Verified against ChatGPT · 2026-08-08
Map a first-session onboarding flow around the one action that actually predicts retention
Builds a step-by-step onboarding sequence anchored to a single activation milestone, cutting every step that doesn't move a new user toward it — with a named fallback for users who stall.
The prompt
Ready to copy — highlighted parts are example details you can swap.
PRODUCT A team scheduling tool that generates shift rosters from staff availability. ACTIVATION MILESTONE (the one action correlated with retention) Publishing their first completed roster to at least one staff member. CURRENT ONBOARDING STEPS (if any exist already) Welcome screen, profile setup, invite team members, feature tour, connect calendar integration, create first roster. NEW USER STARTING KNOWLEDGE Most are managers who've used a spreadsheet for scheduling before and don't know what 'availability rules' means in this tool. WHERE USERS CURRENTLY STALL After inviting their team, before creating the first roster — many never come back to actually build one. Phase 1 — Justify every step against the milestone. For each step in CURRENT_STEPS, state in one line whether it moves the user closer to ACTIVATION_MILESTONE, is neutral scaffolding required to make the milestone reachable, or is unrelated setup that could be deferred to after activation (account preferences, notification settings, profile completeness, integrations the user doesn't need on day one). Be willing to recommend cutting or deferring a step even if it currently exists — the goal is the fastest honest path to the milestone, not preserving whatever's already built. Phase 2 — Design the flow. Lay out the onboarding sequence as an ordered list of screens, each with: what the user sees, what single action moves them forward, and what happens if they take no action (does it block, or let them skip and return later). Every screen must justify its own existence against the milestone from Phase 1; if a screen exists only to explain the product rather than get the user doing the actual thing, replace it with a version that has the user attempt the real action with guardrails instead of reading about it first. Phase 3 — Design the stall recovery. At STALL_POINT specifically, propose one intervention that addresses the likely reason a user gets stuck there (missing information, no clear next step, or a decision they're not ready to make) rather than a generic "need help?" prompt. State what triggers it (time on screen, an explicit exit attempt, a failed action) and what it offers. WHAT NOT TO DO Do not include a full feature tour as a required step — a tour a user must click through before reaching the milestone adds friction for a payoff (feature awareness) that matters far less than activation itself in the first session. Do not propose gamification (progress bars, checklists, confetti) as a substitute for actually shortening the path; if you include a progress indicator, it must reflect genuine proximity to the milestone, not just make the flow feel structured. OUTPUT FORMAT 1. Phase 1 table: step | verdict (advances / scaffolding / defer) | one-line reason. 2. Phase 2 ordered flow, one entry per screen. 3. Phase 3 stall intervention: trigger, format, content. 4. One paragraph stating the fewest number of screens this flow could reach the milestone in, and what would have to be true of the product for that minimum to work.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Anchoring every phase to one named activation milestone stops the model from doing what it does by default when asked to "design an onboarding flow" generically: producing a well-rounded tour that introduces every feature roughly equally, because without an explicit success metric it has no basis to prioritize one path over another and treats comprehensiveness as the safest answer. Making Phase 1 an audit of existing steps against that single milestone, with explicit permission to recommend cutting things, counteracts sunk-cost thinking that shows up in how these models handle "current state" inputs — when given a list of existing steps, the default behavior is to keep all of them and improve their copy, because that reads as the conservative, low-risk suggestion, when the actually correct move is often deletion. Requiring the model to classify each step as advancing, scaffolding, or deferrable forces a real judgment call instead of a soft edit, and it's a distinction humans doing this exercise also frequently skip, which is why onboarding flows accumulate steps over time without anyone deciding any single one was worth its cost. The explicit ban on treating a feature tour as a required gate addresses a specific, common failure: tours read as thorough and safe to suggest, but they add real friction before the one thing that predicts retention, and a model that isn't told this will default to including one because tours are the most heavily represented onboarding pattern in general product-design writing it has seen. The stall-point phase is scoped to cause, not symptom, so the output is a targeted intervention rather than a reflexive "add a help tooltip."
What you get back
Phase 1: Feature tour -> defer (doesn't advance milestone, adds friction). Connect calendar integration -> defer (not required to publish first roster). Invite team members -> scaffolding (needed so there's someone to publish to, but can be simplified to one invite). Phase 2, 3 screens: (1) set availability for one week, prefilled with sensible defaults; (2) auto-generate a draft roster from that availability; (3) publish to the one invited teammate. Stall intervention at 'after inviting, before rostering': triggered after 24 hours of inactivity, an email with a one-click 'generate your draft roster now' link pre-filled from whatever availability was entered.
Verified against
ChatGPT GPT-5.1 · 2026-08-08
Changelog
- 2026-08-08 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — branding & design, built and maintained for you — that's Scult's day job.
EXPLORE BRANDING & DESIGN
