Q&A: Why Call Recordings are Useless Without Analytics
We’re all used to hearing that our calls may be recorded for training purposes, but most people don’t give a lot of thought to what that really means--including, sometimes, the businesses doing the recording. Just like any data, your recorded calls are only useful if you can make sense of them and gain insight. Speech Technology Magazine’s editor, Theresa Cramer, talked to Jeff Gallino, CTO at CallMiner, about the information buried in your recorded calls and how to make the most of it.
Q: What types of calls should enterprises be recording?
A: Many enterprises record all customer service or sales calls, and they should for a number of reasons. But the value doesn’t come from just recording interactions – it comes from analyzing them, and this is where many companies fall short. According to DMG Consulting Research, quality monitoring is often limited to only 1 to 3% of calls because it’s too costly to assign more resources to the task. This inadequate sample size (which is often prone to human bias and inconsistencies in evaluation) doesn’t provide accurate analysis, and so thousands of call recordings from the call center stay in the dark – as does its rich data.
AI-powered speech analytics give enterprises the power and ability to analyze 100% of call recordings, shining an objective, transparent light on every conversation and customer voice. When these words become structured data, actionable insights emerge – and it’s what you do with this data that can really transform the enterprise.
Q: What kind of information could traditionally be gleaned from call data?
A: Before speech analytics, quality monitoring consisted of call center supervisors listening to a very small fraction of calls to glean transactional information such as length of call, time and date, agent and supervisor, and in some cases, holds and transfers. If call centers were connected to CRM, they could learn about call outcomes such as a sale, account cancelation, etc., but there was no way to extend visibility to really capture trends.
Traditional call centers also used a net promoter score (NPS) and other similar surveys – if they were lucky enough to get responses – to understand customer experience, though that often depicts just a fraction of the data available to accurately highlight the customer experience (CX) of a company. Surveys are a good start to gaining insight into CX, but the response rate is only between 5 and 15%. Perhaps the biggest shortcoming of using surveys alone to assess CX is that they often only attract the extreme ends of input from either very happy or very angry customers. Therefore, information traditionally gleaned from call centers was often skewed. Call center employees typically had very little insight into their customer conversations and its overall CX – and some call centers still operate like this.
Q: What new types of information are we now able to get from call data?
A: A modern call center that has implemented speech analytics technology can move beyond anecdotal information based on a small sample of calls. Today, critical insights that inform and influence the customer experience, assess and impact call center agent performance, and provide a new level of business intelligence are all possible through the call center. For example, we can now have a clear look at what the call was about and why the customer was calling, rather than relying upon agent dispositioning – seeing as research shows that agents frequently pick the top disposition code, regardless of what it is, even if the call had more than one.
With speech analytics, we’re also able to learn how the customer felt during the call and their overall sentiment about products, services and policies. Insights into agent performance through augmenting metadata to associate agent quality scores with each agent, time of day and other metrics are now available. Was the agent compliant, did they represent the brand appropriately, and did they deliver a positive experience? All of this information is now at the fingertips of call center supervisors.
Missing from traditional call center measurement was context that allows for driving efficiency and improvement. For example, call centers now have information on why some calls are long, why some are short – and can improve processes to accommodate each, such as moving short calls to self-service channels. When calls contain long periods of silence, we can analyze and discover what exactly is driving that.
By analyzing 100% of calls, organizations can track customer and agent behaviors over time in order to make adjustments and speed time to action.
Q: What are enterprises missing out on if they aren’t making the most of their call data?
A: If enterprises leave their call recordings in the dark, they are missing out on a plethora of insight that could lead to improved first-call resolution rates and enhanced CX, opportunity to cross-sell and up-sell to interested customers, real-time alerts of product issues, better agents and less turnover, a clearer understanding of its audience and their needs, and so much more. Not extracting data from call recordings is a lot like being a billionaire with no access to your bank account – the wealth is there, but you can’t benefit from it. Companies who will win over its competitor’s customers will give itself access to the wealth, and business will certainly flourish.
Just as brands are perfecting DX, they have to rethink every aspect from search to delivery. Eli Finkelshteyn, CEO and co-founder of Constructor.io, weighs in on the challenges brands face in the voice arena.
People are increasingly looking to customer service to solve complex problems quickly, on a multitude of platforms -- and the brands that can find a way to do all of those things will build long-term affinity with their clients. Here's how companies can achieve success by keeping customers around with proactive customer service.