As soon as a customer submits a support ticket, the clock is ticking. Support agents know that they need to close the case, and they need to do it fast.
Motivating agents to better know the playbook and make improvements to response time is one thing. But how do you go from baby-step improvements to a great leap forward in customer experience?
Major Improvements Require Systemic Analysis
Making improvements based on customer feedback for your Customer Support team is like trying to see the forest through the trees.
A Customer Support leader’s job is to step back out of the forest to get a better sense of the whole. On top of this, Customer Support leaders have tons of customer data at their disposal. It’s not always easy to get valuable insight from vast amounts of data, especially when it comes to the voice of customer feedback. This calls for some kind of deeper, machine-assisted analysis.
Leverage Machine Learning to Extract Insight from Comments
As I talked about in my last post, machine learning algorithms make sorting out qualitative feedback quick and easy. They mine the data for you so you can plan out and execute steps to improve based on the insights the algorithms turn up.
For Customer Support, there are two customer feedback sources that can be extremely valuable for machine learning text analytics to mine:
1. Emails/messages from customers reaching out about support issue, descriptions and comments from support tickets
When customers write in about their issues, machine learning algorithms tag comments by categories (“technical”, “refunds”, or “dashboard”, for example) in real-time. This automatic categorization allows you to see what subjects are trending overall, as well as dig into specifics. It can also help you route particularly unsatisfied customers to the appropriate follow-up channels.
For example, let’s say you receive a support request email with this description:
“Tried to check out but the payment window that popped up kept saying to click a button that wasn’t there.”
This would be tagged with the categories “check out” and “bug”. After noticing that both of these categories are trending, you could route all the tickets tagged with these two categories to someone on the product team to handle as a larger issue.
Having these requests categorized in real time saves time for everyone involved. If you’ve previously had your customers categorize support requests for you, this is one less step for them to go through. Alternatively, if you’ve had agents categorizing support requests, they get to skip this step too!
On top of the time saved, the machine learning algorithms identify all of the relevant categories and will often catch more nuance than the static categories you have to give as options in a drop-down menu.
2. Customers’ responses to CX surveys
Having a score like CSAT is a great way for Support teams to objectively evaluate their work and see improvement by glancing at a number. The most valuable part of customer experience surveys, though, is in the qualitative feedback—the why behind the score.
If you receive more than a few hundred comments a month, tracking them in spreadsheets quickly becomes unwieldy. This may be the time to turn to a CX platform that can mine the “why” from comments using machine learning.
If you pass Support agent IDs as a property to your customer feedback platform, you can pair up the tagged text and sentiment analysis with ticket closure times to see what types of issues they handle the fastest. This would allow agents to understand their own strengths and weaknesses when it comes to product knowledge.
You should also keep in mind, it’s not always about improving the Support team’s performance. In fact, one of our Wootric customers analyzed their CSAT feedback comments and came to understand that their Support agents were almost too good. Their actual CSAT score was incredibly strong, but their comments were being tagged as mixed sentiment. Digging into this discrepancy revealed that while folks were happy with the support they received from agents, they were really looking to solve their issues on their own.
As a result, this customer created a more robust self-service knowledge base. This had the effect of reducing the number of support tickets, so that agents could focus on more sophisticated or situational customer issues. Customers were happy, agents were happy, and company resources were being used more efficiently.
A New Source of Insight in Minutes
When you ask customers for feedback, you don’t have to resign yourself to hours and hours of work to mine the valuable insight you know is in there. With machine learning, you’ll bring structure to qualitative feedback in minutes and have a new source of insight at your fingertips.
Even with your Support agents working at 110 percent, there’s always room to improve to make their lives easier and the company more efficient. The key to prioritizing improvement projects lies in the voice of your customer.