Agent Connect Strategy

Continue efforts to move away from being perceived as a Survey, Feedback and Quality Assurance tool towards being seen as a holistic, enterprise Workforce Engagement Management solution for the frontline.

Context

Upon acquisition by Medallia, Agent Connect offered a basic 1:1 coaching functionality. Team leaders could create session drafts, add notes, metric snapshots, and action items.

Over a period of 2 years, Product Managers and I gathered feedback, ideas, and challenges from coaches and organizational leaders during user research sessions.

Contact Center leaders understood the impact that consistent, meaningful coaching had on reducing agent attrition. However, they struggled to determine if their managers were consistently great coaches, especially as macroeconomic conditions forced them to reduce management headcount, requiring managers to lead larger teams with the same amount of time available for coaching tasks.

Team leaders desired a way to automate and expedite the preparation for 1:1 sessions, specifically coaching with context: surfacing QA reviews, comments, and customer feedback that indicated an agent needed improvement in specific areas, such as communication or empathy.

Frontline employees expressed interest in flagging feedback themselves, sharing their notes with managers before 1:1 sessions, and providing bilateral feedback—a way to "rate" their coaches.

Data Insight

Agents who feel their coaching is lackluster are 80% more likely to leave the company within 12 months.

Source: Medallia's Internal Research

Explorations

Collected insights allowed us to explore multiple approaches to improve coaching. Prototypes were used to solicit additional feedback from our customers and eventually decide on the most impactful effort. The tested approaches included:

  • Coach Dashboard: This explored ideas for recurring 1:1 sessions, highlighting team members most in need of coaching (e.g., those with sudden performance drops), and reporting on the number of sessions, metrics, and kudos shared with team members.
  • Frontline Dashboard: This explored ideas for bilateral feedback, the ability to share notes for upcoming sessions, and a centralized action items and lessons center.
  • Coaching Wizard: This explored the idea of a step-by-step guide where an AI model surfaces the most impactful pieces of context for a coaching session, such as metrics, customer feedback, and QA reviews.
  • Suggested Topics: This explored the idea of utilizing an AI model to suggest coaching notes based on past sessions, changes in metrics, comments in QA reviews, and feedback.
The Coach Dashboard exploration focused on identifying members in need of coaching and provided reporting and guidance for coaches.
Dark mode was a recurring request from our customers. We explored the viability of this effort when moving Coaching to the Alchemy Design System.
The Employee Dashboard explored a centralized spot for action items and lessons, with the ability to view upcoming sessions and provide bilateral feedback.
Coaching Suggestions explored summarizing previous sessions and listing potential topics to cover in the current session. We felt this had the most potential impact for coaching.
Carousel exploration provided a compact way to present a single recommendation to use or discard within a coaching note.
Stack is a visual variation of the carousel but suffers from the same issue of showing just a single topic.
The Vertical List resonated with both PMs and customers, as it allowed viewing multiple suggestions and cherry-picking the most useful ones. Unfortunately, it pushed session content down.
The Module moved to the right panel and established a new rule: the large, white container is the content of the actual session, while the widgets on the right are helpers and suggestions. Explorations in the Alchemy Design System helped convince stakeholders to push for an entire page overhaul.
The Slideout anel combines the pros of a vertical list without mixing with the content of the actual session.
The Coaching Wizard explored a step-by-step, full-screen experience for session preparation. It resonated with customers, but we considered it a nice-to-have after launching an MVP.

Validation & Prioritization

Together with the product and engineering teams, we iterated over multiple AI prompts in an attempt to summarize previous coaching sessions and provide coaching suggestions. This trial-and-error method validated the technical feasibility of such a solution with satisfying results.

Through validation of the prototypes with customers, we decided that Suggested Topics would be the most impactful effort for Agent Connect.

Initially, the goal was to keep the 1:1 session in the existing design or at least decouple it from moving to the Alchemy Design System to avoid scope creep. However, rounds of explorations in the new design system helped convince leadership that launching this powerful feature deserved elevating the design and fixing numerous usability issues the page struggled with. The marketing team voiced their support, as it would help make a big announcement much easier.

Product Goals

  • Introduce an AI Coaching Assistant, aligning with the perspective regarding AI/ML at Medallia—AI cannot completely replace a human in mission-critical programs that drive customer and employee experiences. Instead, AI should assist, activate, and empower employees to be more efficient and effective in completing their jobs.
  • Reduce the amount of time coaches spend preparing their coaching agenda/notes for each coaching session.
  • Offer a summary of previous coaching sessions between a coach and a coachee, surface trends and insights, and suggest potential coaching topics to avoid a blank-page effect.

Design Goals

  • Create a modern, engaging, and empowering experience for both coach and coachee.
  • Resolve the draft and sharing flow to reduce confusion and make coaching more effective as a feature set.
  • Reduce anxiety connected to sharing sessions by coaches by clearly communicating which sections are shared with employees and which stay private.
  • Resolve action items within the coaching UI to ensure new, open, and closed items are easily found.
  • Limit the number of pages a coach needs to use simultaneously to prepare for a 1:1 session.
  • Improve the workflow for including files and attachments from sources outside of Agent Connect.
  • Improve the experience of the actual session between coach and coachee, especially when screen sharing is used to discuss the 1:1 agenda.
  • Offer ways to increase adoption of related Agent Connect features, such as the email digest.

