Balancing efficiency and quality: Freshworks’ customer experience advisor
In the latest instalment of our podcast series, MYC'D UP with Tech Leaders, we speak to Colin Crowley of Freshworks about the right and wrong ways for companies to use chatbots - and why ChatGPT won't be replacing humans.
In the latest episode of MYC'D UP with Tech Leaders, we speak with Colin Crowley, customer experience advisor at Freshworks.
Prior to his most recent role, Colin held a number of senior customer-focused positions across an array of companies in a customer experience career that has spanned over a decade.
This has provided Colin with not only a wealth of experience, but also a fabulous insight into almost every step of the CX chain – from customer loyalty manager to senior vice president of customer experience.
With a focus on the chatbot arena, Colin unpacks some of the major pitfalls that prevent companies from fully maximising the potential of the AI technology, as well as speaking candidly about the continued importance of human agents in a customer service environment where expectations have never been higher.
The great thing about chatbots and the conversational space, is that now quality and efficiency are no longer at loggerheads – you can have high quality and high efficiency at the same time.
MYCUSTOMER: Hello. And first of all, I'd just like to thank Colin for joining us today. So I'm just going to jump in with our first question. In a recent FreshWorks consumer study, you found that nearly all customers – I think it was 95% – expect service issues to be resolved in under 30 minutes with a chatbot. Which doesn't sound particularly unreasonable. What do you think are the kinds of issues that can cause chatbots to fail to deliver a quick resolution?
COLIN CROWLEY: Yeah, that's a great question, and obviously very topical, as a lot of companies look at using chatbots to scale their support operations. I think one thing is kind of sticking to two basics of a bot. So you have these things called ‘answer bots’, which are essentially chatbots that just fetch FAQ responses. And sometimes that could be representing those in the chat window itself, or that could be linking off to an FAQ article. And that's a great place to start when you're trying to increment improvements with your chatbot adoption and tiptoe into the space and kind of test the waters and see what resonates and what doesn't, and what you can answer for customers in what you can't.
So it's great to start with that. But as a longer-term strategy, that doesn't really work as well, because it's just a much more clunky environment for customers where you're very limited in terms of the types of answers you can provide if you're just linking people off to the FAQ, and involves extra clicks for customers, and so on and so forth. So sometimes people just kind of stick to basic answer bots, and don't really think outside of that box, so to speak. And as a result, they don't really get the return they expect on chatbots.
So that's one thing that you have to keep pushing forward and innovating, and making sure that you're not just settling for an answer bot, as opposed to truly a chatbot. With a chatbot you have more built-in conversational flows with yes/no questions and things of that nature, that allows you to get more targeted answers to customers that are more relevant to them. I’d say another thing is just bad training and chatbot programming. So a lot of the success of chatbots can depend on very basic things – it can even depend on the phrasing you use. I'll use an example in my experience, because I spent about 15 years heading customer experience organisations and customer support teams prior to joining Freshworks in the software space. And I remember in one prior example, very, very common that a lot of companies that ship products have a lot of customers asking about the delivery of the product. So when am I going to get my order or my delivery or what have you. And so I worked for a food service, where we shipped boxes of meals to people. And we had these great FAQ articles that kind of answered people's questions about the delivery of their order. And we noticed that the Chatbot wasn't returning the appropriate articles, even though we had all these great articles. And it ended up just being a phrasing issue, because customers were referring to these deliveries as boxes. So they were asking: “well, where's my box?”, “where's my order?”, “where's my delivery?”. So we were using internal lingo and that small thing prevented the chatbot from really understanding the customer intent.
So the smallest things like that, like phrasing, if you're not perfecting those and not testing your chatbot to make sure you understand if it's appropriately connecting to either articles you have available, or appropriately connecting to certain conversational flows, that can make a big difference. And that's really about making sure you're constantly training and iterating on the bot and getting qualitative feedback about the chatbot experience. So if you do that you can put yourself in a really good spot.
