In 30 years of working in contact centres, Ed Creasey, VP of Global Solutions Engineering at Calabrio, has seen the industry through many technology introductions. The advent of AI-driven analytics is no different; except that in some respects, it kind of is. Fear not, he is here to explain:
"AI can now cater to increasingly advanced use cases, but that's also going to have unintended consequences – just like any other technology," he says.
"That said, there's also a specific set of known AI-related consequences that we can anticipate if we want to use AI-driven analytics in the contact centre."
So, what's changed, what hasn't, and how can contact centres navigate the ocean of new possibilities? Let's dive in.
What's Changed
To see what changed for contact centre analytics, we must consider the following key aspects:
Observing the AI: Most contact centres use analytics to measure agent interaction and performance – but what about bots?
"Businesses are investing in customer-facing bots, but they're not analysing that experience, which means these bots could be causing CX issues that slip under the radar," Creasey explains.
"It's essential to look at the customer journey through all touchpoints and ensure your analytics covers both agent and machines."
Data-Powered Decisions: With modern-day contact centre analytics providing AI-powered summarisation, evaluation, categorisation, and root cause analysis at scale, contact centres can then use their output for generative Business Intelligence (BI). This allows them to ask questions and draw insights at the press of a button.
"By telling your BI tool which KPI you want to look at, you immediately get automatic visualisations of the data and relevant insights, making data-driven decisions easier than ever,” Creasey says.
Cornerstone KPIs: With AI able to tell contact centres anything about their data, Creasey and Calabrio believe it's important to stop and think: What do we want to measure, and how? To do this, contact centres need to define Cornerstone KPIs to measure CX, each having its own signals.
"If it's a bad customer experience from an agent, a negative signal to look out for would be a human agent's unwillingness or inability to help," he says.
"For a bot, it could be repetition, making a customer rephrase or repeat themselves too."
What Hasn't Changed
According to Creasey, what tends to stay the same when it comes to technological innovations – in this case, AI and analytical technology – is the appearance of underlying problems.
Here are two areas requiring special attention:
Agent Experience: When setting up automated quality measures to assess agent performance or suggesting coaching, contact centres must be extra careful of potential AI bias.
"We've seen cases where AI made decisions that put different groups at a disadvantage due to being trained on insufficiently diverse data," Creasey notes.