Product · onboarding
Three screens instead of five
A working prototype of the shortened signup, and the two questions we are proposing to stop asking. Click through it — it runs.
The prototype
The notes on the right follow along as you move through it.
Create your account
One field fewer than last time. We can infer the rest.
What will you use it for?
This one question replaces the old role, team-size and industry screens.
You are in
No verification wall. We email the link and let you keep going.
Screen 1
Two fields, and neither is a password
We dropped the password and send a link instead. That removes the field with the highest abandonment in the current flow and takes the password reset path with it. Name stays because we use it in the workspace and guessing it from an email address goes wrong often enough to be worse than asking.
Screen 2
One question, not three
Role, team size and industry were three screens that fed one segmentation model. The model only ever used them together, and this single question predicts the same segment about as well. "Still deciding" is a real option — the current flow forces a guess and then we treat that guess as a fact.
Screen 3
No verification wall
The link is sent, but the account works before it is clicked. Verification gates the things that need it — inviting people, custom domains — rather than the first minute. This is the change most likely to be argued with, and the section below is the argument.
What we cut
Today
Five screens · 44% drop-off
- Email and password
- Verify your email before continuing
- What is your role?
- How big is your team?
- Name your workspace
Proposed
Three screens · target 25%
- Email and name
- What will you use it for?
- Done — workspace named from the email domain
The verification change is the risky one
Letting an unverified account in is how you collect junk signups, and our current numbers do not tell us how many we would get, because we have never allowed it. The mitigation is that an unverified account cannot invite anyone, cannot attach a domain, and is deleted after fourteen quiet days. If that turns out not to be enough, this is the one change to reverse first — it is independent of the other two.
What we would measure
- Completion rate, screen by screen. The target is 75% end to end, against 56% today.
- Junk signups per week. If this rises more than fivefold, verification goes back.
- Segment accuracy. Compare the single question's segment against the three-question one for accounts that answered both during the overlap.
- Time to first document. The number that actually matters. Everything above is a proxy for it.