Expanding Year Up United’s capabilities to connect more historically disadvantaged young adults to their dream careers.

I owned the full end-to-end research and design of the 12 month long Predictive Matching project as the Lead UX Designer. I addressed key usability issues, recommended business process improvements, and transformed the UI of our primary “Matching” tool on Salesforce.

For staff at Year Up United who increasingly rely on heavy amounts of data to make informed decisions for the ever-growing pool of young adults we serve, this revamp was long overdue.

The new designs I implemented exceeded our anticipated KPI expectations:

  • Reduced time spent on task by 47%.

  • Saved our staff time spent per business cycle by 16.3%.

  • Increased the number of young adults we matched to an internship per hour by 35.3%.

  • Increased staff satisfaction of our tool from 55% to 88%.

  • Increased perceived value of our tool from 40% to 91%.

Background

Year Up United, a nonprofit organization committed to empowering underserved, low-income young adults by providing them with professional training and corporate internships, plans to scale its operations by 10x. This ambitious growth goal demands a transformation of its internship “Matching” process and product experiences to improve efficiency, scalability, and consistency.

Core Problems:

  1. Operational Inefficiency: The Matching process for internships is time-intensive, requiring significant staff hours. Inconsistent processes across markets hinder scalability. Staff were often performing their Matching duties outside of Salesforce, such as recording important information offline or using Excel spreadsheets.

  2. Suboptimal Tool Leads to Suboptimal Matching Decisions: The primary Matching tool has consistently garnered considerable criticism. Staff have complained that it was confusing, hard to work with, and made things feel more challenging rather than supportive. Coupled with the fact that Salesforce requires a learning curve, our staff were facing an uphill battle.

Matching Tool Research & Design Improvement

To fulfill the current needs of our Matching staff, I set to work improving the design and experience of the Matching tool. I needed to find main usability issues and pain points with the platform in order to address them.

Goal

  • Create and test with staff an improved UI that emphasizes visual information and addresses current usability issues

Researching Matching and how to polish it

I synthesized from our survey responses these initial sentiment scores and metrics:

Introducing “Fit Scores” to the design

I set up meetings with Ajay, Walter, Tara, and AdeptID to discuss the feasibility of this idea.

I informed them based upon my user research the importance of the Internship Survey to our users and how these numerical performance ratings were a necessary facet for Matching.

I recommended that these ratings should be used as visual scores on the new design to help users quickly zone in on quantitative data that mattered most to them during Matching.

AdeptID and Ajay confirmed that this approach would not be a problem and an API would be created to automate quantifying the Survey ratings into Salesforce.

Thus, “Fit Scores” were born.

Designing a new Matching Tool solution

To begin Matching, staff open the Matching Interface object in Salesforce. Staff first scan and verify which Internships need to be Matched. With this view, it is very easy to lose one’s place or misidentify Internships as text sizing and limitations cut off key pieces of info. Additionally, the heavy amount of text entries cause significant information overload, overwhelming the user.

Users

  • 21 Program staff

  • Located in Austin, Boston, NYC, and Arizona offices

In this phase, I worked closely with my developer counterpart Ajay, who was well-versed in the Salesforce technical environment.

Given the unique working environment of Matching, I knew that this design would have to be a custom build on Salesforce to fulfill the requirements I envisioned.

I wireframed a three-panel view where users would be able to access Internship, Student, and the Matching functionality all in one comprehensive view. I believed this would help users more easily differentiate information based on their needs.

After confirming with Ajay that this new information architecture was technically sound on Salesforce, I demoed it with my broader product team.

The proposed design received an enthusiastic greenlight from our Product Manager Tara.

Constraints

  • Design must be Salesforce compatible and can be shipped to developers with minimal overhead

  • Design must not be a “List” view

Part of this organizational move towards operational efficiency meant that Year Up United was deeply interested in investing into AI and machine learning to bolster our Matching tool’s capabilities.

We hired a talented team of specialists from AdeptID who acted as our consultants and SMEs on AI and to assist in building an API that would seamlessly integrate into our product.

Project Structure

Our executives communicated to us the urgency for producing an updated Matching Tool to roll-out across the organization that would address current pain points of the platform and any business process recommendations for the full Matching timeline.

