CS AI best practices
Back to Articles

Best Practices for Automating Customer Service

Improve your customer experience with automation tips from the experts.

September 25th, 2017


Becca Le Blond

Board rooms are abuzz with the possibility of automation. It promises cost-savings and improved customer service (CS). If you haven’t been asked 'what’s your strategy for automating our CS?', you soon will be. We’re sharing tips we’ve developed through shaping automated CS solutions for our clients, along with best practices other CS leaders are advocating.

Benefits of automating customer service

We’re really excited for the potential of automation and artificial intelligence (AI). In fact, we’re already helping our clients use them to improve their customer experience. But, as with any potential disruption to status quo business functions, there is a temptation to stick your head in the ground and hope it doesn’t take hold. That didn’t work for Blockbuster, and it won’t work for you.

How exactly can automation improve your customer experience?


A key ingredient to a healthy bottom line is loyal customers. Returning customers spend 67% more than new customers, on average.

So how do you foster loyalty within your customer base? Personalisation. Per a Salesforce survey, 73% of UK customers say they’re likely to change to a competitor for a personalised customer service.

Automation makes personalisation possible on a large scale. No, software won’t deliver hand-written notes to every customer describing why they’re your favourite. But it can use a customer’s service history with your brand to offer customised product suggestions (hi, Amazon) and special offers. Bonus points if that offer code includes their name.

Using automation to personalise your CS will engender greater customer loyalty, saving you the need to attract new customers at 4-10x the price.


This, alongside cost reduction (that’s up next), tends to be the headline attraction of automation. Helping customers reach great resolutions faster should be the goal of any CS automation strategy. We’re using chatbots and interaction workflows to do that.

By nature, bots will respond to messages almost instantly. Workflows route a customer to our client's chatbot at the start of the conversation. 

customers prefer personalised customer service customers prefer personalised customer service

Then, when the bot has completed its data collection mission - or determined it can’t help that customer - it will pass the interaction on to a human operator.

This process happens in a matter of seconds, and the limiting factor for efficiency becomes how quickly the customer responds. We conducted research on how much time we could save with a bot that gathers simple identity data at the start of an interaction. One of the headline results was that we can save 46 minutes for every social media conversation. Supporting human agents with data gathering AI frees up their time to deal with complex queries that AI can’t (currently) handle.

Faster response times means faster resolutions, both of which mean an increase in operator efficiency. The biggest incentive for maximising team productivity? The great potential for cost reduction.

Cost reduction

Reducing wastage within customer service is often the target when businesses look to cut costs. That used to mean preparing to sacrifice service quality to boost profit margin. Automation means that’s no longer true.

Bots have been taking over relatively simple customer service interactions, in the place of human agents. For example, Mastercard announced KAI at the end of 2016, which makes it possible for customers to make banking enquiries within messaging apps. No more need to waste time making your way to your local banking branch, or listening to an interactive voice response (IVR) system as you (hopefully) navigate your way to an answer. For businesses, these bots mean saving on wages and office costs.

And so we hit the stem of the most controversy around automation: the possibility of job losses. AI’s are going to get more intelligent over time. In fact, the role of a retail sales person is projected to be fully automated by 2030. For now, your job - and that of CS operators - is safe.

We advocate making human teams more productive, and effective, with AI-assistance. In this scenario, cost reductions are still based on needing less human resource. But you don’t need to worry about bots encountering an interaction they can’t resolve and leaving customers in an endless, unresolvable, loop.

Not convinced? This strategy can save one of our biggest clients £9.1 million a year.

We’re firm believers that the best way to improve customer satisfaction is to use a bot that enhances their experience, without losing the human touch.

Improve customer satisfaction (CSat)

automate customer self service automate customer self service

The three main drivers of customer satisfaction are response time, resolve time, and status updates.

We touched upon the capability of automation to improve response time and resolve time in ‘efficiency’ above. To give that more context for customer satisfaction, customers that have to wait more than one hour after their original contact for a resolution are twice as likely to give a low CSat score. 

That likelihood triples if the customer has to wait more than six hours. Workflows make automating status updates a breeze.

Say a customer’s order is due to arrive the following Wednesday by 3pm. Intelligent software can set a timer to check the parcel’s tracking once that date and time has passed. Then, if the parcel is delivered, it will send a follow-up message asking for feedback. If the parcel isn’t delivered, it will let the customer know your CS team are investigating why, then reassign the conversation to an operator. Bluntly put, automation makes your CS team look like they’re going the extra mile, with very little effort required.

Ultimately, the key to improving customer satisfaction with automation is to ensure that your implementation improves speed to resolution, without surrendering quality of said resolutions.

