Embedded Files
The problem
The problem
Company X is launching a new contact flow designed to embed the company's empathy model into every customer phone interaction. Customer Service Associates must follow new customer relationship management steps while also demonstrating skills such as active listening, de-escalation, and empathetic language.
Design approach and rationale
Design approach and rationale
This branching scenario was built around a single, common scenario, a frustrated customer facing a time-sensitive delivery failure. I chose this because emotional pressure is the real test of empathy skill. Associates can recall the their training in a classroom; the goal here is to practice applying it when the stakes feel real.
Experiential learning is supported through structural choices. Every decision point presents plausible options rather than an obvious right answer and one or two obvious incorrect answers. The “okay” choice in each branch is a realistic option, and one that is a very likely choice in some situations. The learning happens in discovering why technically correct responses still fall short without empathy. Additionally, Maria's responses shift in response to the learner's choices, making the emotional consequence of each decision tangible rather than abstract. Finally, there is feedback provided for each choice so the learner can see the reasoning behind the result.
The Empathy Score reinforces that empathy is not a single gesture at the opening of a call but a sustained discipline across the full interaction: opening, de-escalation, and close all carry equal weight.
I mapped out the design for this module using Miro so that I could visualize the paths and results for the learner.
Design Summary
Design Summary
The Customer Connection rollout presents two distinct but interdependent performance gaps:
Technical process gap that requires in-flow guidance
Behavioral gap of knowing when to apply empathy under pressure
A single training modality cannot address both. The solution is a layered ecosystem in which each platform is used to meet the separate needs.
WalkMe for the in-flow technical support
WalkMe sits directly inside the CRM and activates in real time as associates navigate the Customer Connection flow. Key implementation elements include:
Step-by-step Smart Walk-Thrus triggered at each new CRM stage, guiding associates through the required process steps without leaving the application.
Smart Walk-Thru automated steps to streamline the usage of the application.
Contextual SmartTips surfaced at high drop-off points to reduce hesitation and error without supervisor intervention.
Custom task list that clears WalkMe data daily.
ShoutOuts to provide information on system updates or changes.
Axonify to reinforce soft skills
For this rollout:
Empathy Model knowledge checks reinforce the empathy principles through scenario-based questions tied to real call situations.
De-escalation and active listening scenarios present branching judgment prompts, reinforcing correct language choices and flagging misapplication for targeted follow-up.
Spaced repetition scheduling ensures concepts are revisited at optimal intervals, driving retention over the 90-day window rather than relying on one-time classroom training.
Integration and measurement
Data from WalkMe and Axonify serve as leading indicators of adoption rates, skill transfer, CSAT improvement and escalation reduction. Weekly reviews in the first 30 days, shifting to bi-weekly thereafter, allow rapid course correction.
AI-enhanced content creation
Targeted GenAI applications will accelerate content development and enable the kind of realistic, high-volume practice that classroom training alone cannot provide.
Second Nature AI role-play simulations Second Nature is purpose-built for the exact scenarios this rollout requires. It will be used to rapidly generate AI-simulated customer call scenarios, angry callers, order disputes, complex CRM interactions, giving associates a live conversational practice environment with immediate feedback on tone, empathetic language, and adherence to the empathy model.
WellSaid Labs for AI voice generation WellSaid Labs would be used at deployment because of the ability to generate or update narration in minutes. WellSaid's natural-sounding voices matter here specifically, because a robotic narrator would undermine the empathetic communication skills the content is designed to build.
AI-assisted scenario scripting for Axonify GenAI (Claude / ChatGPT) will be used to rapidly draft branching scenario scripts and knowledge check questions for Axonify. Using a structured prompt template grounded in real call transcripts and QA rubrics, SMEs can generate an initial draft of 30–40 questions in a single session and focus their time on refinement rather than authoring from scratch, allowing for responsive iteration.
Quality assurance and humanization
AI-generated content introduces three specific risks for this project: factual inaccuracy in CRM process details, tone that feels scripted or cold, and scenarios that inadvertently reflect demographic bias. Three QA mechanisms address each directly.
Prompt grounding in real source material: It is important to work with SMEs to generate source material before any AI building. All AI prompts will be built from actual call transcripts, existing QA rubrics, and approved existing empathy model language. Anchoring the AI generation in real source material reduces hallucination, eliminates generic corporate-speak, and ensures output reflects how associates actually talk to customers.
Structured SME review with a split checklist: Rather than asking SMEs to review AI output holistically, the review is divided into two distinct passes: a CRM/operations SME validates process accuracy (correct steps, current policy, accurate CRM logic), and a customer experience or QA lead reviews empathetic tone and language authenticity. Separating these reviews ensures neither gets absorbed into the other, which is a common issue when a single reviewer tries to assess everything at once.
Pilot testing with a representative associate cohort: Before full rollout to 4,000 associates, all content will be tested with a cohort of 50–75 associates across tenure levels and team demographics. Associate feedback on whether scenarios feel realistic, fair, and reflective of actual customer interactions is a signal that SME review alone cannot provide. Second Nature role-play scores and Axonify proficiency data from the pilot cohort will inform final adjustments before launch.
Sofware Used
Sofware Used
Articulate Storyline
Adobe Illustrator
Page updated
Google Sites
Report abuse