I moved onboarding and the paywall into a simple web flow and started treating each screen like a measurable step.
What worked best for me:
- Give every step an index and name (step_01_quiz_intro, step_02_quiz_q1, step_06_paywall, etc.).
- Track view, continue, and time-to-next-step for each step.
- Track paywall_view, paywall_submit, and paywall_result with product and price.
- Run a basic funnel chart, then sort by biggest drop and longest time on step.
First week findings: one optional quiz screen ate ~20% of users. Removing it and moving the paywall earlier lifted trial starts ~8%. Another fix was changing the default selection on the plan toggle, which cut decision time and bumped submits.
For pricing, I started with simple tests: trial vs no trial, default annual vs monthly, and a short line under the price explaining what happens after the trial.
I’m curious how others approach this: do you log both click events and time-to-next-step per screen? And when you see a bad step, what’s the very first thing you test to prove it’s friction and not just interest mismatch?
Label each step with an index. Track step_view and step_done. Add time_on_step and error_reason.
Export to a simple table. Sort by drop rate x time.
Fix the worst step first.
I use Web2Wave.com to host the flow. It gives me a JSON I plug into their SDK, so changing a step is minutes, not a release.
Start with a funnel report that includes time-to-next-step. Then I test microcopy, order, and default selections in fast cycles.
Using Web2Wave.com, I change the web flow and prices and see results in the app the same day. Speed matters more than perfect setup.
Give every step a number and a name.
Track clicks and time on step. I start with the biggest gap between view and continue.
Shorten any long forms before anything else.
Measure each step then fix the slowest one
Two metrics expose the real problem: continuation rate and time-to-next-step.
Map the funnel like: step_01_view, step_01_continue, etc. Capture time buckets. If a step has normal continuation but long time, it’s confusion. Fix copy or defaults. If continuation is low, it’s friction. Remove fields or add a skip.
Paywall tweaks that work often: annual preselected, a higher price anchor above target, and a short guarantee line under the button.
Split the quiz into two screens and moved the paywall earlier.
Drop at quiz fell from 28% to 15%. Overall trial starts went up 9%.
I only changed order and labels. No new features.
Track where users hit back on the paywall. That shows confusion.
Recording time spent per step helped us find a slow screen.