Open and Closed approaches to speech design – what’s the difference?

Before starting to develop a speech application, it’s important to first understand the nature of the type of customer process you’re dealing with if you’re to secure strong levels of customer acceptance. Inevitably this revolves around the kind of relationship that exists with the customer – is it a closed relationship like a person might have with their bank, or is it a more open scenario, an example here would be between a train passenger and the train company.

Both of these kind of relationships can benefit from voice self-service , however it’s important that you identify which one you’re working with before trying to address the opportunity with a speech-enabled solution.

When a customer already has a ‘closed’ relationship with an organisation such as a bank, the type and depth of engagement is much deeper. When a bank writes or calls you, as a customer you already have a vested interest in that interaction. There’s already an ongoing dialogue in place – if it’s online or phone banking you’ll already be making regular calls or using their website, or with more traditional accounts you may know staff at a bank or receive regular letters and statements.

For banks, insurance companies or other types of organisations managing closed customer relationships, there’s an ongoing imperative to acknowledge the value and commitment already built up in these customer engagements, and the need for any speech-enabled customer solutions to deliver real added value for these customers. This means its important to automate the right tasks and to understand what the intended customer benefits will be right from the start of the project.

It’s also essential to make sure that you bring your customers with you on your journey towards the delivery of speech-enabled services. Banks, for example, already have a range of channels that they use to communicate and interact with their customers, and these need to be optimised in order to maximise take-up of any new services.

Direct one-to-one marketing can play a key role here, with service flyers included with statements, the use of websites and even TV advertising. Another opportunity to support customers is with an Agent Assist service, where should customers using an IVR or speech solution hit difficulties they simply hit a button and are connected to a live agent who can lead them through the service. This requires close integration between automated and traditional contact centre services, but is an approach that’s already working well with some leading financial institutions.

By deploying effective customer ID processes you can also take account of each customer’s competence and/or willingness to work with speech-enabled services, and provide an appropriate level of support. If a customer needed help last time, you can make sure the solution asks whether they need and help this time or if they’re happy to work with the system themselves.

In a closed relationship it’s also possible to implement Demo capabilities that allow customers to play with the system and work out how it can be helpful for their own usage. This level of involvement recognises that these are very important transactions for customers, and they need to build up their own confidence in the system.

Dealing with less-defined customer relationships

For open customer relationships the challenge is quite different. Customers might not be account-based, so there’s no customer record to draw on, but for the caller urgently needing to find a train time, or the status of a flight, the interaction can be equally important.

This can make the design of speech applications more complex, so it’s essential to always bear in mind what it is the customer is trying to achieve. Any application design needs to first determine the caller’s objective, then map that objective to a process, and also feedback to the caller that their objective has been clearly understood. Callers should also be told of the benefits to them of completing the task using the speech application.

Having a less prescriptive speech dialogue at the start of a call can help, which allows both the customer and a speech solution to focus in on the exact nature of the call.

In the utilities market, for example, one service provider has found that it inevitably helps to ask their customers for a meter reading. The company found that when customers called in, they didn’t want to say they had a problem with their bills. However, finding out the current meter reading often gave the utility the information they needed to determine whether a customer had received a bill based on an incorrect reading or estimate – and meant that the speech system could address the issue with the customer.

It’s clear that standard approaches to speech systems design won’t always be applicable in these situations. While the technology may be in place, there’s still a real requirement for solutions that have been designed around the realities of the customer engagement, and offer the right automation approach to deliver an appropriate and personal customer service.