The last thing I'll say is that you can have challenges where people try to overuse chatbots. So when there's a case where something clearly needs to go to a human being because it's far too complex and requires human empathy, and a chatbot really can't solve the problem. That's another case where I think we've all had cases in our own lives where chatbots fall flat. So that's really the importance of understanding and defining human value in conversations. So really understanding where do human beings bring in an important component like an X factor, if you will, that makes a difference.
MYCUSTOMER: Within the same survey, it revealed that over half of consumers said they expected better customer service since they're paying more for everything due to inflation. With chatbots having something of a chequered reputation with customers, is this potentially an area where there could be a quick win in terms of improving service level?
COLIN CROWLEY: Absolutely, yes. So I think that chatbots provide a tremendous opportunity to improve service levels. And I think the emphasis is on the word 'chat' too, because chatbots are inherently paired with a chat as a communication channel. And more broadly speaking, when you consider a conversational engagement, which includes channels in the messaging space as well, such as WhatsApp and SMS and Facebook Messenger, and Apple Business Chat, and Google Business Messenger. So chat and messaging channels combined are where chatbots really flourish. And when you combine chatbots with those conversational channels, you really see the key problem in customer support for decades and decades and decades. And the problem that's really given customer support a bad name in some areas, and that's the balance between quality and efficiency. So when you think about like the olden days – quote, unquote – which was only just really 10 plus years ago, in many respects, where you really had call centres as opposed to contact centres, and you had this balance between wanting to be efficient in your operations, but at the same time wanting to deliver quality and those two things were always at loggerheads. So as a customer support leader, you were counting how many calls your agents were answering in a given hour, and how long their talk times were, and things of that nature. So all these efficiency based calculations. And if you wanted to spend more time with customers, it was directly at loggerheads with those efficiency goals. And typically, because it's easier to put an ROI behind efficiency, you would sacrifice quality for efficiency, inevitably, even if you had good intentions.
So that's where you have a world where people got used to talking to someone on the phone, and that person kind of rushes them off the phone because they have all these KPIs they're concerned about. But the great thing about chatbots and the conversational space, is the fact that now quality and efficiency are no longer at loggerheads – you can have high quality and high efficiency at the same time. And chatbots are a big component of this. So you have high efficiency in these channels, because you have chatbots being able to resolve low level or tier one issues. So you're making better use of your agents to resolve more complex issues and engage with more complex issues. You're then syphoning off and deflecting a lot of contacts from that. And at the same time, you have your agents able to handle multiple conversations simultaneously now, which is a huge game changer where the agents can handle 2, 3, 4, 5 conversations at the same time. So you have efficiency there. But you also have quality because chatbots, if they're handled correctly, can deliver fast answers to basic questions, which are closely correlated with high satisfaction.
So anything in a chat or messaging environment typically has a high correlation between high satisfaction and responsiveness. And chatbots are a great way to get what you call zero contact resolution where you can get an answer to something within seconds, you don't have to wait for an agent. So that's a more quality experience for customers. And as chatbots become more advanced, they're also able to deliver more quality answers. And in addition, you have chatbot technology that faces the agent as well. So it's not just all customer facing, but you have now agent facing bots as well, that assist agents in real time as they're engaging with customers during those agent handover times, and actually able to provide better experiences at scale. So this is all because of that nexus between conversational engagement and chat bots. And it's really solving that key problem at the centre of customer support, and allowing quality and efficiency to increase at the same time. And again, that's been a huge game changer in the customer support world.
MYCUSTOMER: So ChatGPT has obviously been a big story in the news recently, and people are very excited about AI again. When it comes to the potential for AI in customer service, what should be the ceiling for our expectations? Do you think we can ever expect more human than human interactions with a chatbot?
