There’s lots of excitement around bots, and confusion. So let’s clear up a few things.
The primary reason for a bot is to save the customer’s time, not to reduce live support contacts. Bots can reduce live support contacts, but that should never be the primary reason to use a bot, if you care about customers.
Tasks that customers can do faster with a bot than by speaking with a person should be handled by a bot. Tasks that customers can do faster by speaking with a person should be handled by a person. Sounds simple, right?
Bots are good at specific, clearly defined tasks such as collecting information needed to route customers to the best support agent, or to provide frequently requested information. In these cases, bots allow human agents to focus on higher-level tasks, which makes their work more meaningful and efficient.
Don’t try to boil the ocean. All bots have limits. Most bots don’t handle complex tasks well. The scope of what a bot can do must be clear to customers, particularly customers who need to resolve problems. It’s possible to spend millions on creating a bot that’s a very expensive search engine that returns irrelevant answers that drive your customers to other brands.
When implementing a bot, the content behind the bot is as important as the natural language processing and other technical features of the bot. Bot content must be developed in close collaboration with those developing the bot conversation flows and user interface. How accurately a bot understands a customer’s conversation doesn’t matter if it can’t provide good answers.
Use bots only for tasks that the bot can handle with at least 90% confidence. For tasks where the bot is less confident about the correct responses, bots can still be useful internally to assist live agents. This approach can train bots to do more tasks, and train new agents.
Most people prefer self-help for simple tasks. For complex tasks, most people still prefer to speak with a person.
Because your competitor uses bots doesn’t mean that bots work for them. Bot resolution rates depends on the complexity and variability of your tasks. Don’t use bots from retail commerce sites to set expectations about how a bot might work in a support environment, for example. Helping someone decide which umbrella to buy is very different task than helping someone diagnose a problem with their product.
Bots are a competitive advantage, either to you or to your competitor.