How Notion is supercharging its product feedback loop using Enterpret
The product operations team at Notion gathers user insights to represent the voice of the customer, working with product development to prioritize users’ most important problems.
As the user base and company grew, Notion’s manual tagging process (>700 unique tags, tens of thousands of monthly support tickets) was reaching its limits, with two major problems:
- Lack of trust in the underlying data
- Time-consuming to get insights, with no way to visualize and segment feedback data
Enterpret has unlocked the following capabilities for the product ops team:
- Unified feedback platform and taxonomy across support tickets, Slack, Twitter, and app store reviews
- Machine learning identifies feedback reasons and surfaces unknowns, serving as a “second brain” for the team
- Self-serve insights via queries and user segmentation in Enterpret
- Cut the time spent on the monthly user insights report from 2 weeks to 3 days
- Informed a decision to dedicate an engineering team to fix login problems
“The advantage of Enterpret is that we’re not relying entirely on human categorization. Enterpret is like a second brain that is looking out for themes and trends that I might not be thinking about.” - Misty Smith, Head of Product Operations at Notion
Over the past few years, Notion has become an essential part of many teams’ workflows. But as Notion’s customer base grew, it became increasingly difficult to stay on top of user feedback coming in through multiple channels.
We recently sat down with a few members of Notion’s Customer Experience and Product Ops teams to learn more about how they’re tackling these issues so they can learn from their customer feedback at scale.
Challenge: The limitations of manual tagging and insights
In late 2020, the team recognized that their previous efforts to learn from user feedback were breaking down.
Lack of trust in manual tagging
Notion gets tens of thousands of support tickets coming in monthly, in addition to ever increasing feedback from Twitter, surveys, app store reviews, and community forums. Every support conversation was getting tagged by a support agent, but they needed to choose from a set of over 700 tags.
Accuracy was a big concern: with a distributed team and so many tags to choose from, how do you ensure consistency in how they are applied? The product development team was losing trust in the data’s reliability and usefulness.
Time-consuming manual insights
In addition to waning confidence in the underlying data, it was difficult to get insights out of it. They were dumping feedback records into a database, but there was no easy way to visualize feedback data, identify trends, or slice and dice by different user segments.
Solution: Automating user insights with Enterpret
Notion’s product ops team decided to bring on Enterpret to help them derive more learnings from their customer feedback data. Some of the main benefits they found include:
- Having a unified feedback platform
- Leveraging machine learning to surface unknowns
- Self-service insights
Unified feedback platform
With Enterpret’s integrations to different feedback sources, Notion was able to setup feedback from Intercom (support tickets), app store reviews, and Twitter with just a few clicks.
As Emma Auscher, Notion’s Global Head of Customer Experience (CX) puts it:
“Enterpret helps us have a holistic view from our social media coverage, to our support tickets, to every single interaction that we're plugging into it. Beyond just keywords, we can actually understand: what are the broader sentiments? What are our users saying?”
Leveraging machine learning to identify feedback reasons and surface unknowns
After connecting Notion’s feedback sources, Enterpret used natural language processing (NLP) to generate a custom feedback taxonomy for Notion.
According to Misty Smith, Head of Product Operations:
“The advantage of Enterpret is that we’re not relying entirely on human categorization. Enterpret is like a second brain that is looking out for themes and trends that I might not be thinking about.”
The real magic happens with ‘reasons’, where Enterpret’s model identifies why the customer gave the feedback; things like: “want to upgrade to a team plan”, “unable to add members to account”, or “keyboard shortcuts don’t work”.
“Reasons get more granular than our internal tags. Our tags capture features or product areas, but not the actual feedback itself,” says Maya Bakir, the product ops specialist responsible for user insights.
Self-service insights: querying and segmenting customer feedback
“Enterpret has made it so much easier to navigate through our feedback,” says Maya.
“It has been monumental in terms of being able to visualize trends and filter on our data so easily. That’s not something we were capable of doing before. If I’m curious about a specific trend, I can quickly find it in Enterpret. I can see how it’s changed over time, segment by users on different plan types, and see all the raw feedback records.”
In addition to regular reporting on top trends, “we’re starting to use Enterpret more for ad hoc, one-off questions,” says Misty. “PMs are asking me: Hey, what are the top requests related to search? When do users report issues with real time collaboration?”
Enterpret makes it easy to quickly search for feedback data to validate a hypothesis or inform product planning.
Supercharging the product feedback loop
Over the past year and a half, Enterpret has enabled the product operations team to work more efficiently and effectively in their partnership with product development.
Cut the time spent on the monthly user insights report from 2 weeks to 3 days
Every month, Maya puts together a report on feedback trends that gets discussed in a voice of the customer (VoC) meeting. Enterpret has enabled a drastic reduction in the time she spends on the report. “In the past it would take me two weeks. With Enterpret I can get it done in 3 days,” says Maya.
That report, along with inputs from sales, customer success, and product marketing, helps Notion answer questions like: “Do we still agree with where the product roadmap is at? Are we still working on the right things, or is there anything we need to flag from our user base that we need to fix right away or think differently about?”
Using feedback insights to advocate for customer problems
Feedback reasons identified in Enterpret have helped drive more productive discussions with product and engineering. According to Misty, “The reasons help make the problem more concrete in people’s brains.”
‘Login problems’ was consistently a top category of feedback that wasn’t getting addressed. The product ops team was able to say, “Look, this month we noticed there’s a specific type of login problem that is surfacing.”
Because the reason Enterpret provided was so specific, it sparked a really good discussion with the product team. Says Misty, “I can draw a dotted line from those discussions to the fact that we now have a whole new engineering team dedicated to fixing login problems.”
Continuing the partnership between Notion and Enterpret
Notion recently renewed their plan, starting their second year of working with Enterpret.
“What makes the difference is the Enterpret team and how responsive they are: proactively working on issues we have, looking at solutions, and always coming to us with proposals on how we can move forward. That made the decision to sign another year’s contract that much easier. It’s a partnership.” - Emma Auscher, Global Head of CX
What is the team looking forward to over the next year? “I'm hoping that we can empower more of our product managers to go into Enterpret themselves to understand our feedback data,” says Misty, Head of Product Operations.
And for Maya, she plans to continue to partner closely with the team as they iterate on features that make the platform both more powerful and easier to use.
“The Enterpret team is always really quick to take action when I have a question or feedback. They are the most responsive, helpful team I've ever worked with. They have helped unblock me in so many ways that I didn't think would be possible.” - Maya Bakir, Product Operations