Media Communications reduces call drop rate to <1 % from 3%

Overview

How Media Communications put a conversational voice agent in front of roughly > 1,000 calls a day and gave its human team more time for the cases that genuinely need them.

Company: Media Communications, an international online marketing company supporting multiple brands and products across teams in Virginia, Florida, and the Philippines.

Use case: Inbound customer service, including order lookup, product questions, refunds and partial refunds, after-hours self-service routing, and human escalation.

Results at a glance

• Drop calls fell from above the 3% benchmark to around 1%. Dave said the rate usually stays in the 1% range or below and rarely approaches 2%, even at roughly 1,000 calls per day.
• 75–80% of inbound calls are handled by the AI agent. Tier 1 requests can be resolved without a human, while more complex issues transfer to the live team.
• 30 AI agents can answer simultaneously. Peak demand becomes a concurrency setting instead of a queue that overwhelms the floor.
• Customers get help faster. Media Communications has seen fewer complaints, including fewer third-party escalations.
• Refund conversations are more consistent. The AI agent has produced cleaner partial-refund outcomes than the team previously saw from live handling.
• After-hours routing no longer requires maintaining roughly 150 IVR messages. The agent identifies the relevant account and directs each caller to the appropriate self-service URL.

The problem: every Monday, the floodgates opened

Media Communications runs customer service for a portfolio of brands. For Dave, its Chief Customer Officer, the standard is straightforward: every call should be answered.

“It’s important to answer every single call, and if every single call comes in and you’re overwhelmed, especially on a Monday morning—it opens up the floodgates as soon as we start.”

The company’s drop-call rate was frequently above the 3% benchmark, sometimes significantly. Mondays were especially difficult. At roughly 1,000 calls a day, every percentage point represented customers who could not reach the team when they needed help.

The existing after-hours workaround created another operational burden. A traditional IVR directed callers to self-service URLs, but supporting many brands meant maintaining roughly 150 recorded messages. Any change in operating hours required editing them individually.

“Anytime we wanted to change the hours, we’d have to go back in and change 150 messages.”

Consistency across the human team was also difficult to maintain. Training could address a recurring issue, but not every agent applied the guidance in the same way.

“You see an issue that everyone’s having, and you can train and train and train. Some agents get it, some don’t.”

Why voice AI and why conversational

Dave's confidence in AI came from an unexpected place: music. As a musician, he had watched AI compress studio work he had pursued for decades into minutes. He then began producing complete 30-second commercials himself, including video and music.

“If I could do that by myself, what could a company that specializes in customer service AI do—for not only the speed, but the consistency? That’s the main thing. It’s just consistent.”

The initial implementation was not a perfect fit. Media Communications began with a task-based approach, but the turning point came when the team moved to a conversational agent. That shift enabled more flexible routing and natural problem-solving instead of forcing customers through a rigid decision tree.

“The other type is good for a yes-or-no thing, for companies that just want to pass it off to this or that—that’s an IVR. But if you’re really trying to handle business, conversational is the way to go.”

What Jessica does

Media Communications calls its AI voice agent Jessica. The voice was already familiar to the team from its IVR, helping the new experience feel consistent from the start.

Tier 1 support from start to finish

Jessica handles product-usage questions, order inquiries, cancellations, refunds, and partial refunds. She can address some Tier 2 needs as well, and transfers the call to a live agent when a customer has a unique or complex problem.

Account-aware routing

When a customer calls from the phone number used for an order, Jessica can identify the account and the relevant brand. She then directs the customer to the correct self-service URL for account management, refunds, repeat purchases, or product guidance.

After hours, this replaces the maintenance-heavy IVR tree.

“Now nothing changes. Jessica just adapts.”

More consistent refund conversations

The refund workflow delivered an unexpected result. Media Communications found that when Jessica could offer a partial refund, the interaction was often handled more cleanly and consistently than it had been by live agents.

“If it hits the AI agent and they’re able to do a partial refund, it just works cleaner than the regular live agents.”

Concurrency without a queue

Human capacity is inherently sequential. A sudden surge can overwhelm even a strong team. AI changes that constraint.

“If I’ve got a hundred calls coming in and 15 agents, it doesn’t take long to figure out 15 agents can’t handle 100 calls. But Jessica can handle as many as we set it up for. I’ve got 30 AI agents on the line ready to go.”

A dashboard built for operations

Every call receives a plain-language summary in the SigmaMind dashboard. Dave’s team can search by operational keywords—such as refunds, upset customers, or transfers—to find calls that need attention and identify recurring issues quickly.

“Whatever variable we’re looking for, we can address it.”

The metric that mattered most was speed

Media Communications has always prioritized customer satisfaction. It offers full refunds when appropriate, often without requiring a product return. The problem was not the resolution itself; it was how long customers waited to receive it.

“The metric that changed is the time that you get to that customer. The customer is not on hold forever.”

Faster access has changed the customer experience. The business has seen fewer complaints overall and fewer escalations to third parties such as the Better Business Bureau or state Attorneys General.

The conversations themselves also feel natural.

“When they talk to Jessica, they feel like they’ve talked to Jessica. Listening to the calls, I can tell some of the customers don’t even know they’re talking to an AI agent.”

Fix it once, and it stays fixed

Dave is candid that the implementation was not turnkey. The first months required close collaboration, and the original agent design was not right for the complexity of the work.

What ultimately changed his view was the durability of each improvement.

“When we start seeing a metric that shows 30 customers have this issue every time they call in, we fix that and it’s gone. That’s not how it works with a human agent. With AI, you find that issue, you fix it, and it’s done.”

The team still monitors performance, but recurring mistakes have become less common and the operational stress has fallen.

“There are hiccups with all AI programs. But generally speaking, once you get something right with the AI agent, it stays right. And that’s a lot different than a human agent—constantly retraining, constantly monitoring.”

Dave summarized the change in perspective simply:

“Three months ago I said this is never going to work. Now it’s like, wow, this is working great.”

A working partnership, not a vendor handoff

Reaching that point required a joint engineering effort. SigmaMind worked directly with Media Communications’ developers, iterating on the agent and responding quickly as new issues surfaced. Lessons from the engagement also informed improvements that SigmaMind could apply more broadly.

“It’s been a win-win between SigmaMind and Media Communications. You guys have been there all the time—no matter what time of day it is. It’s really been a great relationship.”

What’s next

Media Communications expanded its telephony infrastructure from one server to three to support the additional call concurrency. The company is also considering how AI could support its email ticketing workflow.

For voice, the decision is already made.

“I know that we will continue to use the SigmaMind AI for the calls. That’s for sure. Very, very happy with that.”

From overloaded Mondays to dependable coverage

SigmaMind gave Media Communications a way to absorb peak volume without leaving customers waiting. The AI agent now handles most Tier 1 demand, routes account-specific after-hours requests, escalates complex cases to the human team, and turns recurring service problems into fixes that can be applied consistently across every future call.

“For us, it has been a game changer. It brought our drop-call numbers back into reality. That’s incredible when you’re dealing with a thousand calls a day.”

— Dave Welsh, Chief Customer Officer, Media Communications