Best practices

Okay, you’re sold on automation improving your CX, but how should you start building your strategy?

We have some strong opinions on the best way to use automation in CS. What can we say? It is the future of our industry. But, in the interest of balance, we asked to hear from a range of leading voices in CS on how automation can best improve customer service. Here are our insights:

Should bots take over the role of human agents?

The main factor when making this decision is the complexity of your typical customer query. Are they easily answered with simple data gathering? If yes, a bot is likely up to the task. Banks are (generally) taking the approach of front-end bots; they can handle interactions independently of a CS team. They're able to do so because most of their contacts are simple account balance enquiries.

But if a query requires a bit more context like 'could you recommend an item to match this skirt?' or 'why hasn’t my refund come through yet?' then a bot is going to struggle. We believe in assisting human agents with bots, rather than replacing said agents completely. Mat Patterson, Customer Service Evangelist at Help Scout, advised a similar approach:

AI and machine learning are making great strides, but I expect to see their biggest impact in augmentation, more than purely automation. A good customer service agent can be empathetic and can understand another person who may struggle to explain themselves… Rather than replacing humans, let’s use the unmatched power of machines to enhance their abilities.
What would that look like? An agent who has a comprehensive view of each individual customer, but who hasn’t had to spend their own time collecting and collating the relevant information about their history with the company, their systems, and preferences.

But the possibilities don’t end there. Here’s what customer service expert, Shep Hyken, said:

At a higher level, AI is being used to assist and support a customer service representative (CSR) helping a customer. The computer 'listens' to the conversation, and, in real time, makes recommendations to the CSR, who then responds to the customer. AI is evolving… We’re not quite to the point of fluid verbal conversations with a computer, but we’re getting closer!

For now, bots simply aren’t able to take on all customer queries. We suggest augmenting your current CS teams with AI solutions that support operators and improve CX. At least until AI can replicate the humanity of human operators.

Automated FAQ’s

Intuitive automation of your FAQ section has the potential to set your brand apart. Note the ‘intuitive’ there. A wide-held frustration with searchable-FAQ systems is how difficult some make it to find an actual answer.

Simply offering an FAQ section isn’t enough. Not if you want your CS to stand out from the crowd. Fifty-five percent  of all customers find online self-service portals hard to use.

Millennial customers try to find their own solutions before contacting a business in frustration. They’re used to technology reducing the need for them to ever contact a human.

That doesn’t mean you need to ignore this section if your current customer-base is on the more-experienced side. When they have a question about products or services, 73% of all customers will search online before contacting a company directly.

How do we suggest automating your FAQ’s? Use a customer’s account history. Are they waiting on an order? They’re most likely looking for an update on their delivery, so make that section prominent when they enter the FAQ section.

Test automation regularly

Just in case it doesn’t go without saying: any automation needs to be tested regularly and thoroughly, even long after it goes live. Nothing is perfect, and the same goes for any automation you implement.

We put hundreds of hours into testing our bots under development. Chatbots rely on a natural language processing (NLP) engine to derive meaning and intent from messages. We fed ours thousands of different questions, hoping to trip it up. We then worked out how it happened, and implemented rules to stop it from ever happening again.

Dedicated teams of testers should be mystery shopping your automated support systems, looking for ways to make it stumble, even after it goes live.

That said, don’t let this put you off. Teams with experience developing bots will already know to do this for you. And, once live, you already have a great resource for testing: your operators. Their in-depth knowledge of how your automation should work puts them in a great place to spot anything amiss, and they can report if customers mention an error.

Collect customer feedback

The best way to make sure automation really is improving your CX is to ask customers for feedback. One of Customer Service Keynote Speaker Chip Bell’s top automation tips is:

Create a constant feedback pipeline to customers to get an ongoing read on how [they] react to customer automation.

If you want to hear back from a wide customer audience, incentivise feedback and offer rewards for all responses. Think a discount code, or entering their name into a prize draw. You can get even more reviews by offering incentives to your operators to gather as many as they can. Your automation method could be working exactly as you expected it to, but the only way to be really sure it’s improving your CX is to hear back from your customers themselves.

automation balances cost and customer service quality automation balances cost and customer service quality

Automation and artificial intelligence are the future of customer service, and it’s time to get your strategy ready. AI’s aren’t currently ready to handle everything a customer can throw at them, but they can offer new elements to your CX that improve satisfaction and boost productivity. 

Want to know more? Check out our page on automation, or contact us to start your journey to a new, better customer experience.


Improve your customer experience with insights from our monthly newsletter. Subscribe today to get started.

Stay Updated
Please type a valid email
Please remember to verify you are not a robot.

Close to try this form.

Please enable cookies to use this form.

Send Email