Company: General Motors
Company background: Our goal is to deliver world-class customer experiences at every touchpoint and do so on a foundation of trust and transparency. Our vision is a world with zero crashes, zero emissions and zero congestion Our diverse team of 155,000 employees brings their collective passion for engineering, technology and design to deliver on this ambitious future. And the bold commitments we’ve made are moving us closer to realizing this vision.
Contact: Christine Darbonne
Title: Experience Delivery Manager, CoE & Social Care, CX Org
Related URLs:
Kudos Category: The Innovator
1. What critical business need were you trying to solve with your use of the Khoros platform and our automation capabilities? How had you or others attempted to solve for this need in the past and what different approach did you take to bring this innovation to life?
GM has worked to use the Khoros platform and automation capabilities to address three critical business needs, listed below:
- At GM, we want to keep up-to-date with all the content related to our top priority “Safety” topic. We have many safety-related tags that cover both the English and Spanish languages. Using Safety tags - both internally and externally - with Brandwatch, we’re able to assist customers who might be dealing with important safety-related issues that require high priority assistance from our team. As we identify specific terms through Brandwatch listening, we then add these terms as tags to the Khoros database. Afterward, they’re then prioritized in the system when the criteria is met. Combined, these Safety tags have been used over 40K times YTD. While these tags can help us spot broad conversations, we also utilize them to target precise terms or phrases such as accidents or other serious cases.
- Combining and collecting user inputs from all of GM’s Messenger & Modern Chat bot channels into a streamlined, single-source for our NLP was important. After this addition, we were then able to empower ourselves with the help of Suggested Training Examples to create and train for Intents. The inclusion of bolting on Universal Classification and Keyword Extraction to our AI Orchestration project also yielded significant benefits. For example, it enabled us to use different sets of tags to monitor an external Metabase Analytics dashboard, identify what methods were working best, and note how frequently each topic was being triggered compared to the rest. A handful of these trained and verified Intents, such as Order Status Intent, have been migrated to existing bots which, overall, has recently increased self-help capabilities with order stats request, for instance.
- We enabled tagging and routing based on a customer’s situation rather than just a topic or brand as we had previously been doing. Working with our internal advisors through interview sessions of their experiences with customers, we learned their thoughts on situations where customers expected quick response times. We identified those situations using Lucene rules and created and refined tags to identify them as part of a Best In Class pilot. We then configured the system to do any combination of things from bypassing our Moderation work queue and dropping into the Care work queue or creating an alert for any incoming matches, to boosting the priority to ensure faster response times. It allowed us to identify “live” events where customers have little tolerance for delay, and we adjusted to ensure we were responsive to customers who are expecting to stay engaged in conversation to resolution rather than step away and continue as convenient.
2. How did you take the Customer Experience into consideration as part of your innovation to ensure engagement and overall customer satisfaction with their interactions with your brand?
Customer experience is always a central consideration that guides our engagement and customer satisfaction strategy. We strive to put the customer at the center of everything we do. With this in mind, we approach improving their experience by thinking from their perspective. Where are they likely to escalate quickly? It could be something related to why their car won’t start within an app or if they’ve already spoken to a phone agent to address an issue without success. Or when they’re likely fine with a short delay in support responses for questions such as the proper tire pressure for their vehicle or how to change the clock in their car, for example.
3. How did you overcome any doubts or resistance in your organization around your innovation? How did you use data to either create a business case or prove the efficacy of your innovation or both? Please include metrics if possible.
For the situational Best In Class testing, we piloted tagging at first to verify that the situation matched what we thought it would. Then, we would pilot a bump in priority to see the effects. We’d spot-check to see if the tags and process were valid and then tweak as needed. There were three of us watching these tweaks and managing team communication over about six months before we made it a permanent part of our configuration with plans to evolve it. We also learned that not every situation needed a priority boost or specific routing. Some customer pain points that arose from the advisor interviews were addressed with a thoughtfully worded Response Template that hadn’t previously been in place. This worked well as an efficiency play for advisors as well as ensuring customers frustrations were addressed consistently.