Intel / article

Voice Wins Complex Stressful Conversations

Source: elevenlabs.io
Author: ElevenLabs editorial sitemap
Date: Published Nov 2, 2026

Context

ElevenLabs, which sells customer-service agents that work over both voice and chat through its ElevenAgents product, reports that voice beats text chat once support conversations get long. It compared voice calls with chat sessions across hundreds of its customers' production agents that run both channels, using conversation length as a stand-in for how complex an issue was. In conversations longer than two minutes, voice succeeded 10.9 percentage points more often, a 31.7% relative increase, and nearly two-thirds of agents did meaningfully better on voice. In conversations under a minute, chat did better by about 9 points, and between one and two minutes the channels were roughly even (ElevenLabs post).

Voice also finished long conversations faster. Among successful conversations over two minutes, voice averaged 4.6 minutes against 7 minutes for chat, and 78% of the 90 agents with enough such conversations on both channels resolved them faster by voice. The company suggests that part of the reason is that people speak about three times as fast as they type. In a separate user study, 98 participants rebooked a simulated canceled flight with both a voice and a chat agent in random order, and 85% said afterward that they preferred voice, naming speed of resolution and empathy among their top reasons.

ElevenLabs compared each agent's voice results with that same agent's chat results, using agents that handled at least 50 conversations in each channel between May and July 2026. That design keeps a business that mostly uses voice and happens to succeed often from making voice look better overall. Success, however, was judged against criteria each business set for itself, ElevenLabs acknowledges that duration is an imperfect measure of complexity because chat customers may pause between messages, and the study comes from a company that sells voice agents. Its latency finding is a correlation: in 315 agent conversations that switched to a faster , a 16% cut in average end-to-end latency went with a 26% lift in success rate, and the share of calls ending within 15 seconds fell from 21.1% to 3.8%.

Key takeaways · AI-distilled

ElevenLabs compared voice and chat within the same production agents. In conversations under a minute, chat succeeded about 9 points more often; past two minutes, voice succeeded 10.9 points more often, a 31.7% relative increase.
Among successful conversations longer than two minutes, voice averaged 4.6 minutes to resolution versus 7 minutes for chat, and 78% of the 90 agents with enough long conversations on both channels resolved them faster on voice, ElevenLabs reports.
ElevenLabs links latency to early hang-ups: across 315 agent conversations that switched to a faster LLM, a 16% cut in end-to-end latency correlated with a 26% lift in success rate, and calls ending within 15 seconds fell from 21.1% to 3.8%.
In an ElevenLabs user study, 98 participants rebooked a simulated canceled flight with both a voice agent and a chat agent in randomized order. 85% said they preferred voice, naming speed of resolution and empathy among their top reasons.
The comparison used agents with at least 50 voice and 50 chat conversations from May to July 2026, scored against each business's own success criteria, and compared each agent with itself to avoid mix effects. ElevenLabs notes duration is an imperfect proxy for complexity.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.

Read the source elevenlabs.io