Read our second and final instalment of 2-part blog series, where we continue to focus on initial steps you and your business must take to successfully transition from traditional to NextGen IVR.
In our recent blog, we discussed how to decide which customer processes to automate. Now that you’ve followed the 4 steps and done the preparation, you know which customer journeys will benefit most from automation.
So what’s next? Building the automation into those processes.
And that’s what I’ll explain in this article.
I find it’s much easier to use an example rather than speak in the abstract, so let’s look a common scenario – a retail company introducing automation so customers can self-serve to find out their order status. I’ll focus on inbound calls here, but the lessons are relevant to other customer communications across sectors.
Step 1: Centralise data relevant to the customer experience
We’re looking at automation around inbound calls, which means it involves the retailer’s IVR system. To deliver a positive self-service experience, the system needs access to comprehensive data in real-time. This includes:
Centralising data doesn’t mean overhauling all your back-end systems. Rather, it means having a solution that sits on top of those systems, bringing everything together in a single view for the IVR to draw on. If you take a piecemeal approach to data – or it’s not available in real time – automation won’t offer the quality experience you want to deliver. You’re then likely to see increased call volumes because people need clarification after failing to self-serve.
Step 2: Set up the automation platform (like Natural Language Processing)
Natural Language Processing (NLP) is artificial intelligence that allows technology to understand people using normal conversation and writing. The aim in our example is to allow callers to self-serve, interacting with the IVR system as if it were a human agent. Now we have access to the single view of data, we can set up the NLP platform so it automates what’s required to achieve that.
For the conversational element to work effectively, you need schemas – NLP automatically learns and adds to the library ongoing, but the foundation needs to be in place. And this means understanding how people are actually asking questions. So look at the questions people ask on the website and social media. Get chat transcripts. Listen to call recordings (you can convert recordings to text to add to the library). That way the NLP will be able to automate effectively from go-live.
Step 3: Connect channels to the automation platform
In this example, NLP is the platform and voice is the channel (since we’re looking at automation for an inbound call process).
As long as the NLP solution is channel agnostic (and we always recommend choosing one that is), you can roll it out to other channels as required. This means the same platform can drive SMS, email, push notifications, WhatsApp, Facebook Messenger, RCS and more.
There are clear benefits to this.
For one, it’s easy to add channels over time because the NLP already has its library ready to go. And the channels all feed into the single library, so you get exponential improvements in effectiveness as it learns from more interactions.
You can also be more flexible and proactive with your channel management. Your solution should remember previous interactions across channels, so it can draw on the customer’s history no matter how they’re communicating with you. If the call lines are busy, you can then easily offer the option to switch channel, so they can seamlessly self-serve for faster resolution. And if you see people regularly checking order status after a specific interval, you can send proactive texts to pre-empt queries. All this contributes to a better customer experience – and reduced inbound call volumes.
Step 4: Monitor successes and bottlenecks for continuous improvement
Don’t just leave the automation once it’s gone live. You wouldn’t stop training agents, and in a similar way, you need to look after the solution to ensure it’s delivering the best experience. There should be a continuous process of review and optimisation.
The Engage Hub Customer Journey Tracker makes this easy – it incorporates data that goes through the Engage Hub platform as well as data from third-party systems (that single view of data we talked about in Step 1). Then it visualises customer journeys so you can see what happens at each touchpoint.
Here’s the example for the “Where is my order?” query for a retailer, covering a week of inbound calls:
Paracel data from the logistics platform – which has to be orchestrated into an easy format.
Start with the automation quick wins – then build up to full digital transformation
This phased approach shows how to get results from automation, building on successes to achieve quantifiable improvements in customer satisfaction and operational efficiency.
For another example of how this worked in practice (with award-winning results), read this story about MBNA’s experience automating fraud prevention processes or check our latest whitepaper The Importance of Connected Customer Journeys for more insights.