COLIN CROWLEY: No, I don't think we're ever going to get there at all. No, no, I think very much the idea that human beings, it kind of reminds me of the famous Mark Twain quote when there was a newspaper that kind of published his obituary, and he wrote in that reports of his death have been greatly exaggerated. And I think you see a similar sort of thing with the human factor, where every time there's a new technology that comes around, people wonder if human beings are being deprecated in some way. So no, I don't think that's ever going to take place because there are just too many unique characteristics that human beings bring to every conversation, especially when it comes to emotional intelligence and empathy and so forth. And we're in a world where more and more customers have higher expectations when it comes to quality. So the need for that is increasing, not decreasing. So no, there's always going to be that need for human engagement and chatbots are never going to be able to be more human than human beings because of that.
And I think it's also worthwhile to pinpoint the fact that chatbots really, at the end of the day, only go so far, and many companies actually use chatbots very effectively to solve very basic issues. So you think of people doing things like automating their scheduling process, let's say if you’re a primary care provider, the amount of efficiencies you can get just from tying a chatbot into your scheduling system to enable people to make appointments and change appointments and cancel appointments is huge. Just in that alone, that's a relatively basic use of chatbots. So a lot of the big wins with chatbots are in that basic area. And to the extent it's more complex when you're bring in an API, as you're tracking your chatbots, your order management system, and empowering customers to make changes to their delivery address, or to change something about their order, or to cancel their account or cancel their subscription and those sorts of things. So yeah, most chatbots still play in that space, you get tremendous wins just from those spaces, even before having to do anything that's much fancier.
And the other thing is, I'll just also reiterate this x factor again. So that's also another key area that we always have to bear in mind. And again, I think it's important that companies define what that x factor is for them. And sometimes it's just something that you have to do with trial and error: a&b tests, qualitative deep dives into each conversation. So the chats that your agents are having with customers, and you can easily identify, they're the instances where your agents are bringing in something to the equation that a chatbot would not be able to provide. So they're able to bring in a human anecdote, as an example that maybe made a difference. So they were able to pivot a conversation away from a basic issue and turn it into a sale. So I would encourage companies to look at your actual chat transcripts and look at how many of those are basic questions where a customer comes in and says: here's my question,and the agent gives an answer and you say, we’ll check, that could be a great chatbot solution, versus those occasions where human beings are doing more, and the conversation is above and beyond simple question and answer. And that helps you to identify how your human beings are being used in your organisation in a really valuable way, where you may be able to say: how can I double down on that value, and make sure that even if that issue is repeatable enough and simple enough that it would normally be a good chatbot solution, maybe because there's that human factor at play, I should be more conscious and more careful about trying to replace the human being with the chatbot. So involving that kind of incremental approach to things and being very thoughtful about how you define that x factor. And that helps you to understand the boundaries of where things should exist with how you can use chatbots in your organisation.
MYCUSTOMER: I think actually, what you're just saying there leads on well to our next question. You're talking about basically finding the right tools for chatbots not overusing, and not underusing, just finding the right area. Would you say that it's fair to say that a lot of the criticisms of chatbots are less about the technology itself and more about its application? What is the best application of chatbots and their most appropriate place in the customer journey?
COLIN CROWLEY: Yes, I would definitely say that's true. So it is in the application. And it is always true of different technologies anyway, where there's a bit of a pendulum swing involved, where if something new comes down the pike, and the pendulum swings in one direction, and everyone's focused on adopting whatever that new technology is, and then ends up being over adopted or adopted in ways that perhaps aren't as careful as they should be. And then that creates a bad experience. And then the pendulum swings back too far in the other direction. So I think we've seen that with chatbots, the pendulum has swung, and now we were at a point where it's like more right there in the middle. As more and more people appreciate the fact that chatbots have a place. And it's just a matter of defining that place more intelligently and making sure you have appropriate resources to define what that place is for your organisation. And sometimes that's really wrapped up in having a great provider with a native chatbot solution. And other times, it's also about making sure that you have a supporting infrastructure available either through your software provider. For example, we do a lot of onboarding and coaching at FreshWorks for bots, because we know it's a difficult area for customers, but it’s so important to make sure you have that institutional support to understand where that pendulum is in the centre.
