Self-service and voice channel: 4 steps to identify use cases to automate


Is your calling customer service clogged with requests, and are you looking for solutions to reduce the flow of calls handled by your agents? Thanks to AI-powered voice agents, intelligent voice assistants, it is now possible to automate the processing of repetitive customer requests and make their experience more fluid. But where do you start a voice automation project? Let's take a closer look at the methodology to adopt in order to identify the call flows to be automated first.

To reduce the volume of calls to be handled and develop self-service, the implementation of natural language projects is an attractive solution to reduce the workload of teams, reduce costs and increase agent productivity. The Covid-19 crisis has only amplified this trend.

But how do you build the roadmap for a contact center transformation project? Despite their willingness to explore the potential of AI, many companies are hesitant about how to proceed. Several of them have even called on calldesk to conduct a diagnosis of call flows in their contact center, with a threefold objective:

  • identify the needs of callers and prioritize the call patterns to be automated
  • estimate the ROI of a voice agent project in order to present it to the management
  • create the specifications of a project adapted to their contact center
  • Too many times, projects fail in companies that identify the solution before having properly defined the problem. As promising as they are, voice agents must first and foremost serve a business need.

    In 3 days, a call flow diagnosis allows our Product and Customer Success teams to study the experience of your callers and your agents to understand their needs and propose the most appropriate next actions.

    To allow you to be inspired by this methodology, this article presents the steps to follow to identify the call flows to be automated and to establish the scope of a future voice agent project in your contact center.

    Step 1: Analyze your contact center call flows

    The first step is to analyze your contact center call flows. Indeed, delegating the treatment of an interaction to a voice agent is relevant when the intervention of the human agent is of little or no interest for the caller. Incoming call flows can therefore be classified according to their added value. This exercise is called a call flow audit.

    How to retrieve the information needed for the call flow audit?

    If necessary, workshops to prioritize the reasons for calls can be held in collaboration with our business teams to help you delimit the reasons for your incoming calls according to the company's objectives and your customers' expectations. Bringing in an outside perspective provides a clear view of your callers' customer experience on the phone channel.

    It is also possible to collect this information through meetings with the different actors of your contact center, from agents and supervisors to top management. Combining this data allows you to model your customers' call journey on the telephone channel.

    This consulting process can be reinforced by double-listening, detailed mapping of the IVR or by auditing the pre-call and post-call tasks of the agents. It is also possible to use some technical tools, for example by capturing calls in order to understand and group the reasons for calls from customers as well as the emotions that come out of the conversations.

    Finally, our business teams also perform a technical analysis of your contact center architecture. An essential step before any voice agent project, this preliminary analysis phase allows us to evaluate the constraints and rules related to the company's information system.

    Step 2: Prioritize call flows according to their value

    What criteria should be used to prioritize call flows?

    The value of calls and caller requests must be understood according to a double logic.

    On the one hand, we must analyze the value of these call flows for the company: does the call fall at a decisive moment in the commercial relationship with the caller?

    For example, an unhappy customer will tend to want to talk to a consultant to get a quick answer and be comforted. This call has very limited value to be handled by a voice agent. High value-added requests are the ones your agents should be able to focus on.

    On the other hand, you have to analyze the value of the call for the customer and choose if this value is rather low or high: does the customer always tend to want to do this operation by phone? Despite the existence of alternative channels, a customer may systematically want to exchange with the phone to perform an operation and this operation has a lot of interest to be processed by a voice agent. On the other hand, if the call has only a small interest, both for you and for your customer, the objective will be to eliminate this flow rather than to automate it.

    For example, if many customers call you because they don't understand their invoice, it's better to work on making it clearer: you will treat the root of the problem, and thus be able to reduce the number of incoming requests!

    Step 3: Decide on the treatment of the different call flows

    The implementation of a voice agent solution must satisfy a need to be effective. Once you have identified all the different reasons for your customers' calls according to their added value, you need to decide how to deal with these flows. Because, even if it were possible, it is not recommended to automate all calls!

    In the context of a diagnosis carried out by calldesk, once the field mission is over, we proceed to the detailed analysis of the data collected and to the creation of a report, for a presentation to your teams.

    The final result is the creation of a call automation matrix, where the use cases are divided into 4 main categories:

    • Polluting calls to be dissuaded
    • Excessively complex paths to be simplified
    • High value-added calls to be exploited
    • Unproductive calls to be automated


    This automation matrix will define the perimeter of automatable calls and voice agents to be deployed.

    Step 4: What criteria to use when selecting the use cases to automate

    Once the matrix has been established, a qualitative analysis of the various use cases eligible for automation allows us to prioritize them and select the most relevant ones to automate.

    This includes:

    • verify the pertinence to your customer experience
    • evaluate the technical feasibility of the project
    • and estimate its operational impact in order to calculate the ROI.

    First of all, regarding the customer experience, it must remain simple and relevant from start to finish. A voice agent relies on a chat script, and the goal is for it to be tailored to your customer's need. If you don't have a simple script to solve your customer's problem, or if unforeseen events often occur in conversations, perhaps it's best to leave this use case to your agents for the time being, as they are better able to adapt to specific requests.

    Next, you need to evaluate the technical applicability of the project, both from an IS and technological point of view. Integrating a voice agent with the company's IS (CRM, telephony tool, etc.) can be time consuming. At this stage, discussing with the IT department in order to understand their technical constraints can eliminate certain cases of use.

    More generally, we advise you to start by identifying a use case that requires little or no integration, so that the bot can be deployed in just a few days. This will allow you to test the effectiveness of the solution, while having invested a minimum of resources upstream.

    Finally, you must evaluate the operational impact of the voice agent implementation and calculate the ROI of the operation. As each use case is associated with a certain volume of calls and an average processing time, you can evaluate the time required by your agents to process each of these requests. It is therefore possible to quantify the financial value of each use case, and to evaluate the relevance of implementing a conversational voice agent that would reduce the cost of call processing. The ideal is to analyze the financial result of the operation if the voice agent is able to absorb 70, 80 or 90% of the flow, in order to have several working hypotheses.

    Once completed, the diagnosis of your call flows is a valuable tool to help you decide on the implementation of a voice agent project. Whether you do it yourself or call on calldesk, this diagnostic allows you to prioritize the use cases to be automated in your contact center.

    In the case of a diagnosis conducted by calldesk teams, this short 3-day mission allows you to develop a clear roadmap for voice channel automation. At the end of the diagnostic, you will have a clear picture of whether or not to implement a voice agent, and which use cases to deploy first.

    If your company decides to launch a voice agent project, it should also be noted that the project will have a better chance of receiving the support of the contact center field teams, because they will have been included in this diagnostic process from the beginning.

    Would you like to benefit from the support of one of our experts to determine if one of your use cases can be automated? Contact us via this page, and we will be happy to discuss your use cases. See you soon on calldesk!

    Subscribe to our newsletter!

    One email per month dedicated to customer service. Unsubscribe at any time.