“I’m sorry, I didn’t get that.” “Can you repeat that?” “I think we’re breaking up.”
These are all familiar phrases you may say on a phone call—even (still!) in the year 2025—that can lead to frustration, confusion, and miscommunication.
As the majority of consumers continue to turn to the voice channel for high priority conversations (personal, high-value, or urgent), the stakes are clear. Calls remain the place where consumers decide whether they’re getting the support they need, the quality they expect, and, ultimately, whether they’re going to sign on the dotted line to create or renew their contract.
Reliability and clarity are critical to customer experience and brand trust. When voice quality slips, it doesn’t just disrupt the conversation—it undermines the relationships you’ve worked hard to build and the technology stack you’re relying on to facilitate those interactions.
The impact of call quality on AI deployments
Your AI deployments can only be as good as the channels they are deployed on. If your calls are choppy or unreliable, every downstream AI tool you’re using risks poor performance. That’s because AI depends on clean, accurate data.
As the saying goes: garbage in, garbage out. As word error rate increases, Natural Language Understanding (NLU) engines can see significant declines in their ability to detect intent, sentiment, and compliance—often dropping by 5 to 20 percentage points, depending on the complexity of the task.[1]
Here’s how poor call quality can undercut the effectiveness of your AI tools.
Impact on voice bots
Voice bots rely on accurate speech recognition to understand and process inquiries. If words are dropped, garbled, or delayed due to network issues, your bot’s confidence score plummets. Instead of helping, bots in this situation can instead frustrate customers further, leading to call transfers, extended handle times, and diminished trust in automation.
Impact on AI assistants
In sales and service teams, AI assistants can take notes, capture action items, and generate post-call summaries that guide next steps. But when jitter or packet loss causes keywords to vanish or phrases to distort, critical details can slip through the cracks. That’s missed opportunities, muddled follow-ups, and ultimately, lost revenue.
Impact on AI fraud detection and prevention tools
Fraud prevention AI relies heavily on clean voice signals to operate. Tools like voice biometrics and deepfake detection analyze subtle audio and speech patterns. If call quality is degraded, those patterns become harder to detect, leaving gaps in your fraud defenses and exposing your business to unnecessary risk.
The bottom line: strong AI deployments are only possible when built on a strong, reliable voice foundation.
Threats to call quality
In our 2025 Enterprise Communications Landscape Report, we found that reliability & quality was the number 1 challenge keeping IT leaders up at night. This underscores an unwavering commitment to delivering seamless, high-quality communications experiences—and the factors working against it.
Here are some of the top threats to know:
Jitter
Jitter is any variance in the delivery of packets on a data network, which can cause calls to ‘break up.’ Words may sound jumbled and distorted, leading to frustration, misinterpretation, and confusion.
Latency
Latency is the delay between when someone speaks and when the other person hears it. If there is high latency, the caller and receiver may begin talking over one another or they may believe the other end has disconnected.
Packet loss
Packet loss occurs when a packet of data is lost and the audio never makes it to the other side at all. Missing syllables or phrases can change the meaning of a sentence, lead to repeated clarifications, and erode both patience and trust.
What you need to look for in your communications provider
To protect both your human-to-human and AI-powered conversations, your provider should offer:
Access to voice quality metrics
Look for visibility into jitter, latency, packet loss, and Mean Opinion Score (MOS). These metrics help you proactively identify network issues, understand their impact, and troubleshoot before customer experience suffers.
Built-in redundancy
Networks fail. Carriers drop. The difference is whether your provider has intelligent routing and redundancy in place. A well-architected network automatically reroutes traffic to minimize or even eliminate disruptions to ensure customers never feel the impact of an outage.
Building the foundation for AI-ready communications
The future of enterprise communications is undeniably AI-driven. But AI doesn’t operate in a vacuum. It depends on the quality of the underlying voice infrastructure. Reliable, crystal-clear calls are the prerequisite for accurate bots, effective agent assist, and trustworthy fraud prevention tools.
As you evaluate your communications strategy for 2025 and beyond, ask yourself: are you confident your voice provider can rise to the occasion? The businesses that invest in network reliability and call quality now will be the ones whose AI truly delivers tomorrow.
[1]: Improved Spoken Language Representation for Intent Understanding in a Task-Oriented Dialogue System, June-Woo Kim, Hyekyung Yoon, Ho-Young Jung
