You Think You Know Why Users Churn. You Probably Don't

Here's a question worth sitting with: when was the last time your product team had a genuine, unscripted conversation with a user who didn't love your product?

Not a support ticket. Not an NPS follow-up. An actual back-and-forth where a confused or disengaged user walked you through exactly where you lost them. If the honest answer is "a while ago" - or "never, really" - you're not alone. According to Maze's Continuous Research Trends Report, only 39% of product teams have weekly touchpoints with their customers. Product discovery coach Teresa Torres puts it more bluntly: she still meets product managers every day who have never talked to a customer - or worse, aren't allowed to.¹

Meanwhile, demand for user insights is going in the other direction: 55% of product professionals say demand for user research increased in the past year, according to User Interviews' State of User Research 2025 report.² More pressure to understand users. Fewer resources to actually do it. And a growing pile of survey data that feels like insight but often isn't.

That gap is the problem. And it's exactly what Product Fruits Discoveries was built to close. We have covered this feature in our recent webinar, you can either watch it here, or red the main takeways below.

‍

Collecting Feedback Is Easy. Understanding Behavior Is Hard.

Surveys have a place. They're useful for tracking quantitative signals - NPS, CSAT, feature satisfaction scores. If you want to know whether sentiment is trending up or down, a well-timed survey will give you that.

But if you want to know why a feature isn't getting adopted, why a segment is churning at a higher rate, why users activate but never reach the "aha" moment - surveys start to break down. Fixed question paths lose nuance. Branching logic is rigid. And even a carefully designed survey can't chase a thread when a user says something unexpected.

The result is a kind of confident ambiguity: you have data, but the meaning behind it stays just out of reach. You guess. You hypothesize. You ship features based on the loudest voices rather than the most representative ones.

Only 16% of organizations have fully embedded user research into their processes and decision-making, per the User Interviews 2025 report.² The other 84% are, to varying degrees, flying partially blind.

Enter Discoveries: User Research That Runs Itself

Product Fruits Discoveries is a new feature - currently in Early Access - that approaches user insight differently. Not as a survey. Not as a fixed questionnaire. More like a user interview that scales.

Here's the core idea: you describe the problem you're trying to understand, and Elvin - Product Fruits' built-in AI - takes it from there.

Elvin guides you through a brief conversation to clarify your goals, then generates a set of research topics: the specific things you want to learn from your users. Think of them as the agenda of a user interview, except Elvin runs the interview for you - inside your product, with real users, at scale.

Start your free trial of Product Fruits →

How It Actually Works

Step 1 - Define your unknown.

You don't need to be a UX researcher or know how to write perfect interview questions. You just need to articulate the problem. Low adoption of a feature? Drop-off at a specific step in onboarding? Users not reaching a key milestone? Describe it in plain language. Elvin asks a few clarifying questions to make sure it understands your goal, then generates the discovery topics for you.

Step 2 - Choose who to inquire.

Discoveries use the same segmentation engine as every other piece of content in Product Fruits, so you can be as precise as your data allows. Target by any user property you pass through the identification snippet - plan type, role, signup date, custom attributes - and combine that with page-level or behavioral conditions. You could ask only users who've been active for 30+ days but haven't touched a specific feature, or only admins on a paid plan in a specific region. The right respondents make everything downstream more reliable.

Step 3 - Publish to your users.

Once you're happy with the topics and targeting, you publish the discovery. A small, unobtrusive prompt appears in your product, inviting users to share a few minutes of their time. Users can snooze it, skip it, or engage -it's respectful of their attention. When they do engage, Elvin opens a conversational interface and starts asking questions.

Step 4 - Elvin interviews your users.

This is where it gets genuinely different from a survey. Elvin doesn't ask the same questions to every user. Each question is contextual - shaped by the discovery topics and adapted in real time based on what the user just said. If someone mentions a specific pain point, Elvin follows up on it. If someone's answer suggests they're using the feature differently than expected, Elvin probes that. No two conversations are the same.

Users can answer in a word, a sentence, or a paragraph. Partial responses are still captured and still useful. The whole experience feels closer to a natural conversation than a compliance exercise.

Step 5 - Generate insights.

Once you have enough responses (as few as four to start), you can generate an AI-powered insight report. This isn't a summary of what people said - it's a structured analysis that combines narrative findings with hard, calculated data points, tied back to the specific topics you defined. Every finding is linked to the source conversations, so you can verify the reasoning and dig deeper where needed. You can export it as a PDF and share it across your team.

What This Means in Practice

Say you're a product lead at a CRM company and you're seeing low adoption of your integrations feature. You don't know why. You could send a survey - but what would you even ask? You'd have to guess at the possible reasons first, write questions around those guesses, and hope you covered the right ground.

With Discoveries, you describe the problem to Elvin: "We're seeing low adoption of our integrations. We don't know why." Elvin helps you shape the discovery topics - things like understanding what's blocking users from setting up integrations, what integrations they actually need, and what their expectations are around the setup process. You publish it to the relevant user segment. Elvin has the conversations. In a matter of days, you have a structured insight report telling you what's actually going on - not what you guessed might be going on.

The difference isn't just efficiency. It's the quality of what you learn.

You Don't Have to Be a User Research Expert

That's the part worth emphasizing. Building good research questions is genuinely hard - we covered five classic ways it goes wrong in our previous post on bad survey questions. Getting the right users, asking the right things, synthesizing the responses into something actionable: traditionally, this takes time, skill, and often a dedicated researcher.

Discoveries is designed so that none of that is a prerequisite. Elvin guides you through the setup. Elvin runs the conversations. Elvin synthesizes the report. Your job is to understand your product well enough to articulate what you don't know - and act on what you find out.

For product leads, CSM leads, and SaaS founders who need user insight but don't have the bandwidth or headcount for traditional research, that's a meaningful shift.

Try It Now

Discoveries is available now in Early Access inside Product Fruits. If you're already a Product Fruits customer, you can access it through the Elvin section in your administration panel - look for the Discover area under Elvin Copilot.

If you're not yet a customer, there's no better time to see what the platform can do.

‍

Start your free Product Fruits trial →

Learn more about Product Fruits →

‍

The gap between what users tell you and what they actually experience is where churn lives, where feature adoption dies, and where roadmap decisions quietly go wrong. Discoveries doesn't close that gap overnight—but it makes it a lot smaller, a lot faster, than anything else in your current stack.

Early Access means the product is live and usable, with more capabilities on the way. The best time to explore it is before everyone else does.

‍

Sources

¹ Teresa Torres, Continuous Discovery Habits - producttalk.org | Also cited in: Maze, Making Better Product Decisions with Teresa Torres

² User Interviews, The State of User Research Report 2025 - userinterviews.com

³ Maze & Atlassian, Continuous Research Trends Report - maze.co

‍