This content is part of the Essential Guide: Guide to AI in customer service using chatbots and NLP

Survey: Customer service chatbots aren't crowd-pleasers -- yet

A recent survey by found that less than half of consumers find chatbots effective. What does that mean for CIOs trying to integrate the technology?

If you've left a conversation with a chatbot feeling frustrated or unsatisfied with your service, you're not the only one.

That's according to a recent survey by and sponsored by customer engagement solutions provider eGain Corporation. The survey, which was conducted in late 2017, garnered responses from 3,000 consumers in the U.S. and UK who had used one or more customer service chatbots in the prior 12 months.

The results were less than glowing -- 53% of respondents found chatbots to be "ineffective" or only "somewhat effective" in helping them resolve issues or get information. That's not an encouraging number, but Anand Subramaniam, SVP Global Marketing at eGain and co-author of the survey, expected worse.

"What I found surprising was the relatively high percentage of respondents that thought chatbots were effective or very effective -- 47%," he said. "This is because the current capabilities of typical chatbots are limited, at best."

Say no to chatbot silos

Many respondents (59%) said the biggest pain point with customer service chatbots was having to repeat information and context when being escalated from communicating with a chatbot to a human agent. The second biggest pain point (32%) was chatbots getting stuck and not knowing what to do next. The authors of the report say this is likely a result of what they call chatbot silos -- chatbot deployments that are completely disconnected from human agent-assisted touchpoints.

Other notable results of the survey include the fact that younger consumers thought that the technology was more effective than older generations. For comparison, 22% of Generation Z consumers labeled chatbots "very effective," while only 12% of Boomers and Silent Generation consumers labeled them so. The study's authors said this is probably because digitally-savvy Gen Z and millennial consumers generally have a more positive attitude towards technologies like chatbots.

U.S. consumers were also harsher in their assessment of customer service chatbots than UK consumers, with 14% of U.S. consumers rating the technology "not effective" versus only 5% of UK consumers. The authors of the report suggest the reason for the disparity is because U.S. consumers ask chatbots more complex questions than UK consumers.

Erwin Van Lun, CEO and founder of, said the goal of the survey was to understand the barriers consumers face when using chatbots.

"The survey findings will help technology providers improve their offerings while helping businesses select and deploy chatbots the right way to make it easy for consumers," Van Lun said.

Chatbots must evolve

As Subramaniam and Van Lun mentioned, improvement is the name of the game. Gartner predicts that 25% of customer service operations will utilize chatbots by 2020, but it's clear that the technology has a long way to go before it reaches its full potential.

Subramaniam said customer service chatbots need to improve in the following areas:

  • Intelligence to understand user intent, answer questions, and resolve issues
  • Integration with human-assisted touchpoints along two dimensions:
    • Transferring context (what the chatbot has learned/determined during its interaction with the customer) to human-assisted service upon escalation
    • Using the same knowledge base as human agents do, to provide consistent answers

CIOs: "De-risk" your chatbot implementation

What tips does Subramaniam have for CIOs looking to improve the customer experience of chatbots? He said they should start by "de-risking" any chatbot implementation. He gave the following recommendations:

  • Make sure the vendor has refined the technology over time, and incorporates best practices into the deployment.
  • Ask for a free production pilot, where there is no obligation to buy anything.
  • Try out the chatbot for a few weeks using these pilots. Ask the vendor to provide expert guidance to realize business value, free of charge, during the pilot.
  • Check if product performance can scale. If the chatbot is not fast during peak demand, consumers will walk away.
  • Ask the vendor what technology they use for knowledge management, learning and reasoning in the chatbot. Also ask them if they integrate out-of-the-box with other customer touchpoints and backend systems.
  • If you are a global business, check to see if the product supports multilingual natural language understanding.
  • Collaborate with other customer experience stakeholders such as the CMO, the customer experience officer or the head of customer service, and the chief digital officer -- if you have one -- to make sure your goals, metrics and timelines for the chatbot project are aligned.

Dig Deeper on Digital transformation

Cloud Computing
Mobile Computing
Data Center
and ESG