As such, my objective for this was to create a research study that delivered solutions for both UI design and user workflow improvements.

Staff Business Workflow Improvement

Improvements for just the interface will not fix a broken process. Our research activities were two-pronged in that they not only garnered insights onto the Tool, but also the user experience with the Matching workflow holistically. I needed to understand how our staff worked and what obstacles they faced to come up with recommendations on how to subvert them.

Goal

  • Research full end-to-end Matching workflow

  • Produce process improvement recommendations to address workflow pain points

Timeline

I collaborated closely with Ajay, and together we used the Lightning App Builder to recreate the Figma UI onto Salesforce. In this phase, we engaged in 3 working sessions where Ajay and I built how elements should interact and primary user flows. I also enlisted support and advice from Walter, who was an expert in the types of information our Program Managers regularly engaged in.

Our efforts produced a working Salesforce prototype, which I planned to use for usability testing.

The most important task we wanted to test was matching a Student to a Seat. In the sample task flow below, a user is seeking to fill a Seat with an appropriate Student candidate. Let’s say they want to match an open Apple role.

I conducted the follow user research activities to achieve the goals set out for me in this project. Now, let’s break this down as to how this was all executed.

Identifying pain points with the Matching experience

We had no quantitative historical knowledge and KPIs to leverage about how our Program staff overall felt about Matching, experienced its process, and their perception of the tool available to them.

As a quick way to collect this information, I created and launched two surveys: one to capture insights at the beginning of a Matching cycle before the launch of an updated prototype, and one timed for after Matching and testing the new prototype.

Key questions to establish this baseline included:

  • How much time (in minutes) did you spend in Matching meetings?

  • How many total Matching meetings did you have this week?

  • What went well/did not go well with Matching?

  • On a scale of 1 - 5, how valuable do you perceive the Matching Tool?

  • On a scale of 1 - 5, how satisfied are you with the Matching Tool?

Staff were expected to fill out this survey once a week over a 4-week Matching cycle. The expectation was the same for after launching our prototype.

My immediate priority was to first understand the following:

  • How do our staff currently use and respond to the Matching Tool?

  • What are the main challenges our users face with the Matching Tool?

  • How does the Matching process execute from beginning to end and what are its goals?

  • What is the overall feedback and perception from users on the Matching process?

The preliminary insights from the surveys and observation studies established a clearer background into Matching. To gather comprehensive qualitative data, I interviewed 5 Program staff members with questions such as, but not limited to:

  • What actions do you take to prepare for Matching?

  • What are the main milestones for Matching?

  • What kind of data is useful for Matching?

  • What aspects of Matching are challenging to you?

  • How do you use the Matching Tool currently?

  • What would make the Matching Tool more useful to you?

I then performed an affinity mapping exercise to uncover major themes and pain points. Several important insights emerged:

Defining further the workflow challenges of Matching

Our users additionally shared qualitative information on pain points associated with Matching. As these were survey responses, they were not detailed, but provided me with an idea of what to investigate.

I prepared to substantiate my findings further by performing both user interviews and observation studies of our users, to find out more about their behaviors, concerns, and the type of information they prepare and present for Matching meetings.

Redesigning the messy UI with a revamped information architecture

The very first thought I had was to create a “one-stop shop” layout that allowed users to easily access all pieces of info and perform necessary actions in one page. I was particularly focused on feedback I had received from users on how to maximize visual cues and breaking down content into a digestible format.

Process Issues

  • Data sharing is not centralized and staff have varying access levels; different departments have access to different info

  • Lack of internship opportunities are a significant hindrance to effective Matching, as there are more Students than Internships available

  • Outside of meeting during Matching sessions, there is no standardized framework for cross-collaboration with other working teams

At this point, I was ready to begin designing for a new Matching Tool concept. However, I needed my Business Services Manager counterpart Walter’s advice and expertise on how to address the process-oriented problems of Matching. To aid in our collaboration, I constructed a full end-to-end process flow of our Program Managers to help visualize the Matching timeline and tasks associated with it.