But yeah, I would say it's all really in the application, and that there are a lot of great uses for chatbots. And we've already mentioned some of them. One of them, again, is solving basic single response questions. So those types of things are great for chatbots especially because you can be more targeted in how you respond. If you look at an FAQ article, which will be large and clunky, and have multiple answers within that one article. Whereas in chatbots, you can build basic conversational trees and even use NLP with your chatbot where it can understand more specific answers to questions. Another thing is, chatbots are great for accessing information for customers. So sometimes customers just want to get their tracking information or they want to see when their order is going to be delivered. So that's an easy thing that you can tie in your order management system and the chatbot can represent that information to customers. Chat bots are great for gathering information. So sometimes you're in businesses where you routinely have to collect certain pieces of information, let's say in the financial space, if you have a banking company and someone wants to submit a chargeback. So there's a bunch of basic questions you have to ask or maybe basic security questions. And chatbots can gather that information so your agents aren't spent doing it. And then there's the agent handoff, so your agents are reserved for more emotionally advanced conversations. Also you have any basic issues linked to an order management system like I mentioned before, such as a change or order cancellation or subscription cancellation, great examples of chatbot usage – purchasing, as well. So more and more people now are kind of using chatbots to notice people who experienced purchase paralysis when they're surfing a website – where they're loitering on a certain page too long. And a chatbot can kind of pop up some helpful FAQ questions that may be targeted based on the page that they're on to help answer questions about your shipping processes or your return process, or maybe offer some information about the products or services that you're offering.
Lastly, I'll throw in that internally, chatbots are really useful for routing purposes. So more and more now, customers want a very, very specific service. So if you contact someone about a delivery issue, you want to go to a person who's a delivery specialist who may be more empowered to resolve your issue than someone in a general pool of agents. And it can be very difficult sometimes to understand which customers should go to which agents. So chat bots, and answering a few basic questions through chatbots, can be a great way where you can tie that into a routing mechanism where you're making sure that those customers are being sent off to agents who are more capable of helping them for their specific issue and more capable of resolving that contact, and in a single contact routing purposes on the back end is also a great way to use chatbots.
MYCUSTOMER: Just on to our final question now. And this is a question that we will be asking all of our guests on MYC’D UP with Tech Leaders. If there was one particular tool or solution or product type that you think businesses should be investing in that isn't provided by your own organisation, what would it be? And why?
COLIN CROWLEY: Yes, that's a really interesting question because technology really defines the boundaries with any leader but especially if the customer support leader has to operate and realise their vision for their team. So I would say one big thing that we're not invested in currently at Freshworks. And we use different partners for this, but quality assurance technology. So I mentioned already that quality is increasingly important and customer support, because again, customers have higher expectations of quality. And especially because many companies have done a good job producing higher quality customer support. So that creates kind of like a larger ecosystem that informs customers expectations of your company, no matter what stage of development you may be in. So customer expectations are rising and more and more we’re seeing that quality is truly a differentiator in customer support.
It's hard to do quality well. So efficiency is easier to do, because of all these new technologies like conversational engagement, chatbots, and so forth. But quality is hard. It's very hard to understand the ROI under quality. And it's very hard to understand how you measure quality, it's very hard to understand how you make quality tangible and objective as opposed to subjective. And for that, you really need an infrastructure. So quality assurance programmes are a magnificent example of that. And they kind of run the gamut for AI power programmes that are able to look at patterns and certain contacts and highlight certain contacts and which ones should be monitored on a more discreet basis. So you can kind of look for contacts across thousands and thousands and thousands that you really need to look at for some reason. And also software that empowers you to build a quality rubric that's linked in turn to a scoring system, which you can use to evaluate your teams and your agents and your leaders as well and really hold people accountable for quality. So quality assurance software, I'd say is a really, really big thing that would like compliment your use of our fleet at Freshworks as an example, and really empower a customer support leader to have differentiated customer support.