After more than a decade with M1 broadband, I finally asked for it to be terminated today. It’s been the only fibre operator I’ve had since Singapore rolled out the high-speed Internet service but I’ve already added a ViewQwest line and switched over more than a month ago.
Asked by an M1 agent why I’m leaving, I said I was disappointed at the customer service and that M1’s promotional prices were available only to new customers, not existing ones.
At least this time I was passed to a human agent – or what seemed to be a human agent – after being greeted by a chatbot. The last time I tried calling M1, about a year ago, I was held back by an AI bot, which wanted me to explain in detail why I was calling.
Rather than helping with customer service, AI chatbots and AI tools have been poorly deployed by many businesses – including Singapore telecom operators – so they end up destroying years of goodwill.
In the highly competitive telecom market in Singapore, it’s understandable that many players want to cut costs by reducing the “high touch” interactions between customers and their human operators.
Yet, some of these cost-cutting implementations have been downright disastrous. If businesses think that each human interaction avoided – a key performance indicator for some that mindlessly deploy AI – is good news, then they are making a huge mistake.
Each customer pushed away could be someone trying to recontract or seeking help for a complex issue that a chatbot could not resolve.
In a tough market, the way forward does not have to be a rush to the bottom. Finding a way to use AI alongside humans could improve customer service and make it a key differentiator.
One positive example I experienced recently was with Amazon Singapore. After telling a chatbot I wished to speak to a human operator, it just made sure I confirmed which products I had an issue with and quickly passed me on.
When the support person came online, they already had a record of what I was about to ask, saving us both precious time. My parcel from the United States, for some reason, had been turned back at the customs over there because the courier found it damaged, the Amazon staff explained.
Whew! Thankfully, it wasn’t because the US government was stopping the export of 5Gbps network adapters to Singapore!
That resolved, the Amazon staff said a refund would be made to me. Sure enough, that came through in the following days and the chat only lasted a couple of minutes.
A lot likely happened in the background, possibly with automated processes that fit a complex business logic. This way, the customer-facing agent only had to make sure everything was in place when speaking to a customer.
So, AI can and should help businesses improve their customer service. Indeed, it should fulfill its promise of empowering frontline staff do more and focus on the customer.
Crucially, that doesn’t happen overnight. One way is to start with the low-hanging fruit. Think of automating “deterministic” processes such as recontracting or other process approvals that meet standard procedures with robotic process automation (RPA).
Can agentic AI help you do away with writing all these rules, by learning them and then “reasoning” its way to a decision? Well, that again takes time and the potential for mistakes can be devastating if an AI makes a mistake, say, by awarding an insurance claim or sending a refund erroneously.
For those who think you don’t need any human operators any more, consider the U-turn taken by Swedish fintech company Klarna.
After laying off 700 employees in 2022 and boasting that AI is helping it cut costs, it reversed course earlier this year. Quality had been impacted, it acknowledged and it re-hired workers again.
That said, AI agents are becoming more powerful, both in the front- and back-end for customer services. And they would make a big difference if they can take on more of the jobs of human operators.
Key to great customer service, as I’ve learnt recently, is to combine the best that both humans and AI can offer. AI can help with a lot of the process and automation tasks but humans can build trust by intervening when needed.
Businesses that have bought the AI hype and pushed a chatbot upfront to frustrate customers will find that their cost savings are short term. Shareholders will be happy for a while but like all shareholders, once the AI bubble deflates a little, doubt starts creeping in.
Or, rather, reality starts knocking. If you lose a customer, it costs you to win them back, say, through marketing efforts or price discounts. Goodwill is hard won but easily lost – AI is helping businesses lose that quickly.
For me, at least I managed to resolve my issue with M1 quickly this time, if only because I was cutting my subscription. On a positive note, I am not leaving angry because I got my issue resolved quickly, without AI hindrance.
That’s a a low bar, but you hope businesses are taking note of the long-term costs they are bearing by deploying AI without thinking more deeply. If customers are made to be part of a grand experiment, they can’t be expected to hang round until things work out.
