How we're generating meaningful insights from unstructured data

This is the second in a series of blogs describing our journey to deploying CX Insights here at Khoros. If you haven’t already, check out our first blog, and stay tuned for more!

As you know from reading our first blog in this series, Khoros is currently in the process of implementing our own customer interaction analytics solution, CX Insights, and we’re blogging about the process. In the last blog, we talked about analyzing the customer journey and growing our Customer Insights team’s influence. This time, we’re going to discuss our first successful uses of CXI and how we’re planning to iterate on our research approach. We’ll go over what’s worked, what hasn’t, important KPIs at this stage, and much more.

Let’s kick things off with where we are in the process right now.

Last time, we had just gotten started building our CXI implementation team and analyzing our customer journey. Since then, we’ve gotten the CX Insights platform up and running, complete with Gong (our call recording platform) integrations to make sure we’re pulling in data from all our channels. We have already generated our first report, which gave us our first glimpse at the incredible power of the solution (more on that below). We’re working on making key improvements to get even better results when we iterate the process for our next report.

We’ve also just rolled out an effort to record more customer conversations in Gong to ensure that we are feeding our CXI engine with as much customer interaction data as possible to surface even deeper insights from this critical touchpoint.

The CX team is focusing on sharing insights with internal departments to help them understand where friction is occurring, why it’s occurring, and how we can minimize it. Tying future changes to measurable indicators will allow us to see whether the adjustments are improving the experience for our customer or not. Identifying the process adjustments that we’ll be making is not yet finalized.

CXI implementation can be a bit complicated, especially with B2B sales. Here’s what’s gone really well — and what’s challenged us — in our CXI implementation so far.

Successes: Our first CXI report

Let’s start with the positives. After implementation, we had two internal teams of analysts create an insights report using our CXI data. These team members work to support our CXI customers on a day to day basis, so they’re no stranger to the tool. The report provided a fantastic resource for understanding the kind of insights that were being surfaced  from Khoros data, as well as how to approach the data research.

Our CX Insights report validated a lot of what we’ve already captured in surveys and other forms of solicited feedback, and we were pleased to see the consistency there. But even more important for our team were the insights into parts of the customer journey that we normally struggle to analyze because of the unstructured nature of the feedback. For example, CXI gives us the ability to analyze sales calls, surface recurring themes in support tickets, and even search for competitive mentions in customer interactions. This will allow us to peek behind the curtain to better understand key issues our customers and prospects are bringing up in conversations, identify which friction points are leading to the most support tickets, and surface areas of opportunity for sales training and objection handling.

This initial CXI report also gave us insight into small product enhancement requests that the tool identified as suggestions in conversations with our customers. These suggestions were small but frequent, and this volume indicated that there might be opportunities for improvement in these areas.

We identified potentially important trends in several other areas, as well:

Sales calls

Our first report showed us which products sellers talk about most frequently together. We saw that sellers often mentioned certain specific products together but left out other, equally relevant products. This flagged a training opportunity for our sales team to get more confident about selling certain products alongside others.

Support tickets

By analyzing support tickets, CXI flagged automated log-out time as a frustration. The platform will log users out after a period of time for security reasons. The action here is probably to increase transparency and communication around the importance of security for Khoros + customers.

Challenges

So far, we’ve been fortunate not to experience many setbacks, but as with any implementation, there have been certain challenges along the way. Perhaps our biggest challenge is that B2B customer interactions are unique cases, so we had to be careful to tailor our implementation correctly.

One of the main differences between B2B and B2C models is that users don’t contact the company in the same way. For B2C interactions, it’s usually just one person contacting an agent, and it’s easy to see the back-and-forth conversation between agent and contact. This means they can make sense of multiple interactions from the same customer, even when they occur on separate channels. In B2B interactions, on the other hand, we often have conference calls where several people from an account talk to the 'agent' — and sometimes there are multiple agents or sellers on the call as well. This necessitates a different way of interpreting conversations where we look at the feedback more from an account perspective than an individual contact perspective.

Another concern for our team has been ensuring accurate results. The keyword confusion popped up more than we were expecting in our results. But it’s hard for us to tell whether that’s because our customers are confused, or because they’re talking about solving their customers’ confusion. Again, this is likely specific to our own B2B use case, but we’re analyzing these results more carefully to figure out the source of that keyword.

We’ve learned quite a bit through the successes and the challenges of implementation. The biggest lesson so far? We were missing out! The insights we used to get from structured data were helpful, but they were incomplete without insights from unstructured data — and that’s exactly the power of Khoros CXI. Now, insights from unstructured data are turning into actionable feedback that we can use to improve both our customer experience and our internal processes.

We will be collaborating with teams internally and aligning on their priorities. This will inform us on what gaps we can solve for using CXI and allow us to take a more specific approach to our research with the tool. 

There are two important next steps for us. First, we plan to iterate this process, using what we’ve learned from our first report to get better, more accurate reports in the future. As we iterate, we will identify ways to structure our research to be even more efficient and valuable to each of the teams involved.

Second, we will take our new actionable insights to the teams that can benefit from them. This involves developing a presentation plan for other departments and creating assets that they can use to improve their processes. After all, CX is a team sport, and keeping the whole team informed is the way that we can win.