Staff then must select an Internship via the Internship Work Site link on the previous view, which opens a different tab. Only then are they able to see information on available Students to Match. It is here that users review qualified Students, but if they need to do this, they must open each Student profile and then close it, one at a time. The other option is to open this page in multiple tabs to keep Student profiles open.

Our users need to achieve three core actions: find internships, review candidates, and then Match candidates to internships.

From my research, users made it evident that the Matching Tool did not perform all three tasks easily or effectively. Information was simply not easily accessible, forcing users to open an unnecessary number of pages to achieve these tasks.

I recommended to my Product Manager Tara and Walter that we share these results with the Program department to prepare for solutioning. We agreed that they would take this over to Program while I started focusing on design for the Tool.

The idea was to table any process change recommendations until after a prototype had been produced and vetted.

I conducted 9.5 hours of fly-on-the-wall observation sessions of Matching calls. Staff were asked to present their screen so that I could see what information they were looking at. I organized my notes by categorizing observations as Tasks, Process, and Pain Points, while leaving a space to record any questions I had.

From these sessions, I learned further that our users struggled with:

  • Confusion where to find the right information in the Matching Tool

  • Collaboration between client-facing staff and student-facing staff seemed disjointed

  • Longwinded discussions that rely heavily upon evaluating a young adult’s “fit” for an internship and then making a Match

Design Issues

  • The Matching Tool is overly text-heavy, with unnecessary info and limited visual cues to interpret data

  • Users have to navigate through multiple pages to review Student qualifications and even perform the action of Matching

  • Filtering and categorization is non-existent, indicating a need for information organization capabilities

  • Internship Surveys submitted into Salesforce exist only as a read-only document, though users have expressed a desire to use its data to help find qualified candidates more efficiently

With Walter’s assistance, we worked together to pinpoint where our users were experiencing challenges within the Matching process. We utilized my research to serve as a reference for constructing this.

Walter frequently worked alongside Program Managers, so I used his knowledge on their working timelines to corroborate and verify these pain points.

After gathering his thoughts, I created a full user journey map to visualize our staff’s experience with Matching. The journey map emphasized the following based on my qualitative and quantitative data:

  • Significant disconnect exists throughout Matching due to information access limitations

  • The middle of the Matching cycle presents the greatest challenges when Program Managers have to connect with client-facing staff, indicating a poor collaborative environment

  • Quality of Matches suffers because more time is spent on fact-checking and ensuring all teams are aligned outside of Salesforce, which should be the main source of truth

  • The emergence of the “seat gap” phenomenon, where users feel pressure from the fact that there is a limited pool of internships available

I presented this visual artifact to the broader Agile team to highlight the problems inherent in the Matching workflow.

Though the act of actually Matching is the most important function, this action is not readily apparent until users open the Matching tab. After selecting a qualified Student, users must select the Matching tab which then opens up another page. In this view, users are finally able to make a Match. Keep in mind that in order to fact-check or perform one last review, users must return to their previously opened tabs.

Through a quick and iterative design process, I moved from producing wireframes, prototyping, and then usability testing with our target audience.

Let’s first explore how the Matching Tool operated.

Matching info is broken down based upon whether it is Internship and Student-based, with the Matching functionality in its own panel. Users will see birds-eye info on Students and Internships through “profile cards” that provide an overview, while the Matching panel will host more detailed data on these categories. This way, users have the option of quickly scanning data or investigating further, while presenting the Matching action in an accessible location

  1. Both Seat and Student information will be represented by profile “cards” - Users will see a high-level overview of Seat and Student profiles so that they can scan quickly pertinent info during a Matching session. Students will be represented by their photos which humanizes the process

  2. A sorting function based upon the competency staff wish to organize entries - 100% of users we engaged wished to see a way where they could organize Students, from highest to lowest scores, by the 6 performance categories they currently rate

  3. Filtering and search capabilities is introduced to finetune user needs - By adding search and filtering, users can more closely examine and search for specific types of Students to complement the sorting functionality

  4. Providing a way for our staff to Match students right off the bat - Taking the action of Matching is the most important step in Matching after reviewing qualifications, yet this function was hidden in the original view and required users to navigate to a separate tab. In this design, they can Match a Student while still viewing all info they need in the same page

