Voice self-service hasn't improved since 2006, research firm ContactBabel reported in its latest "Inner Circle Guide to Chatbots, Voicebots & Conversational AI."
Business drivers for chatbots, voicebots and conversational AI are based on making self-service a better experience for customers while reducing the costs of unnecessary calls, it said.
And while there was a rapid rise in the proportion of web chats handled automatically and a significant drop in the cost of doing so, this trend has stalled as rules-based bots reach the limit of their capabilities, it said further.
But, the headline finding is not that AI has failed; it’s that investment has climbed while outcomes have flattened, according to Steve Morrell, managing director of ContactBabel.
Some other findings in this year's report include the following:
- Web chat automation has sat at around 50 percent of chats for four years as rules-based bots hit the limit of what they can do. This comes after the level of automation steadily increased every year since 2015, when it was at just 5 percent. It climbed to 12 percent in 2016 and 2017, 19 percent in 2018, 24 percent in 2019, 33 percent in 2021, 50 percent in 2022, and 53 percent in 2023 and 2024. It dropped to 46 percent in 2025.
- Voice self-service still handles only 9 percent to 10 percent of interactions, almost exactly where it was 20 years ago.
- Only 19 percent of operations use voicebots today, yet 45 percent expect to do so by the end of 2027.
- The complexity of web chats is on the rise. Today, only 13 percent of web chat sessions now last less than three minutes, down from 25 percent last year.
- Agent-handled customer identity verification costs U.S. companies roughly 72 cents per call, totaling roughly $12 billion a year.
- Ninety percent of businesses say customers want the reassurance of a live agent.
With around 25 percent of calls coming from customers who have tried and failed to resolve their issues online, the opportunity for improving performance and customer experience by improving and developing self-service is massive, Morrell asses, noting that the next step is to use agentic and generative AI to move the chatbot/voicebot revolution forward.