Context: Master’s Capstone Project
Domain: Financial Services
Timeline: 6 months
Role: Project Management; UX Research
Team: Won-Woo Chung, Annette Hong, Hyun Woo Paik, Se One Park
Overview
PNC Bank’s Corporate and Institutional Banking line of business tasked our MHCI team with designing interventions to their online banking tool PINACLE that would support their clients in performing essential tasks and troubleshooting problems.
The new Quick Help makes help documentation more accessible, and the Task manager eliminates page-switching to execute tasks. Additional supports surface in the form of correctional tool-tips which flag potential errors, and auto-generated tasks (“shortcuts”). The purpose of these tools is to expedite learning and task completion via safety rails and dynamic support.
Solution
Our solution to reduce the number of PINACLE-related calls made to Treasury Management Client Care is a digital service innovation that supports better user learning of the banking platform from the moment they log on. The New Help is designed as a value-added modification to PINACLE Commercial and PINACLE Express that complements the existing workflows in both tools.
Our proposed Adaptive Initiative consists of two main components: Quick Help and Task Manager.
Seamless Support with Quick Help
In our design, PINACLE’s Quick Help is available on every screen of the platform to provide the user with contextually relevant support. Users can click through cards for more information, or type into the search bar below to find what they’re looking for. With Quick Help, context-aware support is available on-screen, where it doesn’t get in the way of workflow. It’s easy for users to access training information and guided walkthroughs when they’re made prominent and immediately relevant.
Novice users will be prompted to engage with Quick Help, training them to look toward the lower-right corner of the screen for help and support. As users gain more experience with the tool, Quick Help becomes less prominent on the screen, but is always available if needed. Quick Help offers users easy, relevant information to complement the trial-and-error method of learning that users tend toward, better supporting them in their process of mastering PINACLE and its capabilities.
Correctional tool tips provide feedback to the user based on their past behaviors, like typical payment range for a payee, and payment processing calendars in order to prevent errors.
Task-First Flow with Task Manager
Users can create, manage, and complete tasks in the task manager on the PINACLE landing page. Over time, the users’ interactions will create a data set that can be used to train a recommender algorithm, which will be able to suggest recurring tasks to the user.
Exploring the Problem Space
PNC C&IB’s Treasury Management Client Care (TMCC) representatives provide support to clients with inquiries on how to use services within the portal as well as general inquiries related to their account activity, amounting to roughly 7,000 client calls per month. PNC tasked our team with application of human-centered design methods to identify opportunities for improvement in a redesigned PINACLE user experience.
Our initial research focused on a seemingly simple goal: how might we improve the online banking experience for commercial banking customers?
Methods
Interviews
We conducted interviews with 10+ financial professionals who use PINACLE about their online banking experiences.
We also interviewed TMCC representatives about their day-to-day experiences handling client calls about PINACLE. This helped us enrich our understanding of the quantitative call long data.
Data
We analyzed and synthesized call log data from PNC’s Treasury Management Client Care (TMCC), identifying the most called-about issues
Tool Audit
We conducted a comprehensive audit of all tools available to PINACLE users portal, as of a demo of the incoming redesign.
Ecosystem Mapping
Financial professionals work with a variety of software in their day-to-day work, and we explored out how these tools interface with each other.
Generative Insights
Through our generative research, we learned the following:
Insight 1: PINACLE users are a diverse group, as are their needs and behaviors, and the tool should support users of all abilities.
Interviewed financial professionals varied in terms of responsibilities, department structures, tech savviness, and attitudes toward security.
Length of call “depends on how competent the client is.” Some clients can follow directions well, but others need click-by-click instructions
Insight 2: PINACLE users are overwhelmed by the number of features, leading to frustration with their corporate banking services.
Most of the interviewed PINACLE users make very limited use of the tool, and feel intimidated by features that go unused. They do not make time for exploration of new features or help documentation.
“Sales representatives sell small companies products more complicated than they need.”
– TMCC Representative
Insight 3: How clients learn to use PINACLE doesn’t match clients’ expectations, and clients tend toward reactivity.
PINACLE users don’t know about the Help and Training resources that are currently available on the platform, or they don’t want to use them because of how unwieldy they perceive them.
“I think that most of the time, client issues could be resolved through the [PINACLE] Help and Training module.”
– TMCC Representative
Insight 4: Time is relative for both clients and TMCC representatives.
Some banking clients perceive calling TMCC to solve a problem as faster than troubleshooting it themselves; others devise elaborate workarounds to get PINACLE to perform the way they need it too. And TMCC reps often spend long chunks of time on the phone guiding clients click-by-click through problem resolutions.
