Voice of the customer (VoC) is a customer-centric framework that help product teams understand their customer’s recurring issues, sentiments and expectations through collating feedback across different channels and analyzing them. While most product teams use VoC to help make product related decisions, it can be tedious to keep track of thousands of feedback. Additionally, performing a thorough analysis over all of this feedback can prove to be a mistake-prone endeavor.

Whether you’re using customer satisfaction scores (CSAT), the Kano model, or just looking at reviews in general, product teams will quickly run into a common set of issues:

Issue 1: Collating feedback

We’ve seen this pattern happen all too often:

Product team:

“Okay, we would like to find out what the customer things about this, let’s find out what they’re saying”

Followed by:

“We’ve been compiling this gigantic excel sheet for 2-3 hours every week for the past month. It’s really tiring”

Compiling feedback by hand over multiple platforms is really labor intensive, and it doesn’t allow product teams to keep track of real time data.

Issue 2: Analyzing and tagging individual feedback

Tagging each individual feedback isn’t only time consuming; as you’ll have to read every single feedback, it’s also prone to human error.

Issue 3: Qualitative analysis

Analyzing qualitative data is usually subjective in nature. Interpreting underlying sentiments, spotting patterns / themes and assigning different importance to these themes is something that’s usually left to the discretion of the person in charge of the VoC. The individual performing this analysis must generally also have read most (if not all) of the feedback, which can again, prove to be quite tedious.

How can AI help?

Over the past 3 months, we’ve spoken to hundreds of product team members, and we’ve found that most of them use AI in some form or another to help with their day to day tasks.

In general, most of these product team members do the following steps to use AI in their workflow:

Step 1: Prime the AI by prompting it with something like “You are a professional product manager, help me understand the following set of feedback”

Step 2: Copy a bunch (up to the token limit for ChatGPT) of pre-tagged feedback into ChatGPT.

Step 3: Ask follow up questions like “what’s the overall sentiment” or “what are some common topics among these feedback”

While this approach works, it typically doesn’t give a complete picture either because of the initial prompt used, or because of the token limit. There are ways to bypass these two issues - by copying more feedback in a subsequent message and refining the initial prompt, but that normally runs into its own set of problems.

How can Commune Help

Commune makes this process simpler by:

• Integrating with feedback sources so collating data is just one sync away (We’re working on introducing more feedback over time)
• Automatically tagging each individual feedback with themes and sentiment so you don’t have to
• Analysing all, or a subset of this data to produce a report in one click.