top of page

BusinessDay: Sorry, the chatbot forgot to take notes

The great deal of data harvested from us must be used for better service from call centres

By Johan Steyn, 1 February 2022

A few weeks back my internet fibre line went down. I am fortunate to live near a hospital, which means I rarely suffer power outages and hence always have internet connectivity. Because I work primarily from home, a lack of internet will mean I can not do my job. It’s comparable to planning to go to the office in the morning and discovering your car isn’t working.

I learned, after a desperate search, that my internet service provider does not have a telephone helpline. I was able to communicate only via a WhatsApp chatbot. In principle, I am content to use this technology because it should be quicker than waiting for a human call-centre agent to help me.

I sent my first message and was pleasantly surprised when the bot recognised me and addressed me by name based on my telephone number. It inquired about the type of support I needed and presented me with a menu of possibilities ranging from billing inquiries to new products and technical issues. I chose the latter.

To my surprise, the bot asked me what products I have with the provider: was it LTE, 3G or fibre? I selected fibre and then it asked me if I have a limited or unlimited fibre account. Again, I selected the latter. Then the bot asked me if there was an outage in my area.

How the heck should I know? It directed me to a website to check if the internet in my area was working: it could be a local problem with the exchange. No problem in my area. Next, the bot assisted me with some fault-finding steps: was my router plugged in, are all the lights blinking, was it connected to the wall socket? Check. Then it directed me to a human, and I was extremely happy, only to discover that I was now at the billing department. What? After all this?

The human apologised and transferred me to the technical help desk. Another human answered and asked me what kind of assistance I needed. Phew! After about an hour of waiting and 10 minutes after interaction with the bot and after providing all the needed information, the poor human was in the dark. He had no insight into the myriad of info I had already provided. I was tearing my hair out!

Our service providers harvest a great deal of data from us and this must be used to enable human call-centre agents or chatbots to provide better service. In my case, the bot should have had access to my data and the interaction could have been faster and more satisfying.

I would have expected the bot to know which products or services I have acquired, where I live and whether there was an outage in my area. It should have transferred me to the correct department where the human should have access to all my data and all the information I provided to the bot in order to quickly solve my problem.

Your customers want interactions with humans who are equipped with data to solve their problems. They may even find a chatbot engagement delightful provided that the experience is linked to their profile and anticipated needs using predictive data analytics and behavioural analysis.

• Steyn is chair of the special interest group on artificial intelligence and robotics with the Institute of Information Technology Professionals of SA.


bottom of page