How to Scale User Feedback

Problem

As a company that aims at small- to medium-sized companies, we are in the numbers game. Ensuring that everyone’s voice is being heard and that our business model replicates best democratic practices is exceptionally difficult. For example, if 100 000 people use our product, would conducting 50 interviews be a statistically solid sample?

In today’s world, things are moving exceedingly fast. By the time one has processed customer feedback from past September, the competition would emerge right out of nowhere. To get accurate feedback, I needed something like a thermometer to show me what our customers think or like. Indeed, one can always run tests, but qualitative insight obtained through a conversation is second to none.

Actions taken

Customer interviews have traditionally been run by UX, design, or product people. As we wanted to democratize the process, we tried to make every customer-facing person capable of being a user researcher. We would arm them with the tools that would allow them to collect feedback on behalf of the product team. For example, if one is a sales rep who takes notes on product feedback in Hubspot, their notes would be automatically sent to the product team’s insights-gathering database where we could process it. The same happens with Customer Success or Support. By building different touchpoints, we scaled our ability to collect qualitative feedback and made every customer-facing employee a potential researcher.

However, that required a developed network of interconnected tools that would allow everyone to push feedback from within their ecosystem. We use Zapier to automate the process and Product Board as our central insight-gathering tool paired with a complex tagging system. Insights are being created automatically from a variety of tools; the sales team uses Hubspot, Support uses HelpDesk, while insights from all social media and an admin platform are emailed to us.

Our philosophy is to listen to the market as much as possible and have a rigorous process of triaging information. That demands building a consistent culture around collecting feedback through training, “lunch and learns,” and nurturing cross-team collaboration. We created training material to help people from different departments that use different tools to collect feedback. For example, when a customer says that they don’t like something, they would be instructed to look for a problem and not a solution. They should be asking, “Why don't you like this button?” and “What is it that you are trying to do with it?”. Applying the Five Whys method would allow us to receive feedback of the highest quality that focuses on a problem.

We also work with virtual assistants in the product team who are responsible for cleaning the data, calibrating tools, and classifying insights. They are an integral part of the product team who own the triaging; they are not contractors who often lack context and understanding of a broader picture. Every month I am meeting people from Sales, Support, and Customer Success to have an open QA session on how they could contribute to Product by collecting and submitting high-quality feedback.

Lessons learned

• It is easy to hide behind data as an objective measure of what people think or like. But data tells you what you already know since you are the one building the analysis around the data. The articulation of the problem is intrinsically biased, and data can be as objective as one is objective.
• By scaling feedback, you will be able to have an open conversation with customers, which is always much desired, particularly when a company scales. The more the company scales, the further you are from the actual customer, making it easier to forget that there are “real” people who are using your tool.
• Maintain that connection with users at all costs. By empowering everyone to share feedback with Product, all the feedback will arrive at one place and be processed.
• Lower technical barriers for people to send feedback. If you tell people to send you feedback, which would require them to go to another tool, 80 to 90 percent of them will not do it. At the moment, we are testing a new solution; a sales manager would record its call, and a tool that transcribes the call into a text will automatically extract the segment where a sales manager says, “I’ll send you feedback to Product.”