If you’re part of an overwhelmed customer service team dealing with a mountain of tickets, then ticket deflection sounds pretty appealing.
A ticket that has been deflected is a ticket that you don’t have to deal with, so the more you can “deflect” the better, right?
Well, we think deflection is the wrong thing to measure when it comes to chatbots, virtual assistants and customer service automation in general. That’s not to say it’s not valid on some level (which we’ll get in to), but at best it should be considered a goal and not a metric.
Here is our reasoning, and you can make up your own mind.
First, what is Ticket Deflection?
Ticket deflection is the name given to a range of tactics that a customer service team can deploy to try and decrease the number of tickets they deal with by allowing customers to self-serve.
The main ways companies do this is by having FAQ pages, knowledge bases, portals, or interactive chatbots that allow customers to solve their issue without speaking to a human.
The idea being that the easier you make it to find answers, the more your customers can help themselves, and the less your customer service agents have to do.
Problem 1: How can you measure something that doesn’t happen?
A ticket that is “deflected” is a ticket that didn’t get created. So there is no way of knowing, for sure, that the ticket would have happened.
If someone visits your FAQs but doesn’t then raise a ticket, do you know for sure that they found the answer they were looking for? You can see how many people looked at a page, but it’s harder to model for tickets that didn’t happen.
Take an example. If you didn’t have any information about your returns policy on your site, you’d probably get a lot of questions about it. As soon as you put a returns policy page on your site, the number of questions would drop suggesting you have “deflected” more questions.
There are some ways around measuring. For example a before and after comparison. If you were getting lots of questions about your returns policy, you could A/B test making the policy page more prominent during the buying journey. If you saw tickets related to that go down, maybe you did successfully deflect them.
Problem 2: Are you just putting obstacles in your customer’s path?
Let’s say a customer has an issue and wants to speak to a human. They go to a live chat and get presented with a number of buttons, none of which relate to their problem.
There is no option to speak to a human, so they pick an option and follow the path presented to them, and they reach a dead end. So they start again, and reach another dead end. And another.
By the time they eventually find an option to speak to someone, they are more frustrated and worked up and create a worse experience for the agent on the end of the phone or chat line. So everyone is more unhappy – the customer, the agent, and ultimately your brand suffers.
This is what bad ticket deflection leads to, sending customers down paths they don’t want to go to, or forcing them to read page after page of irrelevant information to find the snippet they do care about.
If you are targeting a low customer effort score, the more hoops you make customers jump through, the higher your CES is likely to be.
Problem 3: Should you want to “deflect” your customers anyway?
Personalisation is a goal for most retailers, and in an increasingly competitive and personalised world, why would you want to turn down an opportunity to offer a personalised customer experience?
In other words, ticket deflection means you are deflecting CUSTOMERS or potential customers away from agents. Agents who, if they are good, could be the difference between a customer buying or not buying, or a customer choosing to remain a customer or not.
Plus, you are missing out on opportunities to upsell and add value to your customers that ultimately hits your bottom line.
Worse yet, if your deflection is too strong you may end up sending your customers to your competitors because they get so frustrated with your site. The point is that once a ticket is “deflected” you have no influence over what happens, for better or worse.
Remember: ticket deflection is solving an internal problem, but if it is not creating a better experience for the customer then it’s a bad goal.
Problem 4: Deflection is not the desired outcome – resolution is
When a customer has a problem and comes to the support section on the website, ultimately you want them to walk away with their problem resolved. Whether that’s through finding the relevant information on site, using self-service, chatbots, or humans – it doesn’t matter.
So if you are trying to measure deflection as a reduction in tickets, you may not be capturing what really matters, which is ticket resolution.
Your agents can ask if the customer was able to resolve their issue, and you can ask for a score at the end of a chat, or after going through a self-service portal. It’s harder to ask with FAQs, but you can add buttons to ask if readers were able to find the answer they were looking for which at least gives you some data.
Problem 5: Chatbots should not be part of “ticket deflection”
The main problem with using ticket deflection as part of measuring chatbots is that it misunderstands what modern chatbots should be part of.
Traditional chatbots use pre-determined journeys and scripts to talk to customers, meaning that if a customer wants to stray outside of those rails then they will get stuck.
Modern chatbots can be an essential part of the customer service team. Using natural language processing they can decipher what a customer is asking and direct them to find the right outcome.
This involves triaging tickets into ones that the chatbot can solve, and ones that are best left for customer service agents. In this way they are more like virtual assistants, taking some of the work off agents’ plates.
Even more advanced ones use deep integrations with carriers, order management systems, warehouses etc. to be able to do much of the work that agents would have to do. The advantage is that they can do it round the clock, and in seconds.
Plus, rather than just directing customers to a relevant FAQ page, modern chatbots can use generative AI to find and summarise the relevant information for customers there and then.
Where ticket deflection is important
All of this is not to suggest that you should not disable all ticket deflection tactics and bin any chance at self-service. But you should be able to trust your customers to know which questions are easier for them to find themselves vs. asking a human.
Many people will check your FAQs before trying to contact you because it’s likely to be quicker. You can monitor which FAQ pages are most commonly used and make them as easily accessible as possible.
You can even put some of these FAQ options as button options in a chatbot, but you should never force people down one of these routes. You will catch a few people who went to your chat button first, but there should always be an option to talk to a human readily available.
The same goes for return label generating. If it’s possible for your customers to easily do it in a portal, then you should pursue that.
At DigitalGenius we believe that customer service agents should have fewer repetitive tickets – so on that level we agree with ticket deflection.
But we believe that self-service and automation should be about helping customers reach resolutions faster, for the ticket types that can be automated and solved. By clearing out these tickets from the backlog, it means agents can spend more time on the tickets they do receive and provide value to customers.
We also believe that chatbots and virtual assistants are about doing more than just answering questions, but actually doing the legwork and finding out where a package is, offering solutions, giving customers choices, and processing returns, replacements and refunds.
All of which in a way “deflects” tickets, but more importantly resolves them. And it does it faster than humans could possibly do it, and round the clock, meaning that your customers receive a much better service.
So what other ways can you look to reduce ticket volume, here are some quick tips:
Look at what your customers want to know. You have FAQs, but are they the questions customers are most frequently asking? You have to dig into the data to do that. DigitalGenius can analyse your tickets for free using AI and tell you the most common questions you are getting. Sign up for it here.
Surface answers to common questions in the buying journey. You can put delivery, returns, and refund information on product details pages, or on the basket page to keep customers on the right track.
Leverage generative AI to summarise FAQs. Sending customers to a long FAQ page is not a great experience. Being able to pull out and summarise the key information is much better. Watch our webinar to see how Air Up do that with generative AI.
Look at what tickets take the most time, and look to automate them. Sports brand on realised that generating return labels took their agents 10-15 minutes per label. So they looked to automate it, and now it can be done in seconds, saving their team 800 hours a month.