“If I’m sitting in front of a video screen, I’m going to be guilty of doing something else at the same time. If I’m sitting at my desk where there’s a whole lot of work that needs to be done, I can’t give you my undivided attention.”
– Accounting Manager
Synthesis
Over the course of our research process, we synthesized our findings and insights into resources that we referenced throughout our ideation and design.
Concept Map
In order to better understand the system our challenge is grounded within, we built a concept map as a synthesis of the information we collected during calls with our clients.
This synthesis helped us identify the parts of the system most critical for design interventions: the business customers with annual revenues of $5 – $50 million, the associated non-mobile PINACLE products they had access to, and the TMCC call center.
Personas
We built personas based on the data that we gathered from both our service providers and our users. We referenced these personas throughout our process to help us understand our stakeholders’ needs, experiences, and goals. In our iterative design work process, we have revisited these and revised them in response to feedback and new findings in order for us to best understand who we are designing for.
Journey Maps
We mapped the various pain points that PINACLE clients and TMCC representatives experience through journey maps in order to form an aligned mental model of the problem space, as well as to communicate our understanding to our clients and advisors
This process helped us identify a demonstrable gap in our own data: comparing interviews with PINACLE users to the qualitative and quantitative data from TMCC revealed a mismatch; most of our PINACLE user interviewees reported that they do not call TMCC often, if ever, despite the massive call volume from users in their peer group. The customer stakeholders that we have been able to access report fairly high levels of comfort in using technology; their experiences may not entirely reflect the realities of PINACLE users who are frequent callers. This is a gap we hope to close as we move forward in evaluative research of our prototypes.
As with our personas, we revisited our journey maps as our findings and focus evolved in order to revise them to match our most current understanding of our stakeholders’ experiences.
Affinity Diagram
We built an affinity diagram to organize our data points from interviews with financial professionals in able to draw out insights. We also augmented the diagram by adding in data points from our interviews with TMCC representatives in a different color to corresponding groups we already formed. It was interesting to see that some of the insights from PINACLE users were validated by client care team, while others were challenged.
Ideation Overview
After we completed our generative research, we moved on to putting our findings to use through ideation around improved PINACLE experience and validation of our users’ needs and of our concepts.
Need Validation
From our insights, we formed four perceived user needs that we sought to validate through speed-dating of representative storyboards:
- Users need to access accounts in the face of obstacles
- Users need to understand system status in PINACLE
- Users need to connect with relevant support from TMCC
- Users need to know how to perform critical tasks
Our needs validation process revealed that visibility of system status was not as big of a need as our generative research had previously indicated, and that while account access is an ongoing area of friction, that friction is not necessarily a bad thing. The most valuable findings we gained were around the need of PINACLE users to understand how to perform and troubleshoot in the platform.
Key Insights (Evaluative)
Through this evaluative research, we were able to form new insights:
Insight 5: Users want to be able to troubleshoot and recover from their own errors.
Errors within current user flows often result in the need for TMCC support, with limited opportunity for non-admins to course correct.
Insight 6: Users want to learn by example, through demonstration and at their own pace.
Step-by-step instructions are most helpful to PINACLE users when the instructions can work along with them.
Insight 7: Lack of confidence is a hurdle to correctly executing tasks.
Some experience anxiety over the possibility of mistakes when performing high-stakes tasks.
Insight 8: Users want to perform their tasks quickly and efficiently, but need to be properly equipped to do so.
Users perform a calculation between what will resolve their issues most quickly – and sometimes, it’s calling TMCC.
Upon reflection, we realized that all these insights nested well underneath Insight 4 (the mismatch between clients’ expectations of how to learn to use the tool and the reality). The more specific learning from Insights 5-8 would help us to narrow our proposed solutions to our client.
Concept Generation
We did a variety of ideation exercises as a team in order to explore potential directions for our opportunity space based on our findings. From 20 Questions to Crazy Eights, reversed assumptions to reframing our problem within analogous domains, we used multiple ideation frameworks and tools to generate a wide spectrum of ideas. These exercises also helped us to develop artifacts in the form of storyboards that we shared with stakeholders in speed-dating to gather feedback on concepts.
From our concepts, we built storyboards to allow ourselves to develop those early ideas further and gather stakeholder feedback.
Midpoint Check-In
At the midpoint of our engagement, we wanted to offer an opportunity to our clients at PNC to react to our design direction. In order to best serve our clients, our design intervention addressed the insights from our research with the overall objective of reducing the total number of phone calls made by clients to PNC’s treasury management client care team.
From our analysis of the call data, we had a clear goal in mind: to better support users to perform accounting tasks in PINACLE on their own by teaching them how to use the tool more effectively.
Two approaches
Through our ideation and concept validation process, we realized that we were considering both heuristic changes and flow overhauls. Therefore, in our presentation, we presented our selections in terms of micro and macro design interventions.
