If you work on a product team and care about getting new users to stick around, you’ve probably stared at onboarding funnels and wondered, “Why do so many people drop off right here?” This guide is for folks who actually want to answer that question, not just make another dashboard. We’ll dig into how to use path analysis in Amplitude to spot what’s working, find what’s broken, and make onboarding smoother—without getting lost in a sea of charts.
Let’s keep it practical. We’ll cover how to set up path analysis, spot real issues, and avoid chasing vanity metrics. If you want to fix onboarding (not just talk about it), you’re in the right place.
Why Path Analysis Beats Funnels for Onboarding
Funnels show you where users drop off, but they don’t tell you what users are actually doing. Funnels are blunt tools—they assume everyone behaves the same way and follows your “happy path.” Reality check: most people don’t.
Path analysis lets you:
- See the actual steps users take, not just the ones you expect.
- Spot detours, confusing loops, or common dead ends.
- Identify unexpected events that happen before, during, or after onboarding.
In short: it gives you the full picture, not just the tidy version you wish was true.
Step 1: Make Sure Your Events Are Actually Useful
Before you jump into path analysis, take a hard look at your event tracking. Garbage in, garbage out. If your events are vague (“Button Clicked”) or missing key actions, you’ll end up guessing.
What you need:
- Clear, specific events for each onboarding step. e.g., “Started Onboarding,” “Uploaded Profile Picture,” “Connected Bank Account.”
- Consistent naming. Don’t make your future self decode “Step 1 Complete” vs “Form Submitted.”
- Contextual properties. Sometimes, knowing which button was clicked or which screen matters.
Pro tip: Audit your events by going through the onboarding yourself. Does every meaningful action show up in your data? If not, fix that first. Otherwise, path analysis will just show you noise.
Step 2: Map Out Your “Expected” Onboarding Flow
Before you dive into user data, sketch out what you think the ideal onboarding looks like. List the key steps in order, like:
- Account Created
- Email Verified
- Profile Completed
- First Project Created
This isn’t just busywork. It’ll help you compare your assumptions to reality—and spot where things go off the rails.
Step 3: Run Your First Path Analysis in Amplitude
Now for the fun part. In Amplitude, head to the Paths chart. Here’s how to get started:
- Choose a starting event. For onboarding, this is usually “Signed Up” or “Started Onboarding.”
- Set the analysis direction. Most onboarding flows are best viewed “forward” (what users do after signing up), but sometimes “backward” (what did users do before dropping off?) can be eye-opening.
- Adjust the path depth. Don’t go too deep at first—3 to 5 steps is usually enough to see patterns without getting overwhelmed.
- Filter by segment. New users only. You don’t want to mix in returning users who already know the ropes.
What to look for:
- Popular alternate paths. Are users skipping steps? Taking detours?
- Loops and drop-offs. Where do people get stuck or bail out?
- Unexpected events. Are users doing something you didn’t plan for (like hitting “Help” right after step 2)?
Pro tip: Don’t obsess over every strange path. Focus on the ones lots of users take, or that lead to clear drop-off.
Step 4: Dig Into Pain Points and Detours
Once you spot patterns, it’s time to ask “why.” Here’s how to get more from your path analysis:
- Zoom in on high-drop-off branches. Click into the steps where most users abandon ship. What happens right before or after?
- Compare segments. Do users on mobile bail at different steps than those on desktop? What about by country or signup source?
- Overlay properties. Use event properties (like device type or plan) to see if issues are isolated or widespread.
Watch out for:
- Loops. If users keep toggling between “Add Payment Info” and “Error Message,” you’ve found friction.
- Dead ends. If a big chunk of users end up on “FAQ Opened” or “Support Chat Started,” your flow isn’t clear.
- Skipped steps. If users skip “Email Verified,” you might have a loophole—or a bug.
What not to do: Don’t chase every single quirk. Focus on the biggest pain points. Fixing those will move the needle; the weird one-off edge cases can wait.
Step 5: Prioritize and Test Fixes
You’ve found the rough spots—now what? Don’t try to fix everything at once.
- Rank problems by impact. How many users are affected? How bad is the drop-off?
- Brainstorm fixes. Is it a confusing UI? Missing instructions? Technical bug? Sometimes the solution is as simple as clearer copy or a progress bar.
- Make small changes. Tweak one thing at a time, so you can see what actually works.
- Set up new events if needed. If you realize you’re missing key data (like “Clicked Help” vs. “Exited Flow”), add it before testing changes.
Pro tip: Document what you change and why. It’s easy to forget—and hard to learn—if you don’t.
Step 6: Re-Run Your Path Analysis—Keep Iterating
After shipping changes, rerun your path analysis to see what’s changed. Are fewer users dropping off at the old pain points? Are new detours popping up?
- Don’t expect perfection. There’s always friction somewhere.
- Look for new patterns. Sometimes fixing one bottleneck reveals the next one.
- Celebrate real improvements. If more users finish onboarding, that’s a win—even if it’s not 100%.
What to ignore: Don’t obsess over tiny improvements or rare paths. Focus on the big blockers.
What Works, What Doesn’t, and What to Skip
What Actually Helps
- Tracking meaningful events, not just button clicks.
- Focusing on the most common paths and biggest drop-offs.
- Making small, testable changes—then measuring the impact.
What Doesn’t
- Overanalyzing edge cases or rare user journeys.
- Relying on funnels alone (they hide too much real behavior).
- Blindly copying onboarding flows from other products—your users and app are unique.
What to Ignore
- Vanity metrics like “number of screens viewed.”
- Super granular events that don’t inform any decisions.
- Overly complex path charts (if you can’t explain it to your grandma, it’s probably not actionable).
Wrap-Up: Keep It Simple, Keep Moving
Optimizing onboarding with path analysis in Amplitude isn’t magic. It’s about making your data actually useful, spotting real friction, and fixing what matters. Don’t get sucked into analysis paralysis or endless dashboards. Start with clear questions, make small changes, and keep iterating.
Remember: onboarding will never be perfect. But with a solid approach to path analysis, you’ll get a lot closer—and your users will thank you for it.