How to use onboarding session recordings and interview transcripts to identify activation friction points

Most product teams know their onboarding completion rates are lower than they would like. Dashboards show where users drop off, but they rarely explain why. Analytics can tell you that 40% of new users never complete workspace setup. They cannot tell you whether users were confused by the terminology, overwhelmed by the number of steps, or simply did not understand why the step mattered.

This is where qualitative data becomes essential. Onboarding session recordings and interview transcripts are two of the most useful sources for understanding activation friction—the obstacles that prevent new users from reaching their first moment of value. Used together, they provide both the behavioral evidence and the human context needed to diagnose problems accurately and prioritize the right fixes.

This guide walks through a practical process for combining these two data sources to find and act on activation friction points.

What activation friction actually looks like

Before diving into method, it helps to be specific about what you are looking for. Activation friction is not limited to bugs or broken flows. It includes any moment where a new user's momentum toward value slows down, stops, or reverses.

Common forms of activation friction include:

• Conceptual confusion — Users do not understand what a feature does or why they need it at this stage.
• Decision paralysis — Users encounter a choice (plan selection, configuration options, template selection) without enough context to decide confidently.
• Expectation mismatch — What users were told during sales or marketing does not match what they experience in the product.
• Unnecessary effort — Users are asked to do manual work that feels tedious or irrelevant before they can see any value.
• Missing feedback — Users take an action but receive no confirmation that it worked or guidance on what to do next.
• Vocabulary gaps — The product uses terms the user does not recognize or interprets differently.

These problems are difficult to detect with quantitative data alone. A user who spends four minutes on a setup screen might be thoughtfully configuring their workspace or might be completely lost. Session recordings and interviews help you tell the difference.

Why these two sources complement each other

Session recordings and interview transcripts each have distinct strengths and limitations. Understanding these tradeoffs is important for combining them effectively.

What session recordings reveal

Session recordings capture user behavior in real time. You see cursor movement, click patterns, scrolling behavior, hesitations, rage clicks, and navigation paths. They are particularly useful for identifying:

• Where users pause or hesitate before taking action
• Steps users skip or attempt to skip
• Moments where users navigate away from the intended flow
• UI elements users interact with that are not part of the intended path (e.g., clicking on non-interactive text expecting it to be a link)
• How long specific steps actually take versus how long they should take

The limitation of recordings is that they show behavior without intent. A long pause could mean confusion, distraction, or careful reading. You are interpreting what you see, and those interpretations can be wrong.

What interview transcripts reveal

Interview transcripts capture the user's own account of their experience. They provide access to mental models, expectations, emotional responses, and reasoning. They are particularly useful for identifying:

• What users expected to happen versus what actually happened
• Which steps felt unnecessary or confusing and why
• What language users use to describe their goals (which may differ from product terminology)
• Whether users understood the purpose of each onboarding step
• What prior experiences shaped their expectations

The limitation of interviews is recall bias. Users reconstruct their experience after the fact, and their account may be incomplete, reordered, or rationalized. They may also downplay confusion to be polite or because they blame themselves rather than the product.

The combined picture

When you watch a recording of a user struggling with a step and then read that same user explaining in an interview what they were trying to do, you get a far richer understanding of the friction. The recording validates or challenges what the user said. The interview explains what the recording could not show. Neither source is sufficient on its own.

Step-by-step process for identifying friction points

The following process is designed to be practical for a team of one or two researchers, though it scales well for larger teams.

Step 1: Define your activation milestone

Before reviewing any recordings or transcripts, get specific about what activation means for your product. Activation is not the same as signup or onboarding completion. It is the moment when a user first experiences the product's core value.

For a project management tool, activation might be completing a task within a shared project. For an analytics platform, it might be generating a first report from real data. For a research repository, it might be tagging and searching across multiple imported documents.

Write down your activation milestone in concrete terms. Everything you analyze will be evaluated in relation to this milestone: did the user reach it, and if not, what got in the way?

Step 2: Collect and organize your data

Gather onboarding session recordings and interview transcripts for the same time period. Ideally, some of your interview participants are the same users whose sessions you recorded. This allows you to cross-reference behavior and self-reported experience for the same individual.

If you do not have matched pairs, that is fine. You can still triangulate patterns across the two data sets. But aim for overlap where possible when planning future research.

Organize recordings by outcome:

• Activated users — Users who reached the activation milestone during onboarding
• Stalled users — Users who partially completed onboarding but did not activate
• Dropped users — Users who abandoned onboarding entirely

Reviewing all three groups is important. Activated users show you what a successful path looks like, which helps you identify what is different about the unsuccessful paths.

Step 3: Review recordings with a friction log

Watch each session recording with a simple friction log open. For every moment where the user's progress appears to slow or stop, note:

• Timestamp — When it happened
• Screen or step — Where in the onboarding flow it occurred
• Observed behavior — What you saw (e.g., "user clicked settings icon three times, then scrolled up and down the page")
• Friction type — Your best hypothesis from the categories listed earlier (conceptual confusion, decision paralysis, etc.)
• Severity — Whether the user recovered and continued, needed multiple attempts, or abandoned

Resist the urge to watch recordings at 2x speed for this exercise. Friction is often subtle—a two-second hesitation, a mouse cursor circling before clicking—and easy to miss when skimming.

