StrategyMediaInvoca’s Signal Discovery uses machine learning to conduct conversational analysis on inbound calls
Invoca’s Signal Discovery uses machine learning to conduct conversational analysis on inbound calls
Call attribution and analysis can be particularly tricky for large organizations. We spoke with two thought leaders at Invoca about their recently launched Signal Discovery and reviewed a case study from a client.
Tracking and analyzing inbound calls is a big pain point for marketers, with multi-channel attribution falling short of other channels such as website analytics in terms of reliability and accuracy.
Data quality issues abound. More than half of marketing leaders in a 2018 worldwide survey reported that multi-touch attribution was one of the biggest gaps in their marketing research efforts.
Call attribution and analysis can be particularly tricky for large organizations that get tens of thousands of calls each month.
Invoca, an AI-powered call tracking and analytics platform, recently launched Signal Discovery to address the issues of accurate multi-channel attribution and call analysis for enterprise and mid-market organizations.
ClickZ spoke with two thought leaders at Invoca about Signal Discovery and reviewed a case study from a client that has successfully implemented this new tool.
Considered purchases rely on inbound calls
Considered purchases—like automobiles and homes—rely on inbound calls to facilitate the buying cycle from discovery to consideration to the final sale.
“With considered purchases it’s helpful to remember that inbound calls are critical,” explains Ian Dailey, Invoca’s Senior Director of Product Marketing. “Consumers want to speak with a human being. You need to consider that your data from your contact center is as important as your data from your website.”
Rich call analytics have not been available in the past, leaving a big question mark when it comes to inbound phone calls. With Signal Discovery, Invoca provides marketers with actionable first-party data gleaned from their call centers.
“Conversational data from phone calls is one of the last untapped resources of first-party data,” says Dailey. “Signal AI helps marketers automate the process of extracting insight from these calls.”
Mapping conversational data
Signal Discovery uses machine learning to uncover new behaviors from first-party call center conversations. The tool helps marketers automate the process of getting insights, in real-time, from a large volume of calls.
“When Signal Discovery hits a threshold of about 20,000 calls, it moves to unsupervised machine learning hours,” explained Sean Storlie, Invoca’s Senior Director of Product Management. “It outputs a conversational map. The first thing a marketer can do is navigate this map to look at past calls and understand what the calls look like and what’s really going on in these conversations.”
Each color in the above map represents a different topic. Similar topics are grouped using related colors (e.g., sales topics, customer support topics, complaints, etc.) Users can apply alerts or “signals” to new inbound calls so they can begin to diagnose issues across the entire buyer experience. This is a powerful way for marketers to collaborate with the customer experience team around the total buyer experience.
Marketers can drill down on any specific topic by clicking on a colored circle in the map. This produces a word cloud that shows the most commonly used words and phrases in that topic. The user can then give the topic a label (e.g., customer support issues).
Once the signal is created, it can be used to flag calls that meet the topic criteria on an ongoing basis. The signal can also be pushed to various marketing platforms and tools via one of Invoca’s many integrations.
“You can create an audience in Google Ads to suppress ads from showing for certain keywords based on the signal created in Signal Discovery,” explained Storlie.
The tool allows users to track and diagnose issues in real-time then “activate” each signal by pushing the call data to third-party platforms such as Google Ads, Facebook, and Salesforce.
University Hospitals success story
Dailey reviewed a case study involving University Hospitals, a large medical organization in Cleveland Ohio with 18 different hospitals, 55 health centers, and over 30,000 employees. They receive over 400,000 calls every month.
University Hospitals had twelve employees listening to calls for many hours each week. The team had been trying to understand what calls were triggering new appointments by analyzing keywords and call times, but they were doing this manually.
“Once they started using Signal Discovery, they realized that their previous method was woefully inaccurate and inadequate,” said Dailey.
The hospital staff learned that 29% of calls were being routed to a messaging service with a conversion rate that was 50% lower than a live conversation. Once they adjusted their call-routing process and improved agent training, they saw a 580% increase in appointments scheduled. They also saved upwards of 40 employee hours per week since staff no longer needed to listen to and attempt to analyze calls.
Signal Discovery is currently only available to mid-market and enterprise organizations with a large volume of calls, but smaller businesses have the option of using pretrained models which reduces the need to reach the 20,000 call threshold.
Storlie confirmed that they’re currently researching a model that works with a lower volume of calls. “Our preliminary research shows that we can push call threshold as low as 5000 and still gain valuable conversational insight.”
Using conversations from inbound calls to develop new first-party data about high-intent buyers, taking action on this data across existing martech platforms and accurately attributing leads and sales to the appropriate channel is a bit of a unicorn for marketing managers.
Signal Discovery promises to do all three using the power of machine learning. It helps marketers connect the dots more efficiently and effectively than ever before.