Introducing Call Control: An A.I Call Centre Sentiment Analysis Module from Panda

As promised, Panda is back with a critical new addition to our line of brokerage intelligence products! Over the past few years, we’ve been very interested in how A.I systems can be harnessed to help brokerage teams work smarter and more efficiently. 

Two examples of such efforts are Panda’s Next Call A.I and Document Verification modules for our flagship CRM product. Each, in its own way, works to provide insights, identify potential issues, and prioritise tasks, helping brokerage team members perform their duties more effectively.

Our newest addition to this growing family of smart systems for brokers is called Call Control, an innovative client sentiment module that helps brokerages make the very most of every incoming call fielded by their operators. 

Call Control increases efficiency and reduces reliance on manpower. It helps brokerage management teams improve the quality of service provided by their call centre agents by recording calls and providing managers with actionable insights into how their agents handle customer calls. The system also helps brokerage teams measure the performance of various marketing campaigns, as well as performing market research, by listening out for customisable keywords used by customers during calls.

The content of these calls is a valuable source of customer insight that can be mined to help businesses better understand their clients’ needs and deliver superior support, as well as identifying potential sales opportunities.

The central idea that unites all of our A.I products, is that a modern brokerage is an abundant source of highly actionable data. These data usually slip between the cracks of outdated brokerage systems that don’t have the capacity to store, process, and, most importantly, make sense of them.

How it works

In the case of Call Control, the system analyses recordings of customer interactions that are longer than 2 minutes in duration. Then, with the help of Google’s transcription algorithm, these recordings are turned into text. Finally, the content of these transcriptions is processed by a Natural Language Processing algorithm that scores the interaction along a series of metrics, for instance, assigning the individual words used in the call a percentage score along the scale of Positive, Neutral, and Negative.

Below you can see how the AI evaluates the content of a specific call, determining the general sentiment of the call to be 50% neutral, 50% positive and 0% negative.

The system also provides a number of other useful metrics by which to score customer interactions including: talk to listen ratio, patience average, as well as flagging when support and sales reps talk for too long (representative long monologue), or when a client has divulged a lot of uninterrupted information to the rep in question (client long story). Furthermore, the system also informs the user whether the call duration was within the acceptable range as defined by the business in question.

Fully customisable

All of the above metrics are fully customisable by management, allowing the system to provide different kinds of insights depending on business model and customer interaction strategy. Most importantly, the keywords used by the Natural Language Processing engine to define these sentiment ratios can be customised by the end client. This means that the system can be used to listen for whatever is meaningful to the decision makers in an organisation.

This allows for many innovative uses of the system beyond providing a data-driven approach to improving how your front line support and sales teams interact with clients. For example, keywords specific to the brand can be flagged, allowing the business to scan telephone interactions for mentions of its own in-house products or services. The system can also be used to identify clients that may be more receptive to a call from a sales agent by focusing on conversion-related keywords. Is business development considering the addition of a trending symbol? “Listening” for mentions of it in customer interactions ahead of time can be helpful as part of a broader market research strategy.

Intuitive to use

We’re as proud of the interface and its functionality as we are of the product itself because it very much follows the design strategy we’ve adopted for our other modules that effectively make a whole new wealth of data available to the team member using it, without them having to learn how to incorporate a completely different piece of software into their workflow.

As with our other modules, the user is able to easily drill down into the data. Keywords are clickable for quick navigation of the recording, making interacting with stored calls very efficient for other team members such as a salesperson tasked with following up, or a manager monitoring their team.

Managers can leave comments to specific agents pertaining the the content of a recording, as well as providing an overall rating the quality of the interaction as feedback. Lastly, we make it easy to sort through the recording database so that it doesn’t become unmanageable as the number of recorded interactions grows. Users are able to filter database results by duration, by agent, and more.

See for yourself

We’d be delighted to discuss how this exciting new module can help your brokerage make the most of your front line staff. Whether or not you currently use any Panda products, our success team is ready and waiting if you’d like to take Call Control for a spin.