Step 4: Code interview transcripts for friction themes

Read through interview transcripts and highlight passages where users describe difficulty, confusion, frustration, or unmet expectations during onboarding. Apply descriptive codes to each passage. Start with codes based on your friction categories, but allow new codes to emerge from the data.

Pay particular attention to:

• Moments where users describe what they expected to happen versus what did happen
• Language like "I wasn't sure," "I didn't know why," "I almost gave up," or "I had to figure it out on my own"
• Steps users describe as unnecessary or redundant
• Comparisons to other products ("In [other product], I could just...")

Also note positive signals—steps users describe as easy, clear, or satisfying. These anchor points help you understand what a low-friction experience looks like in your product.

Step 5: Triangulate across sources

This is where the two data sources come together. Look for convergence—friction points that appear in both recordings and transcripts. These are your highest-confidence findings.

Common patterns of convergence include:

• Multiple recordings show users hesitating on the same screen, and multiple interview participants describe that screen as confusing
• Recordings show users skipping a step, and interviews reveal users did not understand why that step was necessary
• Recordings show users navigating to help documentation mid-flow, and interviews confirm users felt the in-product guidance was insufficient

Also look for divergence—cases where one source contradicts the other. A user might describe onboarding as "pretty easy" in an interview, but their recording shows multiple wrong turns and a rage click on a dropdown menu. Divergence often signals social desirability bias in interviews or reveals friction that users have normalized and no longer consciously register.

Step 6: Map friction to the onboarding flow

Plot your findings onto a visual map of your onboarding flow. For each step, note:

• How many recordings showed friction at this step
• How many interview passages referenced difficulty at this step
• The predominant friction type
• Whether users who encountered friction here ultimately activated or did not

This map gives your team a shared artifact that is far more useful than a list of findings. It shows not just where friction exists but where it clusters and where it correlates with failure to activate.

Step 7: Prioritize and act

Not all friction points are equally worth fixing. Prioritize based on:

• Frequency — How many users encountered this friction point?
• Severity — Did users recover, or did this friction lead to abandonment?
• Proximity to activation — Friction that occurs just before the activation milestone may be more damaging than friction early in the flow, because the user has already invested effort.
• Feasibility — Can this be addressed with a copy change, a flow adjustment, or does it require significant engineering work?

Present findings to your product team with specific evidence: a recording clip showing the behavior, a transcript excerpt explaining the user's reasoning, and a clear description of the friction type. This combination is far more persuasive than abstract recommendations.

Practical tips for making this process sustainable

Record onboarding sessions by default

If your product does not already capture session recordings for new users, set this up as standard practice. Many session recording tools allow you to filter by user segment and session number, making it straightforward to isolate first-time onboarding sessions.

Build interview questions around the onboarding flow

When interviewing recent signups, walk through the onboarding steps chronologically. Ask users to describe each step in their own words before you reveal what the step was intended to do. The gap between their description and your intention is where friction lives.

Use a shared repository for findings

Friction analysis produces a large amount of qualitative data—timestamped observations, coded transcript passages, and cross-referenced findings. Storing this in scattered documents or spreadsheets makes it difficult to revisit or build on. A dedicated research repository like Dovetail allows you to tag and search across recordings and transcripts, making it easier to spot patterns across multiple onboarding studies and track whether friction points have been resolved over time.

Revisit after changes ship

After your team addresses a friction point, repeat the analysis for that specific step. Watch new onboarding recordings and ask about the step in subsequent interviews. Friction analysis is not a one-time project—it is an ongoing feedback loop that should be embedded in your onboarding improvement process.

Common mistakes to avoid

Watching only failed sessions. If you only review recordings of users who dropped off, you will not understand what successful onboarding looks like. The contrast between activated and stalled users is where the most actionable insights emerge.

Treating interview quotes as ground truth. Users' accounts of their experience are valuable but imperfect. Always cross-reference with behavioral data before making high-confidence claims about what happened.

Skipping the problem framing step. If you have not defined your activation milestone clearly, your analysis will lack focus. You will find friction everywhere—because friction exists everywhere—without being able to distinguish the friction that matters from the friction that does not.

Presenting findings without evidence. Telling a product team "users find step three confusing" is easy to dismiss. Showing a 30-second recording clip alongside a transcript excerpt is much harder to ignore. Build your case with paired evidence from both sources.

Bringing it together

Identifying activation friction is one of the highest-leverage research activities a product team can undertake. Small improvements to onboarding compound over time—every user who activates instead of churning represents long-term value.

The combination of session recordings and interview transcripts gives you the behavioral specificity and human context needed to diagnose friction accurately. Recordings show you the problem. Interviews explain it. Together, they point you toward solutions that address root causes rather than symptoms.

The process does not require sophisticated tools or large research teams. It requires discipline in data collection, rigor in analysis, and a clear definition of what activation means for your product. Start with your highest-drop-off onboarding step, review 15 recordings, read 10 transcripts, and map what you find. The friction points will surface quickly—and they will be specific enough to act on.