  5. A section that breaks down the complex information taxonomy of Matching data - Effective Matching is substantiated by a significant number of documents and data on Internships and Students alike. Here, users can closely look at specific Student and Seat information based upon categories with more ease, such as job descriptions, student personal info, and the Internship Survey ratings that fuel Fit Scores

  6. Fit Scores that provide visual data for how suitable a Student is for a Seat - Users can see quantitative scores on the 6 performance categories listed in the Internship Survey, optimizing their Matching efficiency with data-driven scores and a vital link between the information they collect and the tool they use

    • Green - high fit

    • Yellow - middle fit

    • Red - low fit

They see an open role and then click on it. The Make a Match panel on the far right then auto-populates the Seat Details section, where the user can view specific information on the open Apple role, including location, supervisor name, official job description, and length of internship contract.

The Student UI panel by default sorts Students by alphabetical order. Today, the user wants to find the best student overall for this Apple internship. So they then hover and select on the Fit Score Overall category. The Student entries are now re-organized according to who has the best scores overall for the Apple internship.

After they scan over the available students for the Apple role, the user decides that Daniel Rivera is a prime candidate due to his high Overall Fit Score of 95%. They then click on his card, which further populates the Make a Match panel with his background information under Student Details (if the user decides they want to check his credentials) and opens the Matching functionality.

The user, satisfied with the data-driven Fit Score information, proceeds to Match Daniel to the Apple internship. They then hover to the Make a Match panel and click on the Match functionality. The status of the internship has now changed from unmatched to Matched status. They are now complete with Matching this particular internship and can continue on to the next open role.

In four simple steps, the user is able to complete the all-important task of Matching without having to navigate needlessly in and out of multiple pages. With our prototype live, I moved on to the next stage and prepared my usability research.

Testing the usability of the Predictive Matching prototype

I scheduled and conducted these tests virtually with our 5 participants, needing only their laptop or computer, the Zoom application, and the Salesforce prototype. On my end, I used the Userlytics platform to record, annotate, and proctor these sessions.

Synthesizing the data from the usability tests

The results of the usability test generally showed high favorability among our users.

Users indicated a high level of approval of the new Predictive Matching Tool in perceived ease of use, satisfaction, value, and confidence. I observed from these sessions that though there were cosmetic and minor critiques to improve the UI further, there were no critical usability issues that were severe enough to merit a significant redesign.

I performed an affinity mapping exercise to further refine any usability testing insights. Critique was generally centered around the Fit Scores, the information display of Students and Seats, and the Matching Panel. I discovered the following:

With the right training, anyone could pick this up. I especially love the Fit Scores.
— Program Manager, Boston

The last question that participants were asked during testing was:

What are your overall thoughts on this new interface? What can be improved?

What this does so well is basically taking all the steps we need to do into one place. I don’t know why we didn’t do this sooner.
— Program Manager, Austin
The new layout translates info a lot easier. Adding those few tweaks would make this even better.
— Program Manager, NYC

Incorporating user feedback into a finalized prototype

While the prototype was well-received by users, there were still areas of improvement to tackle. I incorporated the following design changes to address user concerns:

  1. Changing the display of Seats profile - Replacing company logo with full internal ID number and length of contract terms

  2. A glance into how a Student is progressing - ing a tagline underneath Student profile indicating their academic progress

  3. Adjusting the Fit Score Sort mechanism - Allowing users the flexibility to sort by multiple Fit Score categories instead of just one

  4. Match Status indicators and an icon legend - Users commented that it was hard to tell which Students and Seats were Matched without some sort of visual cue. This design implementation helps them visually distinguish which Seats and Students are Matched

  5. Providing a “Possible” Match option - Though Program Managers aim to Match all qualified Students to an Internship, sometimes additional time or fact-checking is necessary before finalizing this decision. The “Possible” Match feature allows users the option to revisit a candidate by shortlisting them

Shipping to production & revisiting workflow process improvements

Overall from a user experience and design standpoint, the new Matching Tool was a resounding success. Staff confidence and enthusiasm for the design updates was at an all time high.

