ElevenLabs Compares Hundreds of Support Agents: In Conversations Over Two Minutes, Voice Succeeds 10.9 Points More Often Than Chat

In a study published on its official blog on November 2, ElevenLabs compared hundreds of production agents that run both voice (phone and WhatsApp) and text chat. It included agents that handled at least 50 conversations on each channel between May and July 2026, compared each agent only against itself, and used success criteria each business defined. In conversations under a minute, chat succeeded about 9 percentage points more often; between one and two minutes the channels were roughly even; past two minutes, voice was 10.9 points higher (a 31.7% relative increase), and successful long conversations averaged 4.6 minutes on voice versus 7 on chat. Across 315 agent conversations that switched to a faster model, a 16% cut in average 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 a 98-person study of a simulated canceled-flight rebooking, 85% preferred voice. The research was done by a voice-agent vendor and uses conversation length as a proxy for complexity.

Same Agent, Two Channels

ElevenLabs published a study on its official blog on November 2 that asks a simple question: for the same support agent, does a call or a text conversation get the job done more often? It looked at hundreds of production agents that handled at least 50 voice and 50 chat conversations between May and July 2026, with voice covering both phone and WhatsApp calls. To keep differences between businesses from muddying the result, each agent was compared only against itself, and "success" was judged against the criteria each business defined for its own use case.

The Numbers

  • By length: the study uses conversation length as a stand-in for complexity. Under a minute, chat did better by about 9 percentage points; between one and two minutes the channels were roughly even; past two minutes, voice was 10.9 points higher, a 31.7% relative increase, and nearly two-thirds of agents did meaningfully better on voice.
  • Speed: successful conversations over two minutes averaged 4.6 minutes on voice versus 7 on chat; of the 90 agents with enough successful long conversations on both channels, 78% resolved them faster on voice.
  • Latency: across 315 agent conversations that switched to a faster model, a 16% reduction in average end-to-end latency correlated with a 26% lift in success rate, and the share of calls ending within 15 seconds fell from 21.1% to 3.8%.
  • User study: 98 participants handled a simulated canceled-flight rebooking with both a voice agent and a chat agent in randomized order; afterward, 85% said they preferred voice.

How to Read It

For teams choosing support channels, the data is a useful reference: simple lookups are fine in chat, while refunds and rebookings that need back-and-forth are more likely to be resolved in one go by voice, provided the first response comes quickly, or callers assume the line is dead and hang up. Keep in mind that the research comes from a company that sells voice agents. The authors also concede that duration is an imperfect measure of complexity, that the latency link is a correlation, and that the user study used a simulated scenario rather than real customers.

via: ElevenLabs Blog: Why voice wins for complex, stressful conversations