On the micro level, we could augment the PINACLE user experience through heuristics and tweaking features. We posit that small changes could make a big difference in supporting accountants’ use of the portal.
On the macro level, we introduce novel omnichannel experiences to disrupt current user flow with new features. These interventions present opportunities for proactively equipping users with tools to give them more autonomy and confidence.
Micro Design Approach
Our micro design concepts were not groundbreaking; rather, they drew from established design patterns, like wizards and undo buttons, that we thought could lead to increased user autonomy and understanding of the tools within PINACLE. Designing for these features to be included in the platform, we could enhance users’ experience without changing the core flows of PINACLE Express or PINACLE Commercial.
Macro Design Approach
From a macro perspective, we envision disruptions to the user flow in PINACLE through our proposed, novel, omnichannel experiences. In this case, these interventions, including hybrid human are not limited to specific tasks that might be performed in PINACLE; rather, the macro level design interventions respond to the need for flexibility and confidence-building that our user research indicates.
Faceted NLP search
PINACLE doesn’t have a search bar, which is a common feature in some retail banking products. We propose that adding a search function that utilizes Natural Language Processing to parse help documentation will lead to greater user autonomy and understanding.
For example, if a user wants to export a template for ACH transfer, NLP-powered search results could show multiple options to choose from. If a user wants to learn how to export a template, the result would guide a user to relevant instructions. The result could take a user directly to the page so that he or she could instantly perform a task, or it might show a specific snippet of rearchitected information to collapse the searching process.
Practice Mode
Practice Mode enables users to play with the platform, clicking around and testing out features without any risk of causing damage to the company. In practice mode, the displayed information would be identical to real data to give more realistic experience, but the transactions will not be. The user could choose a specific task to try out and get feedback on from the system. Through practice, users get more used to the features and become confident in executing tasks in PINACLE.
Presenting our Work
We delivered our presentation to an audience of our clients, our faculty, and our peers. Because the majority of the people in the room did not come from an accounting background, we spent time contextualizing the corporate banking space, as well as introducing our personas and other design artifacts that we referenced throughout our process. Our presentation touched on our research process and ideation work and introduced our design directions and possibilities to our audience, walking through macro and micro design approach possibilities.
Our presentation was met with a high level of engagement from our clients. After our pitch, they joined us for an ideation session in which we reflected on our work thus far and looked forward into the summer. While the possibilities that we presented in our slide deck were not off base, together we worked with our client to build out a better model of what we’re actually trying to do: effectively surface data to PINACLE users in order to ease their burden of use and more effectively perform their jobs, reducing the number of calls made to PNC’S treasury management client care.
User Testing
We built and tested multiple, divergent designs before reaching our final concept.
Paper Prototypes
For the first round of testing of our PINACLE support UX, our concept was still in flux and we A/B tested four different paper prototypes with PINACLE users to learn more about how a design intervention might enable or disrupt their flow.
We choose to do paper prototyping for two reasons. First of all, the flexibility of paper allowed our designs to take a lower fidelity form without it distracting users, as nothing was truly clickable. Secondly, in our Wizard of Oz-ed experience, we made use of the multi-modality of paper by inquiring about the allocation and flexibility of space on the screen of our different design interventions. Paper proved to be much more flexible with testing the dragging elements than its digital counterparts.
We built several divergent interface prototypes, including a chatbot for issue resolution; an NLP-driven global search bar; a module that slid onto the screen to provide help; and a drag-and-drop wizard. We also asked each user to co-design a chat or search interface in order to better understand how these tools might be most useful.
From our user tests on paper, we learned a lot about how user workflows might be augmented or disrupted through our design proposals.
Insights:
- Trial and error is not only the default way of learning how to use a new tool; it’s also preferred.
- Users frame their problems in terms of actions, using tasks and verbs to express discrete challenges.
- Use of screen real estate should enable flow through flexibility and modularity.
Clickable Prototypes
In our next iterations, we designed in response to what we learned from our paper-based user tests. We built prototypes in InVision that followed linear user flows and proceeded to test them with users both in person and online through UserTesting.com. Each prototype had a clear machine learning component to them: parsing out text to auto-fill forms to help initiate payments, providing suggestions, and providing help documentation based on user (in)activity.
Insights:
- Some friction is good, but too much treads on inefficiency.
- Contextual, step-by-step information alongside the relevant actions is most valuable to users.
- Users are reticent to make use of search functions because their own expectations are not well defined.
From this testing, we were able to eliminate several features from our possible user experience. Within our target audience, text-entry-based interfaces are seen as time-intensive and less likely to be integrated into users’ workflows. That is not to say that they (in particular, chat) are not worthwhile endeavors for our client; rather, they are not solving the problem we are focusing on, the improved user learning experience of PINACLE.
