Automated customer service through a ChatBot on WhatsApp is here to stay. Consumer self-service interactions on WhatsApp greatly speed up the resolution of many services, such as issuing a duplicate bill or reporting a power outage, for example.
We know that not everything can or needs to be automated, and many services require human interaction and intelligence.
When structuring the ChatBot, we focused on the most important features for distribution utilities — that is, functions that make sense to automate. One of these features is the process known as handoff to human agents, where a person takes over the conversation to help the customer.
In this scenario, agents and managers need tools to make service easier and faster, and to collect information for ChatBot curation and training. That is why the Useall ChatBot is bringing several interesting new features:
1. Service queue control
Every customer needs and wants to be served as quickly as possible. The ChatBot lets customers know when no agents are online — outside business hours, for example. The tool also sends alerts (sound and visual) to the agents who are online, letting them know a new customer is waiting to be served.
2. Multiple agents
One of the most important features is allowing several agents to handle several conversations simultaneously using just one WhatsApp number.
3. Online agent lookup
For teams with several agents, the online agent lookup lets people see who is online to ask for help, share information, or transfer conversations.
4. Detailed conversation reports
Using information is essential to evaluate and improve customer service. Dashboards and detailed service queries show information such as the number of customers served, services performed, wait time, handling time, the number of messages sent and received, and much more.
5. Real-time conversation monitoring
We know automated service on WhatsApp is very good. And we know the ChatBot doesn't know everything and cannot understand many customer requests. That is why monitoring conversations — especially early in the rollout — is important, so we can improve the ChatBot's interactions with customers. Now, we can do this in real time. It is possible to follow customers' conversations and interactions with the ChatBot during automated service. With this, we can assess whether the customer is managing to complete the service and, if there is any difficulty, we can curate and train the ChatBot to improve the experience.
Examples of curation and improvements already made:
The CPF example
In the beginning, we trained the ChatBot to understand the CPF (Brazilian taxpayer ID) as 11 digits or in the standard format 000.000.000-00. By monitoring conversations, we saw that customers type their CPF in other formats, so we needed to teach the ChatBot to understand those formats too.
- Users typed it with blank spaces: 000 000 000 00
- Users typed it without the dots but with the dash 000000000-00, or with an underscore 000000000_00
- Users typed only dots: 000.000.000.00
The Consumer Unit (UC) number example
When the ChatBot asks the customer to type their UC number, it expects the UC number.
However, we noticed that many customers type extra information along with it. So we trained the ChatBot to understand that as well.
- Users type the UC plus a check digit: 99999/9
- Users type text: UC 99999
- Users type a sentence: My UC is 99999
This ChatBot curation is necessary and possible thanks to the monitoring of consumers' interactions with the ChatBot — and it can be done by service agents and managers.
Want to bring these conveniences to your customer service?
Talk to us and schedule a demo of this tool.
Originally published in Portuguese at useall.com.br.