We Cut Call Abandonment from 30% to Under 10% in 12 Weeks — Without Hiring a Single Agent

A healthcare scheduling operation. 110 agents. Patients calling to schedule appointments, confirm visits, reschedule, ask questions — the kind of calls that directly determine whether a clinic runs full or runs empty.

The abandonment rate was 30%.

Three out of ten patients who called gave up before reaching an agent. They hung up, didn't reschedule, missed their appointment, or called back later and clogged the queue again. In healthcare scheduling, an abandoned call isn't just a missed interaction — it's a missed appointment, a gap in the provider's schedule, and a patient who might not come back.

The operation didn't have a staffing problem. It had an allocation problem. And nobody had mapped the workflows well enough to see it.

The Real Problem: Outbound Was Choking Inbound

Here's what was happening on the floor every day.

Agents were making outbound reminder calls. Appointment confirmations, no-show follow-ups, recall outreach. These calls had low connect rates — most went to voicemail. An agent would dial, wait, leave a message, note the record, dial again. Repeat.

While those agents were tied up making outbound calls that mostly went unanswered, inbound calls were stacking in the queue. Patients who actually wanted to talk — to schedule, to reschedule, to ask a question — were waiting. And waiting. And hanging up.

There was no tracking mechanism. Nobody was measuring the cost of outbound calling in terms of inbound abandonment. The operation didn't have documented workflows — it ran on tribal knowledge. Supervisors knew the general rhythm, but nobody had mapped the actual flow of work across the floor to see where the bottleneck was.

That's where I started.

Step One: Map What Nobody Had Documented

There was no documentation when I arrived. No workflow diagrams, no call routing logic written down, no SOP for how agents moved between outbound and inbound. Everything lived in people's heads.

I mapped every workflow from scratch. How calls came in. How they were routed. What agents did at each step. How outbound campaigns were scheduled. What happened when an agent was on an outbound call and an inbound call hit the queue. Where the handoffs were. Where the dead time was.

The map made the problem obvious. Outbound calling was a low-yield, high-cost activity — not in dollars, but in agent availability. Every hour an agent spent dialing outbound was an hour they weren't available for inbound. And the inbound calls were higher value: a patient calling in is a patient ready to act. An outbound call to a number that doesn't pick up is just burned time.

The fix wasn't “hire more agents for inbound.” The fix was “stop using agents for work a machine can do better.”

The Fix: Programmatic Texting Replaces Manual Outbound

The outbound reminder calls — appointment confirmations, recall outreach, no-show follow-ups — didn't need a human voice. They needed a message delivered reliably. Programmatic texting does that better than a phone call in almost every way:

• Texts get read. Calls get ignored.
• Texts don't require the patient to be available at the exact moment you reach out.
• Texts cost zero agent time. Every text sent is an agent freed for inbound.
• Patients who need to respond can call back — converting a low-yield outbound attempt into a high-intent inbound call.

The best part: we didn't need a new platform. The texting capability already existed in the system the organization was already paying for. It had never been operationalized. Nobody had built the workflows, configured the campaigns, or defined the routing logic for inbound callbacks generated by text outreach.

I built all of that. Campaign logic, text templates, callback routing, and the workflows to connect the inbound response to the original outreach reason — so when a patient called back after getting a reminder text, the agent had context immediately.

The Hard Part: Cross-Training 110 Agents

Shifting from outbound to inbound sounds simple on paper. It isn't.

The outbound agents had one skill set: dial, confirm, leave a message, document. The inbound workflow is fundamentally different — you're reacting, not initiating. The patient sets the agenda. You need to navigate scheduling logic, insurance verification, provider availability, multi-site coordination. It's a more complex call.

I couldn't pull all outbound agents off the phones and put them through a full training program. That would take the normal 8-week ramp and leave the floor short-staffed during the transition. So I built a cohort model with a tiered skill approach.

Week 1–2: New skill bucket, minimal training. I created a new low-complexity call type in Five9 — a skill bucket that routed only the simplest inbound calls to the transitioning agents. Basic scheduling confirmations, straightforward appointment requests. Two weeks of targeted training on these specific workflows, not the full inbound curriculum. One quarter of the normal training timeline.

This got agents taking inbound calls immediately. Not all inbound calls — just the ones with the lowest complexity and the highest volume. That alone started pulling calls out of the queue and freeing up the experienced inbound agents for the harder interactions.

Weeks 3–8: Upskill in waves. As each cohort stabilized on the basic skill bucket, I expanded their routing. More call types. More complex scenarios. Additional training in focused sessions, not a second boot camp. The IVR routing logic determined which calls went to which skill level, so agents were never thrown into calls they weren't trained for.

Weeks 8–12: Full integration. By the end of the 12-week engagement, the former outbound-only agents were handling a full range of inbound scheduling calls, routed by skill level based on IVR input. The most experienced inbound agents were freed up for the most complex scenarios — multi-site scheduling, insurance coordination, provider-specific logic.

The key was never disrupting service during the transition. Patients didn't experience a gap. The queue didn't get worse before it got better. It got better immediately.

The Results

Call abandonment: 30% → under 10%. The change was immediate. As soon as agents shifted from outbound to inbound, the queue started clearing. The 30% abandonment rate didn't gradually decline over 12 weeks — it dropped as soon as the allocation changed.

Zero headcount increase. We didn't hire a single agent. The same 110 people handled the same volume — we just stopped using them for work that a text message does better.

Zero additional platform cost. The programmatic texting capability was already licensed. It had been sitting unused because nobody had built the workflows to operationalize it.

Documented workflows that didn't exist before. Every call flow, every routing decision, every SOP — mapped and formalized from scratch. The operation went from tribal knowledge to documented processes that could be trained, measured, and improved.

Still in use today. The workflows and the outreach model I built are still running. They've been enhanced since the engagement ended, but the foundation — programmatic texting for outbound, agents focused on inbound, skill-based routing for cross-trained staff — is the operating model.

No complaints from the patient side. In healthcare operations, that's actually the strongest signal. You hear about problems. Silence means it works.

Why This Matters Beyond One Scheduling Operation

Every contact center has some version of this problem. Agents doing low-yield manual work while high-value interactions wait. The specifics change — maybe it's outbound collections calls instead of appointment reminders, maybe it's manual data entry instead of outbound dialing — but the pattern is the same:

1. 1.Nobody has mapped the workflows well enough to see the allocation problem.
2. 2.The platform has capabilities that haven't been operationalized.
3. 3.The fix doesn't require new headcount or new technology. It requires someone to look at the operation as a system and redesign how work flows through it.

That's process improvement applied to contact center operations. Not a methodology pitch — a workflow redesign that freed 110 agents to do the work that actually required a human.