Automate Call Center Agent Scoring With Voice Analytics Automation

by Emily Nave

The average call center agent handles 781 inbound calls per month (Via ICMI). The average call center has about 289 employees (via Cornell University), so that means most call centers process over 250,000 calls per month. That’s a lot of Quality Assurance checking (QA)!

One study showed only about 1.4%-1.8% of all call center calls are QA’ed, possibly because human scoring and transcription is time consuming and expensive. But with today’s voice analytics systems running machine learning, this whole process can be done instantly – resulting in the automatic monitoring and scoring of every single call that goes through the call center. 100% QA!

What is Auto Call Scoring?

Auto call scoring is the process by which machine learning algorithms train using results previously defined by humans (such as hot lead, rude agent, upset customer, etc.). This training process allows these machine learning platforms to zero-in on the key variables that define each pattern, in order to determine the result of all future calls.

Why does it matter?

As much as they would like to, call centers cannot afford to manually score 100% of their recorded calls by human monitoring. Right now most are doing the “next best thing,” humanly scoring as many calls as financially makes sense. Unfortunately that only amounts to 4-5 calls per agent per month on average. These call centers are losing valuable data every day by not listening to the majority of their calls:

  • Customer opinions of competitors
  • Call Drivers
  • Product feedback details
  • Sentiment spikes
  • Pre-churn signals
  • Insights about user personas
  • Security alerts
  • Sales conversion tracking
  • Agent performance across the call center
  • Trends in topics
  • Etc.

What can you detect? 

1. Script Adherence

There are certain terms agents should be saying on every call, ‘5 year guarantee, ‘zero-down’, ’30 day free trial’, ‘6 month commitment’, etc. Script adherence keeps the team unified and more effective. Certain risky phrases, if spoken by the agent, can even result in large fines. 

Using a voice analytics solution that automatically spots missing phrases, you can easily pinpoint problem areas, coaching opportunities, and team performance in a snapshot view. In a business intelligence tool like Tableau, you can then surface exactly which individual calls need to be reviewed. 

2. Profanity/NSFW Behavior

Swear words are not uncommon in a call center. Sometimes customers are upset and get a little mouthy. But you should know if your agents are the ones with foul mouths, right? If you record in stereo (or channel/multi-party), you can search just the agent side of recordings or set flags to alert supervisors if there are any NSFW terms or phrases on the agent side.

3. Competitor Names

Wouldn’t you love to be a fly on the wall when your customers are talking about your competitors? Now you can! Customers are quite candid on the phone, they mention competitor prices, marketing campaigns and other factors that effect the buying process. Spot competitor names and see what your customers really think.

4. Positive/Negative Comments

Ever perform agent reviews? Imagine having a list of all of the negative and positive comments a customer has said when on the phone with an agent. By using calls that have been tagged with specific time stamps during a “positive” comment and a “negative” comment, it is unbelievable how accurately the machine can pick up those little nuances that define each. Now agent reviews are a cinch!

5. Redact Sensitive Information

Lastly, some call centers can’t even record calls because of the information being spoken. Credit card numbers, social security numbers, mother’s maiden names, addresses and other Personally Identifiable Information (PII) is being exchanged. Instead of not recording calls, call centers can utilize an analytics solution with automated redaction that is properly certified for data security. 

The Payment Card Industry Data Security Standard (PCI-DDS) is the set of standards to which PII data is handled. Companies that handle this data must go through a rigorous assessment in order to qualify for certification. With a properly certified voice analytics vendor, call centers can instantly create a large queryable database of information from their recorded calls.

If your call center analytics solution is secure and searchable, you can analyze 100% of agent behavior in less time, with less effort. To learn more about VoiceBase and how we can revolutionize your call center’s voice analytics, get in touch with our team. 

More From the Voice analytics blog

Predictive Analytics for Strategic Insights

Predictive Analytics for Strategic Insights

Predictive analytics is an advanced form of data mining that leverages machine learning to identify patterns in voice recordings, intuit a speaker’s intent, and predict a future outcome — be it a sale, account cancellation, or one of many customized “X” signals your clients might request.

read more