Action Items
Existing Issues

"It's hard to hold associates accountable for Action Items because they're not pulled over to the next 1:1" "No way to bring Action Items over from previous sessions"

We've heard this complaint reiterated by multiple customers during our interviews focused on coaching. This was a surprise for the PM, since past action items were, in fact, carried over from previous 1:1s when starting a new coaching draft.

However, the existing design displayed each action item in its own container, causing the list to grow vertically and hide them in a separate tab. Additionally, the category was a required field when creating a task, disrupting the flow of adding action items one by one.

Existing workflow struggled from multiple usability issues.

Action Items
New Experience

We've heard this complaint reiterated by multiple customers during our interviews focused on coaching. This was a surprise for the PM, since past action items were, in fact, carried over from previous 1:1s when starting a new coaching draft.

However, the existing design displayed each action item in its own container, causing the list to grow vertically and hide them in a separate tab. Additionally, the category was a required field when creating a task, disrupting the flow of adding action items one by one.

New workflow streamlines creating action items.
The first iteration featured a simple, compact list in a half-sized module.
Additional visual cues were added to communicate the concept of a list.
We explored the concept of pinned reminders that stay visible across all sessions. At this stage, I also arrived at the concept of full-width subsections for the session.
The list was organized into three sections: a button to show completed items, open items carried over from the last sessions, and items created today.
The action item was enhanced to handle additional metadata, gracefully handle long values, and offer alternative variants.
A single component in Figma built with variants, variables, and subcomponents to be easily reused in other features.

Sharing
Existing Issues

During interviews with our customers (and subsequently investigating production instances), we noticed a lot of 1:1 sessions left in a draft state.

Draft sessions are accessible only to the coach. This prevented frontline employees from preparing for the session beforehand and blocked access to valuable notes after the session took place.

Our further investigation pointed to multiple root causes:

  • The share button was placed in an unconventional location.
  • The share status wasn't clearly visible.
  • There was no explanation of why sharing the session is important.
  • Some customers had anxiety about sharing their private notes.

Sharing
New Sharing Experience

I aimed to improve this by making the "Share" button the main call-to-action with an explicit label. The new progress component visually communicated the state of completion with the help of badges across the application in a prominent way. Additional text was added to explain the advantages of sharing and recapping sessions.

To alleviate anxiety about sharing private notes, I introduced an automatic locked state for private notes and placed them in a container separate from the coaching session.

Share badge is displayed across the application.

Coaching Mode

While shadowing coaches during their sessions, I noticed many shared their screen while running 1:1 sessions.

This was a natural way for the coach and coachee to use Agent Connect at the same time—reviewing notes, discussing goals, and creating and checking off action items.

The same page was used for two distinct workflows: preparing for a session and conducting it, resulting in a busy interface and distraction from the main task—focusing on the agenda.

Moving the list of previous sessions to a side panel improved both workflows.

By introducing a Coaching Mode switch, I provided a clean way to run 1:1 sessions using the same page, without adding technical complexity.

Coaching Mode allows for a streamlined experience when screen sharing.

Final Design

The new coaching experience was the result of relentless focus on the customer, hours spent listening to and shadowing our users, and absorbing and organizing insights from ProductBoard. It also involved trial-and-error iterations over design with Product Managers and prompt engineering with the engineering organization.

Rapid prototyping with Figma allowed me to quickly explore and validate distinct interaction patterns.

The final design artifact was assembled in a cumulative prototype that served as a design and engineering spec.

Illustration.

Tiles

System of tiles was created to reuse across Coaching and other parts of the application

Attachments

Building on the top of tiles, attachments are able to handle various file types and preview images, audio and video inline.

My Role

I joined the AgentConnect team upon my return to Medallia as a full-time Product Designer.

In my role, I worked closely with the Director of Product and a team of Product Managers to discover, design, and prototype AgentConnect's new features. Together with Product Managers, I conducted user research sessions aimed at uncovering product gaps, validating information architecture, prototypes, and design concepts.

Components and patterns designed at AgentConnect were contributed back to the Alchemy Design System and used across the Medallia product ecosystem.

While leading design for the Contact Center product group, I introduced a process of organizing Figma files and Jira tickets that improved collaboration between the design, engineering, and product teams.

Impact

During my time* at Medallia AgentConnect:

Business

  • Book of business increased by 64%, becoming the second most profitable product in Medallia's portfolio
  • Total active customers grew by 30%
  • MAUs experienced a 42.5% growth
  • 7 out of 10 of the largest US Contact Centers were using our product

Product Metrics

  • QA Reviews increased by 2.7x
  • QA Assignments adoption rate grew by 37%

* Metrics calculated for Q2 2021 – Q4 2022 timeframe