onboarding
why
- the stay/churn decision is made at the first valuable moment (FVM): the first time the product does the thing it promised.
- every step before the FVM is a friction point where visitors abandon; no polish after it recovers a visitor who never arrived.
- tours, carousels, setup checklists are confessions that the value path is unclear.
- cutting the path to the shortest honest route is the highest-leverage product change available.
rule
- onboarding is the path from arrival to the FVM: name that moment, then remove every step that does not lead to it.
- a tour or tooltip sequence is not onboarding; make the first screen itself performative — the interface teaches by doing.
- account setup, preferences, education screens, payment gates all add friction. visitors who never reach the moment the product proves its value churn.
- optimizing steps after the first value has no leverage — visitors are already gone.
design order
- name the FVM: the shortest interaction after which the visitor has received value — an answer, a generated artifact, a working connection.
- map every step from cold arrival to the FVM.
- delete or defer every step not required for it.
- defer everything else to the moment it is genuinely needed — progressive disclosure applied to product scope.
| step candidate | keep before FVM | defer |
|---|---|---|
| authentication | when identity is required for the value itself. | otherwise — value first, account after. |
| configuration | one choice that shapes the value. | every optional setting. |
| education/tour | never. | the interface teaches by doing. |
| payment | never before demonstrated value. | after FVM, at the upgrade boundary. |
| permissions/integrations | the one integration the FVM uses. | the rest. |
empty states are onboarding
the empty state is the first screen many visitors see; it must explain the absence and offer the next valid action — per truthful product states. "nothing here yet" plus a dead end is a funnel exit, and an untruthful one.
do not
- front-load account/profile forms before value exists.
- build a tour overlay; make the first screen itself performative.
- require payment before demonstrated value.
- defer the FVM behind "complete your setup" checklists.
- measure activation as signup — signup is a step, not the moment.
verification
- walk the live path from cold arrival: count screens and inputs to the FVM; each must justify itself against the table.
- probe the empty state of the FVM surface: it explains and offers the action.
- run one new visitor (fresh session, no help) through the path; every point where they hesitate or ask "what do I do" is a defect.
good and bad
bad
arrival → create account → fill profile → set preferences → watch onboarding tour → connect data source → first workflow run
profile and preferences add 2-3 minutes of form filling before first value; 68% of visitors bounce at the preferences screen without seeing the first run succeed. the team then optimizes the tour to be "snappier," missing the real problem: abandonment before value.
good
arrival → authenticate → connect source → first workflow run → optional profile (next session)
only the one connection required for the FVM blocks arrival; after a visitor sees a workflow complete, profile-edit friction is much lower. first-run completion goes from 32% to 71%.