I presented my full research findings to the Agile team, our executive team, and AdeptID, showcasing the favorable results we had collected.

We secured approval from the executive team to push the new Matching Interface into production. I compiled my work into a handoff document for Ajay and his team to begin refining.

Collaborating with Tara and Walter, we set to work creating a potential future state workflow for our Program Managers.

Referencing the process flow and journey map I had created, I identified potential periods of time within the workflow where standardized working sessions could be introduced between themselves and other Matching teams.

I included in this updated workflow recommendations on the kinds of info that should be prepared for these sessions and the results staff could expect coming out of them.

This phase of brainstorming the new workflow was iterative and went through several phases of reviewing and editing. To the right is the latest version of this workflow, which was approved by the Agile team and is being prepared for sharing with the Program department.

Problem

With less Internships available to Match from, staff are unable to Match all qualified Students, creating less than ideal outcomes for both the business and Students

However, I emphasized to the internal team that our work was not done and that we needed to focus on delivering process improvements in order to fully realize the potential of the updated Matching Tool.

I connected with Tara and Walter on what they were able to gather in their interactions with the Program department.

Reimagining & introducing recommendations to improve the Matching business process

From my research, we knew the following was true about our staff’s challenges with the Matching business workflow:

  • Data sharing is not centralized and staff have varying access levels; different departments have access to different info

  • Lack of internship opportunities are a significant hindrance to effective Matching, as there are more Students than Internships available

  • Outside of meeting during Matching sessions, there is no standardized framework for cross-collaboration with other working teams

With the creation of an updated Matching Tool, the issue of information centralization was solved as all available data related to Matching was now contained in a single repository with universal access for Matching staff. However, the remaining problems of the lack of Internships and a standardized structure for cross-departmental collaboration still loomed ahead.

I additionally recommended the following to address the disruption that was caused by the “seat gap” phenomenon and to maximize the efficacy of the Tool among staff:

Recommendation

From my research, the vast majority of staff rely upon existing and repeat relationships to secure Internship opportunities. I recommended that staff should be trained in headhunting activities to source more companies and increase our available Internship pool.

Compared to before:

Reflections

Problem

Updates to the Matching Tool increases the need for a more standardized process. Matching staff are subject to their specific site’s framework, creating a more disorganized working environment.

Recommendation

Provide mandatory department training sessions to aid in change management, familiarize staff with the new Matching Interface, and designate open-office hours to answer any open questions.

Project Recap

The Predictive Matching project was a complex and high-visibility endeavor that lasted for one full year. At the end of it all, I oversaw a full end-to-end research and design process that delivered crucial design improvements to Year Up United’s primary CRM platform and produced recommendations aimed at addressing shortcomings in the Matching workflow.

Let’s revisit the main problems of what I was solving:

Design Problem

  • The Matching Tool is overly text-heavy, with unnecessary info and limited visual cues to interpret data

  • Users have to navigate through multiple pages to review Student qualifications and even perform the action of Matching

  • Filtering and categorization is non-existent, indicating a need for information organization capabilities

  • Internship Surveys submitted into Salesforce exist only as a read-only document, though users have expressed a desire to use its data to help find qualified candidates more efficiently

Matching Tool Changes

  • One-stop shop information architecture and layout limiting unnecessary opening of multiple tabs

  • Introduced filtering and sorting functionalities to limit information overload

  • Implemented AI-powered Fit Scores visual data derived from Internship Surveys

  • Clear categorization of Student, Internship, and Matching info based on type of content

  • Matching action easily accessible

  • Increased emphasis on visual cues

Business Workflow Problem

  • Data sharing is not centralized and staff have varying access levels; different departments have access to different info

  • Lack of internship opportunities are a significant hindrance to effective Matching, as there are more Students than Internships available

  • Outside of meeting during Matching sessions, there is no standardized framework for cross-collaboration with other working teams

Matching Workflow Improvements

Restructuring the way a company handles a core function takes time and resources. I paved the way for our executive team to introduce changes to the Matching process by recommending the following:

  • Creating an updated standardized workflow where regular touchpoints increase much-needed cross-collaboration, limiting team friction

  • Training staff in sourcing more partner leads instead of relying solely on existing client relationships to increase Internship pool size

  • Introducing mandatory trainings on the new standard workflow and the new Matching Tool to enhance change management

Metrics & results after launching the new Matching Interface

After providing Ajay and his dev team handoff materials to finalize the Matching prototype into Salesforce, we launched it live into Salesforce to pilot across multiple sites.

