ADP Policy Miner

OVERVIEW

For this project, I worked on a web application for ADP's Policy Miner, an automated payroll onboarding experience for ADP and its midsized business clients to transfer from its old payroll system to the new system, Payroll Innovation engine (PI). Policy Miner, which is currently in progress before implementation, strives to empower the involved users by integrating machine learning to provide the payroll practitioner (an ADP client) and implementer (an ADP associate) with a quick and efficient onboarding process.

I partnered with two other UX designers to explore the concept and visually design Phase I of the UI for Policy Miner. Once further improvements are made to Policy Miner, PI will launch as the primary online application for ADP's midsized business clients. 

My Role: UX Designer, UX Researcher
Duration: 2.5 weeks
Tools: Axure, Figma, Sketch, Invision

UNDERSTANDING THE NEED

When our stakeholders (a PM and UX designer at ADP) came to us for help with designing the UI for Policy Miner, my team and I first asked why they feel that Policy Miner is needed so that we could get a better understanding of what problems Policy Miner would be solving and if Policy Miner would truly be the best solution. 

We found out that the current onboarding process at ADP takes six to ten weeks and requires twenty ADP representatives to participate in a training session. Even if one disregards the unreasonable turnover time, the mandatory training session itself is costly and not effective, for many employees have trouble paying attention and do not fully understand what is expected of them to perform the data transfer and effectively onboard clients to the new system. 

THE CHALLENGE

• How to revamp the user's experience and simplify application tasks based on a complex system? How to take onboarding offline to online?

• How to change the stereotype of traditional corporate training and engage the users? How to build trust and convince them the value of learning and encourage them to become more self-sufficient?

• How to provide for collaboration between User 1 (payroll practitioner) and User 2 (ADP implementer) without overloading the ADP implementer into having a customer service role?

• How to prevent system and user errors as much as possible

As you can see, the problem presented itself as a much more complex issue than simply a need for an interface to the new system - much research, testing, and iterations would be needed to 1) predict what the machine learning concept would look like and 2) make Policy Miner a seamless experience for users so that they will not only learn how to use the new system but also prefer it over the previous experience. 

TACKLING THE PROBLEM

After much brainstorming of the problem we can solve for Phase I, my team and I determined the problem to be:
ADP needs an automated, efficient, and reliable onboarding process for both new midsized clients and migrating existing clients because transferring from the old system to the new system is time consuming, costly for ADP/clients, difficult to scale, and prone to error. The current process takes six to ten weeks and requires twenty ADP representatives. Also, consistent language* is needed across policies for proper categorization. Lastly, implementers are unable to manage the overwhelming requests of payroll practitioners through a customer service role.

*Note: When transferring from the old system to the new system, certain "policy" names do not overlap across companies (e.g. "401K" and "Retirement Plan" may mean the same thing, but the system does not recognize that) - this is where machine learning, or Policy Miner, comes in to "mine" the data for the user.

Our solution for this complex problem is: 
By creating an interface for the new ADP payroll engine, PI, we will be able to cut down the onboarding process by half, while preventing system errors before they happen.

THE PROCESS

Our stakeholders provided some assets and some previous research on our target user, the "implementer", but after assessing the assets, we decided that we needed to do more research to confirm the assumptions that had been made. 

We sent out a survey to 11 people and asked the following questions, among others:

Key takeaways:

• Many respondents use ADP, but many of the issues they have also involve the ADP platform

• People complain about having too many manual entry fields

• Employees don't do their part/delay the application

• The amount of time is an issue - especially because payroll practitioners have other duties

• People would like to learn to use a new platform using virtual learning, training tutorials, written guides.

This is how we formed our persona: an ADP implementer. 

Using this information, we decided to conduct Competitive and Comparative Analysis using some of the companies that our respondents use to process payroll.

As our surveyers expressed, upon analyzing these websites and accessing demos of the products ourselves, we found out that platforms like Gusto and Turbotax are popular for good reason: they use a conversational tone and show a sense of progress to the user, keeping transparency and decreasing cognitive load. 

ADP has been a legacy leader in the payroll/HR industry, but we saw now why other platforms were becoming more popular, and we wanted to make that clear to our client. Like the new companies, ADP has great potential in automating such a process. 

LET'S GET TO THE FEATURES

My team and I decided to do a Goals Exercise activity to determine how users would be discovering, learning, using, and achieving their goals through our product.

We decided to keep it as simple as possible, to ease the user into the newly automated process. Uploading files may be easy to get users onboard with - it's straightforward and less work for the user. However, how would we display this information in a way that payroll practitioners are used to seeing? The answer: spreadsheets.

Since payroll practitioners are familiar with using Excel spreadsheets to enter and view their data, we decided to explore a spreadsheet feature that could be easily editable when the user sees or is notified of an error. We believe that this feature would allow for the user to build trust in the system and complete the application with ease. 

Also, since many of our respondents expressed that they would be interested in virtually learning how to use a new platform, we decided to integrate a virtual workshop once the user first begins using the platform. Initially, our idea for the virtual workshop was to be a process in which the whole team is gathered in one place to take the workshop, but through user testing and more interviews, we decided to iterate it into an optional video tutorial that would explain the steps for Policy Miner as well as an explanation for what Policy Miner does for the user. 

We decided to get our ideas on paper and make low-fidelity sketches of the application and its features.

Upon receiving feedback, we decided that Policy Miner would receive a user's data via uploading .xls or .pdf files, then "mine" the data which takes approx. 30 seconds, and lastly, generate potential "matches" for verification - the matches would address any variables that count as policies, whether they are policy plans or labels. The user could then correct the addressed errors or enter information about a specific policy so that Policy Miner would autofill suggestions. As a last resort, a Live Chat assistance would be available for the user if they cannot resolve the issue on their own. 

USER TESTING

We conducted a series of usability tests, both to our survey respondents as well as the payroll implementation team at ADP. 

Key insights from usability testing:

• I like the conversational language - it helps me understand the system better and makes the process less boring

• Spreadsheet feature is a plus! I'm used to seeing this, so I can correct errors easily

• Can I know how much time a specific step/process will take?

• What does inspect mean in the application?

• How will I know if the suggested match is truly a match for my policy?

Taking these insights into consideration, we iterated our layout and designed high-fidelity wireframes. 

NEXT STEPS/FUTURE RECOMMENDATIONS

1. ADP Implementer Workshop Training

2. Develop a roll-out to display to legacy clients

3. Work with marketing on introducing Policy Miner to the market

4. Develop easier ways to upload registry

5. Further develop Policy Miner's collaboration feature

6. Explore rolling out the Policy Miner onboarding process to small businesses

FINAL PROTOYPE

You can view a clickable version of the final prototype here: 
https://invis.io/KPJ9OSIZ7WJ#/298718174_Homescreen