These are the friction patterns we see again and again in B2B SaaS onboarding. Use this as a checklist when you capture evidence. If you're not seeing these in your data, you're probably not looking hard enough.
How to use this: Click New product in Pathmode and name it "Onboarding", then use these patterns to guide your evidence collection. Check what you see, skip what you don't.
Stage 1: The Promise
Where: Marketing Site → Sign Up Click User Motivation: High Primary Risk: Confusion
| Friction Pattern | Evidence to Look For |
|---|---|
| "Is this for enterprise or SMB?" | Pricing page bounce rate, support chat asking about tiers |
| Hidden pricing, unclear tiers | Users clicking between tiers repeatedly in session recordings |
| "Free trial" vs "Demo" confusion | Split traffic between CTA paths, abandoned demo forms |
Evidence to gather: Landing page analytics, sales call notes about mismatched expectations, bounce rate metrics.
Stage 2: The Gate
Where: Sign Up → Account Created User Motivation: Medium Primary Risk: Abandonment
| Friction Pattern | Evidence to Look For |
|---|---|
| "Why can't I use Google?" | Support tickets about SSO, sign-up page drop-off |
| Magic link goes to spam | Delayed verification times, repeat sign-up attempts |
| "Your password needs a hieroglyph" | Password reset rates within first session |
Evidence to gather: Sign-up funnel metrics, time-to-verification data, support tickets about access.
Stage 3: The Setup
Where: First Login → Configuration Complete User Motivation: Medium-Low Primary Risk: Fatigue
| Friction Pattern | Evidence to Look For |
|---|---|
| "What do I call this?" | Users editing workspace names multiple times |
| "I don't want my boss to see this yet" | Users skipping team invites, solo-mode usage |
| "I don't have this data ready" | Setup abandonment at import step, empty workspaces |
Evidence to gather: Setup completion rates per step, time-on-step metrics, quotes from onboarding calls.
Pro Tip: If every step isn't skippable, you'll see the evidence pile up at the first mandatory gate.
Stage 4: The First Strike
Where: Configuration Complete → First Key Action User Motivation: High (survivors) Primary Risk: Competence anxiety
This is make-or-break. The user attempts the one thing your product promised.
| Friction Pattern | Evidence to Look For |
|---|---|
| "Where do I start?" | Users clicking around without creating anything, help doc visits |
| "Did it work? I can't tell." | Users creating then immediately deleting, no second action |
Evidence to gather: Time-to-first-action, help center searches during first session, session recordings of first attempts.
Stage 5: The Ah-Ha
Where: First Action → Value Realized User Motivation: Very High Primary Risk: Disappointment
| Friction Pattern | Evidence to Look For |
|---|---|
| "This isn't what I expected" | One-action users who never return, NPS comments about results |
| "How do I show this to my team?" | Sharing feature requests, screenshot workarounds |
Evidence to gather: Day-1 to Day-7 retention, sharing/export usage, NPS responses tied to first-use timing.
Quick Diagnostic Checklist
Use this to audit your own onboarding evidence:
- Can a user reach Ah-Ha in under 5 minutes?
- Is every Setup step skippable?
- Does the first action produce a visible result?
- Can users share their first result with one click?
- Are you measuring time-to-value, not just sign-ups?
From Template to Action
Found a pattern in your evidence?
- Capture the friction as evidence. It lands in the Evidence Inbox marked Unreviewed.
- Start an intent from it. Creating an intent from an Inbox item opens a one-field brief; click Draft with agent. Or have your coding agent propose the intent from the request and the repository.
- Review the drafted claims (the objective, the outcomes, the constraints) and correct what is wrong. Drafted claims stay suggestions until you keep them.
- Work the Inbox and use each supporting signal as backing for the exact claim it supports. The result is an Agent-Ready Spec whose claims show their basis.
Related Resources
- Evidence 101 is the method behind this template
- Anatomy of an Agent-Ready Spec — What happens after you find friction