Thought Leaders in Big Data: SupportLogic CEO Krishna Raj Raja (Part 2)

Sramana Mitra: How do you go to market? Do you sell directly to enterprises or other companies that you mentioned?

Krishna Raj Raja: Great question. Salesforce is a customer of ours. Salesforce Support uses Support Logic. We have not done joint go-to-market. We sell directly, but we’re forming partnerships with ISP vendors and delivery and support vendors. If you have an offshore agency that delivers support, they care about the quality of support they deliver. We cater to them as well.

Sramana Mitra: So Salesforce is a customer; not an OEM partner.

Krishna Raj Raja: Right now, they’re not an OEM partner. They are a big fan of what we do and a big customer of ours. We’re working towards an OEM partnership.

Sramana Mitra: Let’s take some use cases of how you deliver value. Pick whatever customer you want to work through. Go into some depth.

Krishna Raj Raja: What we are going after is unstructured data in the system of record. If you look at the entire computing industry in the last 30 years, the focus has been on metadata and machine data. That’s what we’ve been relying on.

If you look at any ticketing system, they’re relying on that metadata. This is the first time that we have technology to process unstructured data. We’re talking about voice data, text data, images, and videos. This shifts the paradigm on how you look at your customer relationships.

Customer relationships have been very transactional. It worked in the era of transactional sales. It also worked in the world of transactional databases. Now we have a continuous stream of conversations with customers. Customer communication doesn’t stop with a ticket. Maybe they’re using different channels. You want to have a technology that can go beyond transactional boundaries. You want something to continuously analyze what customers are telling you.

We sit on top of your system of record. We sit on top of Zendesk, Freshdesk, and Salesforce. We pull that data continuously in near real-time. We do a differential sync and process every content that is registered there. We run it through natural language processing. We use deep neural networks and a combination of traditional ML models. We also use a technology from Stanford called spaCy for language parsing. We also use Google’s technology that’s part of the deep neural network stack.

We extract rainbow color of signals. We can identify what is a customer’s sentiment. Is it frustration or impatience? Then we extract flavors of customer urgency. From the tonality of the message, we can extract that. Then we extract keywords found in those tickets. What is the customer talking about? What hardware are they using? Is it a Dell or Macbook?

Then we also look at outbound conversations. What does this support engineer tonality look like? Is it empathy? Are they greeting the customer properly? We look at the progression of sentiment. Is it getting worse? We also maintain signals across ticket boundaries.

From all of this, we used to predict which customer will escalate an issue. We predict escalations well in advance of when a customer escalates an issue. We’re taking into account not only the current sentiment and urgency of the customer but also previous tickets. We consider all factors and predict customer escalations.

We send an alert to the right stakeholders and we tell them that this account is going to escalate and the reasons they’re escalating. We allow various people to swarm on our platform. It could happen not only in support. Sometimes, you have to bring engineering and leadership. They can collaborate with the support engineer to resolve and diffuse customer escalations.

Since we do signal extraction, we also recommend the right subject matter expert who should handle the case. Since we track outbound communication from the support engineer, we can also track which engineer needs training for soft skills. We have a module that allows you to coach support engineers. All of this functionality is in the service of delivering great support experience to the customers.

The signals we extract also help provide analytics using dashboards. We also send real-time alerts to the stakeholders. If it’s an account that you really care about, you get a notification on your Slack before it escalates. Net result is, our customers reduce escalations by up to about 50%. 40% is pretty standard. The mean time to resolution drops significantly. Because we’re doing sentiment analysis on every case, you don’t need surveys. Our customers have stopped using surveys.

This segment is part 2 in the series : Thought Leaders in Big Data: SupportLogic CEO Krishna Raj Raja
1 2 3 4