Onboarding, First-Run & Empty States

The first session decides retention more than any other. Goal: shortest credible path to the user's first success ("aha moment"), not the most thorough tour.

Principles

• Show value before asking for effort: defer registration, permissions, and personal questions until their benefit is obvious (Apple HIG says exactly this for permissions; growth practice agrees for sign-up walls).
• Do, don't tour: interactive first-task beats slide carousels; NN/g's research on mobile onboarding found front-loaded feature tours are frequently skipped and poorly retained — teach in context at the moment of need (just-in-time tips) instead.
• Default to skippable: autonomy (see SDT in Motivation & Behavior); every tip dismissible, tours escapable, and never shown twice after demonstrated competence.
• Progressive setup: get to the core experience with minimal setup; collect preferences through use (implicit) or later prompts (explicit), not a 12-screen questionnaire. If setup is unavoidable, show progress and explain each ask's benefit ("so we can find matches near you").
• The empty state IS onboarding: a blank list/canvas is a fork between activation and abandonment.

Personalization quizzes — trade-offs

• A short setup quiz can work when its answers visibly change the first experience (content feed, plan, sizing) — the payoff must appear on the very next screen, or it reads as data harvesting. Each question is paid for in drop-off; cut any question whose answer you could infer from behavior instead.
• Quizzes also set expectations and create commitment (effort invested → higher perceived fit — an endowment/sunk-cost effect widely exploited in consumer onboarding; label this a persuasion technique and check it against the ethics file). Always skippable, always editable later, and never gate previously free functionality behind quiz completion.

Empty-state design (first-use, cleared, no-results, error)

1. First use: explain what will live here + one clear starting action ("No projects yet — Create your first project") + optional sample/demo content or template gallery (lets users learn by editing, which beats blank-canvas paralysis).
2. User-cleared (inbox zero): celebrate/confirm, suggest next step.
3. No results: state what was searched, offer fixes (spelling, filters, broader query), never a dead end.
4. Error/empty-by-failure: honest cause + retry path (see Feedback, Loading, Errors & Recovery).

Activation mechanics & the activation metric

• Define the activation event (the measurable first-value action) and instrument the funnel to it; remove every step that isn't strictly needed. In practice teams pick the metric by finding the early behavior most correlated with long-term retention, then validating causally by pushing users toward it (standard growth practice popularized by Facebook's reported "7 friends in 10 days" — a company anecdote, not a published study; Amplitude's and Reforge's playbooks formalize the method).
• Track time-to-value alongside conversion: the same activation rate achieved in 2 minutes vs. 2 days retains very differently.
• Checklists work (Zeigarnik + goal-gradient: consider one pre-completed item — endowed progress, Kivetz et al. 2006); email/notification re-engagement should deep-link straight into the incomplete step.
• Beware optimizing activation at retention's expense (nagging users into an event they didn't value inflates the metric, not the habit).

Tooltips & coach-marks — use and accessibility

• NN/g's instructional-overlay research: users routinely dismiss coach marks without reading, and can't recall them at the moment of need — prefer contextual, triggered-by-behavior tips over launch-time overlays; if you must sequence tips, keep it to ~3 and make "skip all" one tap.
• Accessibility: tips triggered on hover/focus must be dismissible (Esc), hoverable, and persistent until dismissed (WCAG 1.4.13); anything keyboard users can't reach, they can't learn from — tie tips to focusable elements and expose the text via aria-describedby or move focus to the tip's dialog when it demands action.
• Don't trap focus in a walkthrough; never block the whole UI behind an undismissable overlay; re-show entry points from a help menu so the information isn't once-only.

Sources

• Apple HIG — Onboarding; requesting permissions in context (developer.apple.com/design).
• Nielsen Norman Group — "Mobile-App Onboarding" and instructional-overlay research (nngroup.com).
• Kivetz, R., Urminsky, O. & Zheng, Y. (2006). "The goal-gradient hypothesis resurrected." Journal of Marketing Research, 43(1).
• W3C WCAG 2.2 — SC 1.4.13 Content on Hover or Focus.
• Amplitude — Mastering Retention/North Star playbooks; Reforge growth programs — activation-metric practice (practitioner sources).
• Material Design — empty states guidance (m3.material.io).
• Eyal, N. (2014). Hooked — activation/investment framing (use with the ethics file).