In the context of customer loyalty programs, contact centres
play a huge role in the customer experience. Tens of
thousands of calls can come in each month from customers
asking about anything from their points balances to resetting
passwords to product availability.
Much like in sport, where the saying goes that you’re only as good as your last match, customers’ satisfaction is often only as good as their last experience. Literally every call matters, with the potential to make or break a customer’s relationship
with a brand.
So the role of quality assurance, and its noble intent of continuous improvement, is crucial. The rise and rise of AI highlights
opportunities for just how far that improvement can go.
The current state of play in contact centre QA
Here’s some of what makes up the role of quality assurers in a contact centre: they must listen into calls or recordings,
transcribe some or all of those calls and, based on their training, look for and address issues they might find in those calls.
That’s no reflection on the people doing the quality assurance – they’re doing essential and positive work within the bounds of human limitations. But as Richard Cramer, Loyalty Director at Achievement Awards Group, points out: “We’re always looking for ways to kind of ‘QA the QA.’ And most improvements are incremental. When something comes along that can help you 10X that improvement, you have to pay attention.”
What’s possible when you incorporate artificial intelligence and data into the process?
For starters, AI can transcribe all 25,000 of those calls practically instantly. So your sample size goes from a tiny percent to 100%. You’re QAing every, single call.
Then AI can look for specific things in those calls and transcriptions. Can it do as good a job as humans? Arguably it can do it better. AI can be trained to look for certain words. For example, are callers mentioning the names of competitor brands? That’s often a sign of dissatisfaction.
AI can also be taught to pick up other things. For example, are callers angry? AI can pick out swear words, raised voice volume, high speech speed – each of which may be indicators of callers’ agitation. By the same token, AI can be trained to look for slow speech or low vocal tone in contact centre agents, which may indicate that they’re unfamiliar with subject matter, or lacking in confidence. It could even do this in real time, alerting a manager, who could intervene appropriately if necessary.
It’s not just red flags that AI can find; it’s also green lights. What are contact centre agents doing well? Can we recognise excellence that we might otherwise have missed? Absolutely, yes, because you’re not missing 99% of calls.
That’s all at an individual call level. At a macro level, it’s obvious that AI will do a far superior job looking at full and aggregated data, picking out patterns from all 25,000 calls.
For example, it’ll quickly pick up if lots of people are phoning because they can’t reset their passwords, which could be a cue to look at the user interface, where a relatively simple change could improve the customer experience and reduce pressure on the contact centre. Or, it might be that lots of customers are enquiring about the availability of a specific product, which might be a cue to reach out to another brand to explore a mutually beneficial partnership. AI can pick these patterns up in a relative blink of the eye, indicating opportunities to improve and incorporate into training, as well as to recognise who and what’s working really well.
One last example of AI’s advantage: it never forgets. Any learning at any time is added to all of its learnings over time. So while humans come and go, need to be trained (and sometimes re-trained), and largely only have a view of things from when they start in the role, AI maintains decades of collective knowledge. Practically, this means it could pick up correlations that people wouldn’t. A simple example: are patterns or behaviours shifting over time? Perhaps customer are phoning contact centres more often than they used to, or are doing so more at different times of day or on certain days of the week.
From menial to meaningful
A common concern with AI is that if it can do humans’ jobs, it will take humans’ jobs. But that isn’t the case here.
Consider, again, the roles and responsibilities of that individual working in QA. In an eight-hour work day, most of those hours might be taken up listening to and transcribing calls. That leaves only a few hours to act on the information and influence positive changes, which is where the value of QA really lies.
With AI doing the more menial work of listening, transcribing and finding, the quality assurer is freed up to do the much more meaningful work of human performance, dedicating all eight hours to supporting contact centre agents and continuous improvement.
Additionally, the admin side of that work can be automated. As Johan Hechter, Business Process Consultant at Achievement Awards Group points out:
“Consider how long it might take to write 10 emails individually, telling agents that issue, x, y or z was picked up in call a, b or c, and that as a result action p, q or r will need to happen. And consider that an inadvertently poor word choice may mean that some of those emails risk falling foul of HR or legal policy. AI can be trained to write those mails in a way that is appropriate and correct – again, in a milli-fraction of the time. Plus, it can file the emails together with the relevant call recordings and transcripts, so anyone needing to access them down the line can do so in a few clicks.”
“Consider how long it might take to write 10 emails individually, telling agents that issue, x, y or z was picked up in call a, b or c, and that as a result action p, q or r will need to happen. And consider that an inadvertently poor word choice may mean that some of those emails risk falling foul of HR or legal policy. AI can be trained to write those mails in a way that is appropriate and correct – again, in a milli-fraction of the time. Plus, it can file the emails together with the relevant call recordings and transcripts, so anyone needing to access them down the line can do so in a few clicks.”
