Automated payroll (ADP)
Automated payroll (ADP)
Embedded Files
Research goal: evaluate if and how automation can help solve pain points in payroll review process
Research goal: evaluate if and how automation can help solve pain points in payroll review process
Current payroll dashboard (test client - dummy data)
Problem statement: Reviewing payroll for errors is a time-consuming and tedious process. There are no features in the product supporting payroll admins with this task.
Assumption:
Automating parts of the review process will help payroll admins save time and make the process less cumbersome.
Research questions:
How do practitioners currently review and resolve errors?
In which cases is assistance with reviewing errors most necessary?
How can automation be a helpful tool in the review process?
Methodology: Multi-phased, mixed method research
Methodology: Multi-phased, mixed method research
My responsibilities:
Project management
Project management
Manage project timeline and scope
Triage research questions
Determine research strategy
Collaborate with external agency and cross-functional team of developers, designers, content writers, and product managers
Conduct research
Conduct research
Create screener and recruit participants
Create survey in Qualtrics
Create discussion guides
Moderate focus groups and 1:1 interviews
Create Pendo dashboard for product analytics
Design, launch, and monitor A/B test
Analyze and synthesize data from multiple sources
Influencing
Influencing
Present biweekly updates to steering committee
Present insights and recommendations to C-suite executives
Facilitate working sessions with core team to ensure research insights are implemented in design iterations
Project challenge: Maintaining research rigor under pressure
Project challenge: Maintaining research rigor under pressure
High visibility and high pressure from senior leadership necessitated creative approach to meet demands while maintaining research rigor:
Redirected opinion-based conversations through research data and advocacy for user needs
Pushed for foundational research while team had jumped to immediate solutioning.
Delivered high-level insights on-the-go while building the usual comprehensive report
Wins
Wins
Shared knowledge among team increased tremendously
Team made major strategic changes based on research feedback, resulting in higher engagement with payroll assistance features
Project challenge: No process yet for A/B testing
Project challenge: No process yet for A/B testing
In-product A/B testing had not been done before at ADP and I recognized that this project was the perfect opportunity to push this method.
To achieve this, I organized and facilitated multiple rounds of strategic discussions with UX leadership, operations team, product governance leadership, and client support team to create buy-in for A/B testing.
Wins
Wins
Launched the first ever in-product A/B at ADP's Workforce Now product, paving the way for future experimental studies, and ultimately maturing the research practice at ADP
Multiple teams soon expressed interest to test incremental label and UI changes
Key findings
Key findings
Detailed research insights and recommendations of this research are strictly confidential.
On a high level, I gathered invaluable insights to answer the original research questions:
How do practitioners currently review and resolve errors?
How do practitioners currently review and resolve errors?
Mapped Jobs To Be Done and general payroll review process
Identified most cumbersome parts of the process
Gathered baseline for time on task
Identified decision makers besides payroll admin
In which cases is assistance with reviewing errors most necessary?
In which cases is assistance with reviewing errors most necessary?
Quantified payroll error frequency and seasonality
Discovered range of acceptable payroll fluctuation
Pinpointed level of fluctuation at which admin will further investigate the data
Specified the steps of further investigation and resolution
How can automation be a helpful tool in the review process?
How can automation be a helpful tool in the review process?
Measured sentiments towards automated assistance
Tested different concepts, some more automatic than others
Reasons why sentiments are positive or negative for different concepts
Where and how (automated) assistance is the most helpful in the product
Engagement rate for new payroll assistance features
Reasons why engagement rate differs for certain features and certain errors
Which label drives highest engagement
Result: Payroll error assistance features are used by 68% of clients
Result: Payroll error assistance features are used by 68% of clients
Payroll error assistance features are currently live and used by 68% of clients each pay period
Based on customer satisfaction scores, the features get positive ratings and are perceived as helpful and are reported to save 20 minutes on average
More assistance features will be developed and shipped over the course of 2025 in different areas in which payroll admins review earnings and hours data
Pipeline of A/B tests
Page updated
Google Sites
Report abuse