7 reasons not to automate customer service… AND WHY THEY'RE WRONG
30 May 2023
6 min read
These days everybody seems to be talking about artificial intelligence as we marvel at its growing capabilities. However, whilst we enjoy discussing the possibilities of AI, many of us still show resistance towards customer service automation in our daily lives.
At the start of the year, we asked 200 senior CX leaders in eCommerce to answer our survey which amongst other questions asked exactly that: “What, if anything, would stop you automating more of your customer support processes?”.
Instead of having a few specific concerns stand out, the responses were pretty evenly distributed among the seven options we offered.
Do any of the following reasons chime with you? Read on to find out what they are and why we think they’re unfounded.
1. The responses would not be personalised
What does it mean to know someone? A lot of people say that you need to travel with a person to really know them. Others say you need to watch them interact with waiters. The lesson here is that you accumulate knowledge from each contact you have with them. It’s this shared experience that you draw from to figure out what they would like to get for their birthday, for example. As Dr. Phil would say, “the best predictor of future behaviour is relevant past behaviour.”
In a well-designed automated system, information is integrated in a manner that it is very easy to access, whether it’s the customer’s shipping information, personal preferences or any previous communication. You can take this further and integrate automation into your business’s CRM platform to guarantee that automation has all the tools it needs to create a personal interaction. A customer service experience that is tailored to that specific customer in that point in time.
And if you’re still not convinced, believe us when we say that with new generative AI tools like ChatGPT, customer service teams can generate a limitless number of options for messages giving interactions a personal feel. This area is actually developing so quickly that Neil Sahota, lead artificial intelligence advisor to the United Nations, told the BBC last February that “we may actually see AI emotionality before the end of the decade.”
Even then, in the here and now, there are broadly two types of customer queries – which we’ll call “Transactional” and “experiential”. Transactional queries would be ones where there is a quick, factual answer. For example: “Do you deliver to France?” – here the answer is Yes or No, and there isn’t a lot of room to add personalisation except in a purely cosmetic way. These are the kinds of questions that automation is best suited to.
For the more “experiential” questions – e.g. “Will this pair of shoes suit this outfit?” – this is where a truly personalised response would come in. These questions are – for now – still best answered by a human. So if more of your transactional queries are handled by an automated system, it frees up your human agents to truly personalise the “experiential” queries.
2. The automatic responses may not be correct
Just like training a human agent, training an AI takes time. But you would not worry unnecessarily about a human getting things wrong if the training is correct.
Automation is a process, rather than something that just switches on and runs without supervision. With AI tools that are trained for specific use cases, much of it can work out of the box, but there will be tweaks to adjust things to fit a specific business.
The more you understand how AI works, the better you’ll be able to make it work for your business. Be prepared to embrace the learning process, knowing that it will pay off in the end.
3. We always want our customers to be able to speak to a human if they want
The question here is not whether customers should be able to speak to an agent, but more about WHEN they would do it. The Harvard Business Review predicted a few years back that the “companies that are successful with service automation will be those that supplement the work of technology with that of actual people”.
The best designed conversations will always build humans in as a default if the AI/automation system cannot address the issue at hand. A sign of a bad chatbot is that you can get stuck going around in circles without the option to speak to a human.
Instead, automated support systems assist your customer service team by tackling the repetitive common issues (these “transactional” queries) your customers face, whilst freeing agents to deal with the most complex situations that require human intervention.
4. The questions we receive are too complex to automate
Automation is not an “all or nothing” decision. Depending on the issue at hand, a customer may just want a quick answer, where a bot is best suited to deal with it, but for other situations they may be happier to wait for a human.
How many of your tickets are that complex? For retail businesses, most of your queries should come from customers asking about products, trying to find out where their orders are, amending orders, or trying to return them to you.
Make sure to offer every option possible to a customer, from self-service tools that a customer can use to find the answers themselves to a chatbot to an agent.
5. We don’t want to issue automatic refunds without human authorisation
Very understandable, but it depends on average order value. Think about your cost-per-ticket (to get a rough idea, think about the salaries of all your customer service agents and divide by the number of tickets they handle).
If the cost of a human handling a ticket is higher than the average order value, you are losing money on each sale. In that case it may actually pay off to issue automatic refunds when relevant.
Again, it does not have to be all or nothing. Set a limit to what automation can refund and require human approval for higher values. Refunds don’t have to be a complex issue if you have the right processes in place.
6. Tickets would not be resolved as quickly
Chatbots don’t take holidays, they don’t care for weekends and have no use for sleep in general. They are always online and thus are able to provide assistance 24/7, whatever the day it is.
For your customer, this means there is support available round-the-clock, regardless of where they are. Their needs will be met at a convenient time for them, not your company’s working hours and not your time zone.
Automated systems can at the very least respond faster than humans, but in a lot of cases they can go further and solve tickets automatically.
Humans can only talk to one customer at a time, but an automated support system is able to help several customers at the same time, being not only quicker, but also saving you significant time and money.
7. It would lead to a worse customer service experience
Is it a worse experience if say 10% of your tickets can be resolved automatically without human involvement, and the other 90% gets quickly checked over by an automatic system before being passed to a human anyway?
Assuming that your human agents always have a backlog of tickets to deal with, it’s surely no worse? The 10% gets answers faster and does not clog up the system. Now imagine if that figure grows and grows.
On top of this, automation can actually help the employee experience. Employees don’t have to deal with the same repetitive questions day in and day out, but can instead use their brains to answer the questions that really matter.
How can automation improve my customer service?
We’re glad you asked. Let’s start by analysing your tickets to find out what impact automation could have on your business.
What’s stopping your business from pursuing automation? Let us know!
