Nobody asked the customer
There is an argument about AI in customer service that many organisations make, and I understand why they make it. The better AI becomes, the argument goes, the more the personal contact with customers matters. The employees in customer service explain bills, advise on products and notice when irritation is covering a real worry. AI should sit beside them and take care of the searching, the summaries and the notes, so that the employee on the phone can give the customer their full attention.
I agree with a good deal of that. I also think it is out of date. It underestimates what AI can do in customer operations in 2026, and more importantly, it underestimates the customer.
In my view, the customer was never asked.
The customer’s experience starts in the queue
Think about what it takes to register a move with your energy company. You find the number, work through a menu and join the queue. You call in your lunch break, because the line closes at four. After 20 minutes someone answers, and I have no doubt that person is friendly, competent and patient. They then spend the first part of the call finding you in the systems that hold your contract, your meter and your billing address.
Everything the argument values happens after that. The warmth is real, and it arrives after the part of the experience the customer actually minds.
An example: registering a move with an energy company by phone
20 minutes in the menu and on hold, 6 while the employee finds the account and 4 to register the move. The line closes at four.
Identified with MitID and registered in the same call, at 21:40 on a Sunday.
Most people I know would like their address changed without giving up their lunch break, and very few of them are looking for a warm relationship with their energy company. For all of them, a call that is picked up at once, at any hour, and ends with the change made is a gift.
The queue is a decision someone made
Every customer service operation decides how long its customers wait. It decides through the number of employees it has, the hours they work and the target it sets for answering calls. For an organisation whose capacity is its employees, those are reasonable decisions. They are still decisions about the customer’s time, and the customer has no part in them.
So when an organisation says its customers value the personal contact with an employee, what I hear is that it has decided, on the customer’s behalf, that the contact is worth the wait. I would like to see the customers asked. My guess is that most of them would describe their request plainly. The address needs changing, the bill needs explaining, the card needs blocking or the claim needs to move. They want it done, and quite often at nine in the evening, when they finally have time for it and the line has been closed for five hours.
The people who need a person wait in the same line
People matter more than ever in customer service, and on that point I agree completely with the organisations I am arguing with. Patience, judgement and the ability to hear the worry beneath the irritation are exactly what a customer needs when a bill is disputed, when a claim has become complicated or when someone has died and the account has to be moved. Those calls deserve an employee with time to give.
The trouble is where those customers are today. They wait in the same queue as everyone else. A customer whose partner has died, and who needs the account moved into their own name, is waiting behind the meter readings, the payment dates and the address changes. An organisation that treats every request as work for an employee has, without meaning to, made the customers who most need a person wait as long as everyone else.
What I think good service looks like
When the design starts from the customer, requests divide by what the customer needs. A large share of them ask for something the organisation’s own systems can do: move the supply to a new address, register a meter reading, confirm when a payment will be taken, send a copy of a policy. Those can be resolved when they arrive, in a service that is always open. The requests that remain need a person, and they reach one sooner, because the queue in front of them is much shorter.
This is what we build at oHallo. oHallo picks up the moment the customer calls, and answers written requests at any hour of the day. Where the request needs it, the customer is identified with MitID or BankID before anything on the account is read or changed. oHallo works in the organisation’s own systems while the customer waits, under the organisation’s own operating procedures, and it can be every bit as polite as the organisation wants it to be. When a request needs a person, the call is transferred with everything said so far, the account and everything already checked on the screen, so the customer does not have to explain the matter twice.
AI that only helps employees answer faster has been measured, and I come back to that study in the next article. Seen from the customer’s side, the three versions of the service compare like this.
AI-assisted figures: Brynjolfsson, Li and Raymond, Quarterly Journal of Economics, May 2025, 5,172 support employees. Time on hold, opening hours and cost follow from every request waiting for a free employee.
I take the preference for a person seriously
Customers are not indifferent to who answers, and I would be wrong to pretend otherwise. Asked in surveys, many say they would prefer a person even if the outcome and the wait were the same. I think that preference deserves a proper answer, and I give mine in a later article in this series. For now I would only point out that it describes a comparison very few customers ever get to make, because today the outcome and the wait are seldom the same.
Ask your customers
The organisations that defend the personal contact are right that customer service is where a company earns or loses trust. My suggestion is that they test the rest of the argument with their own customers. Measure how long a request takes from the moment the customer makes contact to the moment the task is done, including the time in the queue and the hours when the line is closed, and put that figure beside the satisfaction scores.
I think I know what they will find. That figure is the customer’s view of the service, and it is the one the argument leaves out.
Related
The employee is still the unit of capacity
AI-assisted customer service makes each employee faster. I think it leaves the customer where they were, because every request still waits for an employee to be free.
27 Sept 2026Customers want to know a person is within reach
Many customers say they would choose a person even if the outcome and the wait were the same. I take that seriously, and I think it tells us what an organisation has to promise.