Way back when I worked in a startup digital bank, with far less resources than the large traditional banks. Every customer support agent had a tonne of customers to look after. This originally got us to think how could we get people onto the call and off it as quickly as possible. We had the usual KPIs around fast service, but it didn’t lead to a great experience.
Given we wanted to be different to the traditional banks and we were more customer focused, we decided this KPI was totally counterintuitive to how we wanted to build relationships with our customers. So, we decided we wanted the team to have proper conversations with customers and even a bit of banter.
At one point, we organised rugby tickets for a customer. I think the original tickets went to an address where they couldn’t get them before the match. One of the support agents got two more tickets and got into a taxi across Dublin trying to find the customer. That agent never had any permission to book a taxi or leave the office or go and deliver the tickets. She did it because she felt it was the right thing to do. When she came back, we were delighted and celebrated what she’d done. She understood that putting customers first and going the extra mile was what we stood for.
Last week I was in a meeting about AI in customer support and I started trying to work out exactly what we are automating away. In the meeting everyone was rushing to the conclusion that we can reduce headcount by leveraging more AI. Just because we can do something doesn’t always mean it’s a good idea. Yes I know bots are cheaper, faster and always on, never take breaks and they handle the boring stuff freeing up us humans to be more creative. That’s the pitch, and for boring or easy tickets it is mostly true. I’m perfectly happy to ask an AI bot how to find my statements or reset my password as I don’t need a person for that and I don’t want to wait for one either.
But the best support is often when someone uses common sense and decides that following the normal process is the wrong thing to do in the situation. This takes common sense and also takes some courage. I’m not sure how you can ever build that into an AI bot.
There’s no way companies will give AI agents the permission to use common sense. Well, for a start they don’t have any and it would be extremely risky to try giving AI bots too much wriggle room especially for regulated businesses.
AI agents follow pre-defined rules and best practices. In banking and payments, these rules exist for a good reason to stop fraud and protect customers. You don’t want an AI agent deciding that one of those rules is getting in the way of a good customer experience. This would give bad actors an opportunity to exploit the AI bot. Try explaining to the regulator that your AI bot thought it was a good idea to leak account details to strangers.
AI agents will have no trouble answering structured questions but when it comes to situations where judgement is needed, thats a different story. That doesn’t mean a person needs to deal with every single question. It just means there needs to be a very smart orchestration layer between what support is routed to humans and what support is routed to AI agents.
My worry is human support will become a luxury product. A status symbol for VIP and premium customers. Klarna’s CEO even called human customer service a VIP thing.
I expect both worlds will coexist, but I’d imagine the one that gets most traction is the simplistic status version. It’s an easy sell inside any company to justify why elite customers need human support. Plus, it’s easy to rank customers by revenue and status. But I believe companies that are more customer oriented and have a bit of a rebellious spirit will look for more sophisticated ways to route by key human moments rather than doing it only on customer status.
Give the login question to the bot and the frozen account call to a person, regardless of how much that person spends with you. Whoever cracks that will be using humans at the right moments, rather than trying to put a person behind every interaction.
I’m more than happy to get agent support for a lot of tasks. But there are certain times where, if I feel my account has been hacked or something serious has happened, I really want to speak to someone who is competent and knows what they are doing. Someone with common sense ideally.
2 comments