top navigation was integrated into side navigation to avoid confusing hierarchy
Side navbar was redesigned as collapsible to afford more space in the inbox
“Team” and “My Messages” were separated to help representatives discern their own messages from the team
Ability to filter messages by source and/or representative was added
Additional identifiers were displayed in each message field: lead source and assigned rep
Personalized Notifications
To inhibit the noise dealership representatives experience, personalized notifications were added as a way to help silence the excess amount of notifications related to their teammates’ customers.
Testing & Iteration
Usability Testing
Six usability tests were conducted to evaluate the redesign - three were performed remotely via Zoom, and three were conducted in-person. Participants were instructed to complete four different tasks.
Objective
The objective was to test and observe whether participants naturally use newly implemented features for filtering messages.
Tasks / Prompts
Task 1: Filter messages using My Messages
Where would you go to view only your conversations?
Task 2: Filter messages by user
Filter messages to determine how many contacts have been assigned to each user
Task 3: Filter messages by lead source
Determine which lead source has provided the most leads
Task 4: Change notification settings
You no longer want to receive notifications for leads assigned to other reps. How would you change your notification settings to reflect this?
Analysis
An analysis was madve on each task. Success was based on whether a participant was able to complete the task on their first attempt. Participant feedback was considered.
Iterations
Based on testing and feedback, iterations were made to the side navigation bar and notification settings.
Side Navigation Bar
The iconography was reconsidered after several participants expressed confusion and dissatisfaction with the gear symbol being used to represent “Account”. Taking this feedback, a user symbol with a squared background (to differentiate it from “Contacts”) was used to replace the gear symbol for “Account”.
Since most participants associated the gear symbol with “Settings”, most had made the error of selecting “Account” to update their notifications. Given this, notifications was given its own category under “Settings” in the sidebar (with a gear symbol).
Notifications Settings
Under “Team” notifications, participants found the wording of these options to not be descriptive or self-explanatory. “Unassigned Contacts” was changed to “Contacts that haven’t been assigned yet”; “Assigned Contacts” was changed to “Contacts assigned to my teammates”.
Reflection & Next Steps
Reflection / Limitations
Small Sample Size
Multiple personas weren’t expected to be utilized at the beginning of the project, as a single “dealership representative” persona was thought to be adequate. After further investigation, it became apparent the goals and needs of users depended upon their job title / position. As a result, six dealership representatives were considered across three different personas, averaging only two reps per persona. Due to there being a low sample size, conclusions about the personas may not generalize as hoped.
Single Dealership Consideration
The data in this case study was conveniently sampled from a single dealership as a way to streamline the research phase. While this method allowed data to be collected in a short period of time, it may have come at the cost of having a more holistic view of dealerships and the processes they implement. It could be the case that pain points one dealership experiences when using SendSmart aren’t the same pain points of another dealership. Once again, findings may not be as generalized as hoped.
Next Steps
Testing
Further testing should be conducted. In particular, the replaced iconography in the sidebar and the new wording within the notification settings should be tested to confirm if they translate well to dealership representatives.
Missed Opportunity Rate
Missed opportunities should be recorded before and after implementing the redesign to confirm if a decrease in the rate at which missed opportunities occur.
Research
Research involving a larger sample size across multiple dealerships should be conducted to increase confidence in the ability to generalize findings. This will also lead to the development personas that better represent various user segments of SendSmart.
Additional Features
Based on user research, additional features should be considered to improve dealership representatives’ experience of SendSmart. Areas to be explored include the implementation of CRM integrations and individual performance metrics.