Automated Associate Call-Out

Calling out of work used to be a headache for store associates—they had to track down a manager, wait on hold, and sometimes call multiple times just to make sure their absence was recorded.

I designed a structured IVR system that let associates handle this process quickly and reliably over the phone. By guiding callers through a clear conversation and providing confirmation via SMS or a number, I made the experience smoother and more trustworthy. The system now handles roughly 257,000 calls per month, automatically containing 198,000 of them—about 77%.

Timeline

4 Months

Project Type

IVR - Interactive Voice Response

Roles

Product/Conversation Designer

The Problem

Store associates who needed to call out of work had to:

  • Call their store directly

  • Wait for a manager to answer

  • Sometimes call multiple times

  • Risk not getting through at all

Managers had to:

  • Interrupt store operations

  • Manually collect information

  • Log call-outs separately

This was a high-volume, low-complexity task consuming valuable labor time on both sides.

The Opportunity

This was a clear automation candidate because it:

  • Followed a predictable structure

  • Required limited data inputs

  • Needed reliable documentation

200K call out attempts per month

Significant repeat calls due to missed connections

Research & Tuning

Part of my role consisted of building and refining the NLU structure. I needed to design for real language, so before creating new intents, I:

  • Analyzed existing IVR menu data

  • Reviewed how associates were already phrasing call-outs

  • Created new NLU intents to intercept call-out phrases earlier


This reduced friction at the top of the call.

I’m an associate I’m calling out sick

calling out of work

employee absence

I need to call in sick

calling out

I’m an employee calling out sick

Building the Design Logic

Key considerations:

  • Multiple date phrasing patterns

  • Numeric ID confirmation

  • PTO branching logic

  • Invalid input handling

  • API failure responses

  • Re-entry paths without restarting the entire flow

I mapped:

  • Happy path

  • Invalid date paths

  • API failure states

  • Retry limits

  • Escalation conditions

This was a deterministic system — clarity and error containment were critical.

Confirmation & Trust

The confirmation experience was critical. After submission:

  • Caller received a confirmation number

  • Option to repeat it

  • Option to receive SMS confirmation


The confirmation number acted as proof.

I also defined the dynamic SMS that included the:

  • Sales ID

  • Date

  • PTO type and amount

  • Confirmation number


This wasn’t just automation; it was design that built trust.

Impact

Projected:

  • 200K calls/month

  • 100K potential containment

  • 50% containment rate

Actual (2025):

  • 257K calls/month

  • 198K monthly calls contained

  • 77% containment rate

That equates to:

  • Massive reduction in manual interruptions

  • Significant labor savings

  • Improved associate experience

  • Reduced repeat call attempts


This feature turned a friction-heavy task into a structured, reliable system.

Let's connect