Working · 2026-11AnecdotalReferralProduct

Cut steps between join and next invite

Design the shortest path from a new user entering the product to that user inviting the next one.

Claim and evidence

The claimDesign the shortest path from a new user entering the product to that user inviting the next one.

No dated example yet. Treat this tactic as a hypothesis.

What the numbers show

80%–90%

attrition to assume when asking for a username and password (Andrew Chen's rule of thumb)

andrewchen.com

Why it works

Diagram, not data
New user arrives froman inviteUse first, registerlaterRegister beforeanything elseInvites the nextperson
New user arrives from an inviteUse first,register laterRegister beforeanything elseInvites the next person

Every extra step (like registration) loses a large share of people, so a tight loop beats a long tour. Chen advises tightening the loop before adding on-ramps.

How to run it

  1. List steps from landing to first invite
  2. Delay registration where possible
  3. Test removing each step
  4. Add onboarding on-ramps only after the loop works

Before / after

Before
Create your account
Username
Password
Continue
Registration wall before anything
After
Your friend's board
Invite someone
Sign up only to save your changes
Value first, account when saving

Generalised wireframe of the pattern, not a copy of any company's screen.

Ways to run it

Design options to test, not promised results.

  1. Option ADelay registration

    Let invited people get value first and ask for an account only when they need to save or share.

  2. Option BRemove one step per test

    Map every step from landing to first invite and test dropping them one at a time.

  3. Option CLoop before on-ramps

    Tighten the core loop first; add tours and onboarding on-ramps only once it works.

Timeline

  1. 12007 first documented: Andrew Chen
  2. 22026-11 status checked: Working

Dated examples

No dated example yet. Treat this tactic as a hypothesis.

Does it still work?

Workingas of 2026-11

Principle is old (page undated, likely 2007) and still standard in funnel design. source

Risks

  • Page undated
  • No data on the page

Sources

  1. andrewchen.com · andrewchen.com · Andrew Chen · undated
    “Assume up to 80% to 90% attrition if you are asking them to register for a username/password, for example.”
    ✓ quote checked word-for-word on the page

Change log

  • 2026-11-20 — harvest v0.1 (referral-loops)