Product-Led Growth Research: How to Combine Usage Data with Qualitative Interviews

Product-led growth (PLG) is one of the most powerful go-to-market strategies in SaaS — but it creates a peculiar research challenge. You have more behavioral data than ever before: activation rates, feature adoption funnels, time-to-value curves, session recordings by the thousands. Yet most PLG teams still cannot answer the question that matters most: why do some users activate and others do not?

Analytics tells you what happened. User research tells you why. For PLG teams, combining both is the key to unlocking sustained growth — and AI-powered qualitative research is finally making it practical at scale.

What Is Product-Led Growth Research?

Product-led growth research is the practice of combining product usage data with qualitative user insights to understand the behaviors and motivations that drive activation, retention, and expansion.

In a PLG model, users self-serve through discovery, signup, and value realization without ever talking to a salesperson. This means:

• Your product is your primary growth engine
• First-time experience quality determines conversion
• Understanding the activation journey is critical
• Feature adoption directly impacts revenue expansion

Research in this context is not about understanding vague "user needs" — it is about diagnosing specific failure points in a self-serve journey and understanding the exact moments where users find (or do not find) value.

The PLG Research Cycle

Effective PLG research follows a continuous cycle:

1. Identify behavioral patterns (analytics)

Use your product analytics tool (Amplitude, Mixpanel, etc.) to find segments worth investigating:

• Users who activated in week 1 vs. those who did not
• Users who expanded (added seats, upgraded) vs. those who stayed on free
• Users who churned within 30 days vs. those who retained
• Power users who complete 3+ activation moments vs. casual users

2. Interview the segments (qualitative research)

For each behavioral segment, run targeted AI interviews with 15–25 users. Ask about their motivations, decision points, and experience at the moments your analytics highlighted.

3. Build the insight (synthesis)

Combine the quantitative patterns (what percentage of users did X) with qualitative context (why they did X and what they were trying to accomplish) to build actionable product insights.

4. Validate and ship

Run experiments based on your hypotheses, then run follow-up research to confirm the change achieved the intended effect.

Why PLG Teams Underinvest in Research

Most PLG teams default to A/B testing and analytics for a simple reason: it is faster than scheduling and running qualitative research. A/B tests run automatically. User interviews require recruiting, scheduling, facilitating, transcribing, and synthesizing — weeks of effort for 10 conversations.

This is why AI-moderated research changes everything for PLG teams. Platforms like Koji let you run 50 AI-moderated interviews in a week with no scheduling, no facilitation overhead, and automatic analysis. The marginal cost of learning "why" drops dramatically — making qualitative research practical as a continuous practice rather than an occasional project.

Key Research Studies for PLG Teams

Activation Research

The activation moment — when a user first experiences your product's core value — is the most important moment in a PLG funnel. Research it obsessively.

Study design:

• Trigger: Users who just reached (or failed to reach) activation in their first 7 days
• Questions: Walk me through what you did in the product on day 1. What were you trying to accomplish? What felt unclear? Did you reach the moment where you thought "yes, this works"?
• Structured question: Scale 1–10, how confident are you that this product will solve [core problem]?

Koji's structured question types let you collect quantitative confidence scores alongside qualitative reasoning in a single conversational session.

Expansion Research

In PLG, expansion — users adding seats, upgrading plans, activating premium features — is where revenue grows. Understanding what triggers expansion decisions is critical.

Study design:

• Segment: Users who upgraded in the last 30 days vs. users who have been on free for 90+ days
• Questions: What made you decide to upgrade? What feature or moment convinced you? What would you need to see to add more team members?
• Structured question: Single choice — what was the primary reason for upgrading?

Churn Research

When users churn from a PLG product, they do not tell you why. They just stop logging in. Research catches what analytics misses.

Study design:

• Trigger: Users who have not logged in for 30 days (send a link before cancellation)
• Questions: What were you originally trying to accomplish? Did the product deliver? What would have needed to be true for you to keep using it?
• Structured question: Yes/No — did you find an alternative solution?

Feature Adoption Research

You built a feature. Analytics shows 10% of users try it. Research reveals why 90% do not — and whether the problem is discovery, onboarding, perceived value, or positioning.

Study design:

• Segment: Users who used the feature 3+ times vs. users who tried it once and never returned
• Questions: Tell me about the first time you used [feature]. What were you hoping it would do? What did it actually do?

Learn more in the dedicated feature adoption research guide.

Running PLG Research at Scale with Koji

The practical challenge of PLG research is volume. In a PLG model, you may have thousands of users moving through your funnel daily. To understand the patterns, you need to interview at scale — not just 5 users, but 50–100 per segment.

Koji makes this possible through fully automated AI interviews:

1. Set up a study with your research questions and structured question types
2. Trigger via webhook or CRM import when users hit a behavioral threshold (e.g., day 7 without activation)
3. AI interviews every participant — no scheduling, no moderator, available 24/7 in any timezone
4. Automatic analysis surfaces themes, patterns, and representative quotes across all responses
5. Report generation aggregates findings with charts for structured questions alongside qualitative themes

For activation research specifically, the combination of scale ratings (confidence scores) and open-ended follow-ups means your report shows both the distribution of scores AND the themes explaining why users felt that way.

Integrating Research Into Your PLG Stack

The best PLG research programs connect three data sources:

The workflow: analytics alerts you to a pattern (e.g., 40% drop-off at step 3 of onboarding), session recordings show you what happens at that step (users hesitate on the email verification screen), and Koji interviews explain why (users do not understand why verification is required before they can use core features).

Each layer makes the others more valuable. Analytics without research is correlation without causation. Research without analytics is anecdote without scale.

Automating Research Triggers with Webhooks

For mature PLG teams, research triggers can be automated end-to-end. Using Koji's webhook integration and headless API:

• When a user hits day 7 without completing onboarding, automatically send an interview invitation
• When a user reaches a usage milestone, trigger an expansion study
• When a user goes dormant (no login for 21 days), trigger a churn risk interview

This creates a research pipeline that runs 24/7 without manual intervention — capturing data continuously as users move through your funnel.

PLG Research Mistakes to Avoid

Only talking to happy users: Your power users love the product. They will tell you great things. The insight you need is in the segment that did not activate — the harder-to-reach users who left quietly.

Waiting for enough data: PLG teams often delay research until they have statistical significance in their analytics. Do not wait. Even 15–20 qualitative interviews will surface the core themes.

Treating all activation cohorts as the same: Users who activated in month 1 have different experiences than users in month 6 after multiple product iterations. Segment your research by cohort.

Ignoring the pre-signup journey: PLG research often starts at signup. But the decision to sign up (or not) is shaped by everything before — website, content, comparison research. Interview users who considered your product but chose a competitor.

Conflating engagement with value: A user who logs in daily but never achieves their core goal is a churn risk, not a power user. Research should measure perceived value delivered, not just usage frequency.

Starting Your PLG Research Program

The fastest way to get started:

1. Pick one funnel stage — activation, retention, or expansion
2. Identify the top drop-off segment in your analytics
3. Create a Koji study with 5–8 questions targeting that segment
4. Import or trigger participants via CSV or webhook
5. Review the AI-generated report after 20+ responses

Most PLG teams run their first Koji study and immediately identify 2–3 product changes they had not considered from analytics alone. The combination of quantitative signals and qualitative reasoning produces the kind of insight that actually changes roadmap priorities.

Related Resources

• Structured Questions in AI Interviews
• Feature Adoption Research
• User Onboarding Research
• Churned Customer Interviews
• Research Automation: Webhooks
• How to Automate User Research