Similar to our previous survey study, we recruited 21 new Program Managers from New York, Texas, Massachusetts, and Arizona to use in their Matching work. We asked them to follow the same guidelines as our baseline study and submit their responses once a week over a 4-week period.

I collected the following results:

Evidenced from these studies, my updated design of the Matching Tool led to:

  • Increased user satisfaction from 55% to 86%

  • Increased perceived value of the platform from 40% to 91%

  • Reduced the total amount of staff time spent in a Matching cycle by 16.3%

  • Reduced the total amount of staff time spent in Matching meetings by 6.3%

  • Increased the number of young adults we matched to an internship by 35.3%

  • Reduced time spent on task by 47%

As the Lead UX Designer for Year Up United’s Predictive Matching initiative, I spearheaded a comprehensive redesign of our Salesforce-based Matching Tool and introduced business process improvements to enhance efficiency and scalability in connecting underserved young adults with corporate internships.

The project itself was incredibly challenging that required management of many moving parts and active collaboration with multiple stakeholders and team members. Although this was not my first time directly leading UX initiatives, given the long duration of this project, there were moments in time where the team’s immediate needs came into conflict with maintaining UX best practices.

The most important takeaways I gained from being a part of this impactful project were learning the art of when to listen, when to push back, and cross-functional engagement.

One example comes to mind where early on in our project, our team was eager to push the Matching Tool wireframes I had created into production and complete usability testing from there. I knew inherently that there was pressure from the higher echelons to deliver results quickly, but I also knew that taking this approach would inevitably result in broken results and more resources spent. I understood that individually, particularly from my Product Manager and Business Services Manager, that each had their own concerns that I had to assuage.

I listened to their perspective and created a working culture of high visibility and communication, where I would regularly invite team members to demos and working sessions of my progress. I reaffirmed throughout the importance of thorough research prior to releasing designs, and the work I produced in gathering important insights eased them into a proper research into design workflow. As more results came in, my team developed high confidence and trust in my process, and ended up receiving a deliverable that was much more impactful than if they had rushed through it.

Many of the challenges I had encountered were mitigated through my emphasis on communication. If nothing else, my success as the Lead UX Designer here would be attributed to the fact that I had built strong trust with my colleagues, where I not only provided my support and guidance, but was readily available to ask for the same from my teammates. The success of this project is based on the fact that I was able to develop a productive working rapport with my Agile members, as well as showcase the gains made from following an iterative UX process.

Trust is key and can make or break any environment.

Constructing a working prototype on Salesforce Lightning App Builder

After these discussions, I produced a higher fidelity wireframe utilizing the Salesforce Lightning Design System to better illustrate interactive elements and actions, as well as visually represent Fit Scores. I worked alongside AdeptID and Ajay on a consistent basis by setting weekly check-ins to ensure that we were aligned on expectations and technical requirements. I utilized the Salesforce Lightning Design system to mirror components, and in addition to the overhauled IA, I implemented the following design implementations to address user needs:

A majority of users desired to see some sort of performance metric at a glance for how well a Student “fit” for an Internship. Our Program Managers submit a document known as an Internship Survey where they record 6 quantitative ratings on their students academic and professional performance. These ratings help our Matching staff make decisions on which student is best suited for a particular internship.

Although a pivotal document for Matching, it existed at that time purely for record-keeping purposes and had to be frequently downloaded or printed on hand to reference.

An idea then dawned on me that an ideal way to use AdeptID’s AI algorithms while increasing the Survey’s use case for our staff was to integrate these performing ratings into Salesforce and translate them into visual data points for our staff in the Matching tool.

Incorporating Fit Scores into a higher fidelity wireframe

Previous
Previous

WithYou

Next
Next

Goettl