Gainsight Cuts Response Time by 50% and Improves Manager Productivity Using Sentiment Analysis and Auto-Case Routing

Gainsight launched SupportLogic’s NLP-powered Core SX platform as a solution for improving the efficiency of its enterprise and standard support tiers. The platform analyzes the unstructured data and customer interactions in every case, tagging sentiment signals and customer impressions. This data is then used to automate case assignment, prioritize cases by attention signals, and predict potential escalations.

“When we got SupportLogic, the biggest use case was trying to identify cases with negative sentiment from our customers,” explains Davis. “We’ve got a lot of high-paying customers in our paid portfolio, and we have agents that are specifically assigned to help them manage their account and their relationship with Support. We brought SupportLogic in to be able to pinpoint key areas, and help support managers quickly see exactly where they can jump in and make the biggest impact—instead of having to open up account pages and look through every case that a customer has open.”

SupportLogic’s intelligent case assignment tool used Gainsight’s case history to predict the best available assignee for every case by automatically determining bandwidth, agent skills, customer history, schedule overlap, and complexity. After the AI model ingested the case history, the Gainsight team spent a couple of weeks verifying the tools’ suggested assignments.

“We removed the need to have someone sit there and say, ‘this is about problem X,’ and then manually categorize that issue, put it into a bucket, and triangulate which support person it should go to.”

With the team impressed by the tools’ suggested assignments, Gainsight turned them on ‘auto mode’ and saw immediate time savings for their support management team. “We saw the cases come through, and that the skills the tools were indicating were matching the skills that we thought should be on each case. It wasn’t scary because we didn’t start with auto-assign, we started with suggested-assign,” recalled Davis. “That was the big eye-opening moment for us: realizing how many agents we had that were finishing their workload with three, four, and five hours left to go in their shift, and then just sitting back and waiting for their next case to go from pending into open, and taking action.”

With automatic case assignment, the team uncovered even more efficiencies, finding opportunities to assign open cases to idle agents and speed up time to resolution. “So what we’re doing is saying, ‘okay, this team has zero open…let’s go ahead and give it to them and have them start working on it.’ That has absolutely lowered our response time, because it’s taking and giving a case to somebody who’s sitting idle as opposed to somebody that even has one open at that moment in time, and is currently tied up. That’s the biggest thing that we’ve seen—the initial response time continues to keep going down.”

The efficiency improvements didn’t end there. Gainsight took SupportLogic’s sentiment insights and analytics dashboards, and looped in the product and onboarding teams, spreading the voice of the customer throughout the organization.

“Getting our support health scores integrated with our customer success team has been huge. Our services team has been using it as well while they’re going through implementation and onboarding. So we’ve had a lot of project managers that have been able to go in and set up accounts, and they can actually favorite accounts, look at the ones that are going through onboarding, and take a look at tickets at a high level.”

“One of the most compelling things that we’ve shown the product team is that when you aggregate data, you remove the human component. I didn’t have a tool that was able to say, ‘this is the overall positive sentiment or negative sentiment about the product.’ And the way that I’m able to do that is to go into text analytics, type in ‘email configuration,’ search every single email ticket that we’ve mapped in our ontology, and then get an overall sentiment score. That was unheard of before. So they’ve been going in and looking at all their different products and seeing what the sentiment scores are. And then be able to drill into the cases that make up the sentiment score, and see what people are saying about their products via support.”

“Our team is using the text analytics with our product features, and combining that with pulling the data at certain times to gauge what the sentiment is for customers around release time. It adds that sentiment aspect that wasn’t achievable before, where we were only able to see the metadata. But now we’re able to tell which customers are angry, frustrated, or confused.”