Next steps
The insights we synthesized from our early rounds of user testing helped us to focus on what users want, and we proceeded to build out more robust interaction flows that integrated desirable and impactful features, continuing to test and iterate until we produced our final design concept.
Solution Rationale
Upon review of the insights we extracted from our discovery research, need validation, and evaluation of early concepts, we decided that our final solution must do three things: augment user flow, support users’ tendency toward trial and error, and optimize client care efforts.
Augment User Flow
Collapsing navigation
One of our main goals in tackling the challenge of user support on PINACLE was to collapse navigation in accessing help. In the current state of PINACLE Express, users must drill down through several categories of information architecture to access PDFs that provide support documentation about the tool. Through our research, we found that users perceived this as an investment of time that takes longer than calling a customer service representative. Therefore, we wanted to reduce the amount of time it takes to find context-relevant support documentation in order to change the balance of time in the user’s troubleshooting calculation.
“I don’t really use PINACLE for any of this information [ACH]. I go straight to Small Business. It doesn’t really give me any useful information, and it’s just too many steps.”
– Controller
Task-oriented workflow
Throughout our user testing process, we found that PINACLE users frame challenges they face in terms of tasks and verbs, which runs counter to the noun-centric modules and information architecture of PINACLE Express.
“By default, just from being on that page, the task tiles are reminders. But when you click on them, they should also act as shortcuts to take you to the area where you can complete that banking transaction.”
– Controller
Better Trial and Error
Scaffolded Learning
PINACLE users told us time and again that they like to learn by doing; these users prefer to jump right in rather than sitting through a webinar or reading through a guide. They want help that supports them when they don’t know what comes next. The New Help uses “scaffold and fade” learning supports to provide support to new users when they need it and to be available (but less prominent) at all times when they don’t.
“I would definitely use something like Quick Start for training, having my team, people that aren’t really accustomed to accounting, get into this and be able to set up new receivers and wire payments and things like that.”
– Accounting Manager
Supported Autonomy
Throughout our generative and evaluative research, we found that PINACLE users want to feel like they have control over their workflows. Our prototypes that explored predictive functionalities with regard to user actions elicited strong reactions from our testers, most notably from fear of making mistakes. At the same time, users were eager to try out our prototyped task manager, driven by a predictive algorithm that learns from user behavior; they thought it would be a great way to efficiently accomplish tasks. We designed the New Help to complement users’ need for control in the high-stakes realm of corporate accounting with augmentation of user flow. All predictive elements can be controlled by the user and do not automate the completion of tasks; instead, they ease use of the PINACLE suite by understanding and supporting the user’s independence.
“I’ve had predictive options before with our accounting software, and that was how things really quickly and easily messed up… For me, I’d rather be completely aware and in control of what’s going on, especially when it comes to accounting.”
– Accounting Manager
Optimization of Client Care
Offloading Training
Our research showed that the current PINACLE on-boarding and training experience is not meeting the needs of users. Cursory onboardings by sales associates provide only an orientation to the product. After that, clients who are using PINACLE are meant to learn how to use the portal through webinars and tip card PDFS, but they often wind up directing general help and training questions to TMCC representatives (even though they’re not meant to be product educators). Our solution helps offload the burden of client training from TMCC onto the interface. Quick Help appears on every screen to provide context-aware support, but it also scaffolds new users at the beginning and fades as they use the product over time, effectively teaching the client how to use the portal. This will free up TMCC representatives’ time spent on training calls and will decrease the number of overall inquiries.
“There is no on-boarding in PINACLE itself, but the associates on the sales team are supposed to do some on-boarding and welcoming. ”
– TMCC Representative
User Confidence
One of the reasons that clients and TMCC representatives cite for calls to TMCC is lack of confidence on the part of the client. Accounting is a high-stakes domain, and users’ uncertainty of the outcomes of their actions contributes to chronic calls to client care. Our Adaptive Initiative solution supports clients by providing them with access to the information that they need to feel equipped and competent. As users’ confidence in their ability to use PINACLE increases, so too will their positive feelings about banking with PNC.
Next Steps
Our final design solution necessitates and lays a strong foundation for a roadmap toward a more all-in-one PINACLE experience, where nearly every related action taken by the user can be done within the portal itself. Each phase of this road map involves a different level of user versus system input.
Initially, users will create and complete their own tasks in the Task Manager. These interactions will constitute a training data set that will be used to train a recommender algorithm. Once trained, which will be able to suggest recurring tasks to users, personalized according to their regular behaviors.
Taken together, the Task Manager and Quick Help lay a strong foundation for movement toward an all-inclusive PINACLE experience.
Nathalie Rayter 